issue_owner_repo listlengths 2 2 | issue_body stringlengths 0 262k ⌀ | issue_title stringlengths 1 1.02k | issue_comments_url stringlengths 53 116 | issue_comments_count int64 0 2.49k | issue_created_at stringdate 1999-03-17 02:06:42 2025-06-23 11:41:49 | issue_updated_at stringdate 2000-02-10 06:43:57 2025-06-23 11:43:00 | issue_html_url stringlengths 34 97 | issue_github_id int64 132 3.17B | issue_number int64 1 215k |
|---|---|---|---|---|---|---|---|---|---|
[
"libjxl",
"libjxl"
] | The fact that official .deb packages are available for libjxl releases is pretty amazing. It's such a rare luxury that unless this easy availability is explicitly mentioned, most of us Debian/Ubuntu/derivative users will just assume that we'll have to tolerate an old version in our distro's repositories or not get it at all. Currently, the readme doesn't mention that the releases page also provides .deb packages for all the current Ubuntu LTS releases (none of which have libjxl in their repositories) and Debian releases as well. That's a lot of potential users that probably won't even bother trying to install libjxl, since most will think they'd have to build from source (which most users have no capability or willingness to do).
Also, the next Ubuntu LTS is still a year away, and that'll be the first LTS to have libjxl in the repository. So the debs will remain important until then and even beyond, since there will still be users looking for a more up-to-date release than what their distro offers. | Mention availability of release .deb packages for Debian and Ubuntu in README.md | https://api.github.com/repos/libjxl/libjxl/issues/2370/comments | 0 | 2023-04-11T16:40:45Z | 2023-04-20T17:00:36Z | https://github.com/libjxl/libjxl/issues/2370 | 1,662,874,762 | 2,370 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
It seems that the support for FLOAT32 is not stable right now. `JxlEncoderGetError` just returns `JXL_ENC_ERR_GENERIC` when encoding some fp32 inputs. The error can happen on many of my images, but fp16 works fine.
**To Reproduce**
Steps to reproduce the behavior:
To reproduce the error, install `imagecodecs` with `pip install imagecodecs` and run the following python program:
```python
import numpy as np
from imagecodecs import imwrite
buffer = np.load("1.npy")
print(buffer.shape, buffer.dtype)
imwrite("1.jxl", buffer)
```
Here is the example buffer:
[1.npy.zip](https://github.com/cgohlke/imagecodecs/files/11162460/1.npy.zip)
The example is produced in python because I don't know how to give `cjxl` an fp32 buffer. I will be happy to provide a way to use `cjxl` to reproduce the problem if someone give me a hint. The error happens in both lossless and lossy mode.
I got error
```
(1536, 1536) float32
Traceback (most recent call last):
File "reproduce.py", line 6, in <module>
imwrite("1.jxl", buffer)
File "/miniconda/envs/main/lib/python3.8/site-packages/imagecodecs/imagecodecs.py", line 1399, in imwrite
image: bytes = codec(data, **kwargs)
File "imagecodecs\_jpegxl.pyx", line 561, in imagecodecs._jpegxl.jpegxl_encode
imagecodecs._jpegxl.JpegxlError: JxlEncoderProcessOutput returned JXL_ENC_ERROR
```
In my bug report to the python library: https://github.com/cgohlke/imagecodecs/issues/71, the library author says that `JxlEncoderGetError` just returns `JXL_ENC_ERR_GENERIC` in this case.
**Expected behavior**
The library should not crash.
**Environment**
- OS: Linux
- Compiler version: Unknown
- CPU type: x86_64
- cjxl/djxl version string: 0.8.1
| JxlEncoderProcessOutput returned JXL_ENC_ERR_GENERIC when encoding some float32 inputs | https://api.github.com/repos/libjxl/libjxl/issues/2363/comments | 1 | 2023-04-10T23:23:28Z | 2023-09-21T08:09:56Z | https://github.com/libjxl/libjxl/issues/2363 | 1,661,537,780 | 2,363 |
[
"libjxl",
"libjxl"
] | encode：
decode:
<img width="464" alt="图片1" src="https://user-images.githubusercontent.com/41775496/230844447-f1e567f3-bbdc-4989-9216-bfd48446af3d.png">
| Unstable encoding memory usage | https://api.github.com/repos/libjxl/libjxl/issues/2362/comments | 4 | 2023-04-10T06:51:26Z | 2025-01-23T22:52:36Z | https://github.com/libjxl/libjxl/issues/2362 | 1,660,332,087 | 2,362 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
```
JPEG XL encoder v0.9.0 6d38e955 [AVX2,SSE4,SSSE3,SSE2]
Read 21888x14592 image, 394324866 bytes, 184.2 MP/s
Encoding [Modular, lossless, effort: 7],
lib/jxl/modular/encoding/enc_ma.cc:987: JXL_ASSERT: tree.size() <= kMaxTreeSize
Command terminated by signal 4
```
**Expected behavior**
encoding performed within defined limitations
**Environment**
- OS: Ubuntu 22.10
- Compiler version: clang 15.0.6
- CPU type: x86_64
- cjxl/djxl version string: v0.9.0 6d38e955 [AVX2,SSE4,SSSE3,SSE2]
| Encoder creates too-large tree and terminates | https://api.github.com/repos/libjxl/libjxl/issues/2361/comments | 0 | 2023-04-07T06:13:19Z | 2023-05-10T11:50:36Z | https://github.com/libjxl/libjxl/issues/2361 | 1,658,413,454 | 2,361 |
[
"libjxl",
"libjxl"
] | ❯ uname -a
Darwin zm2.local 22.3.0 Darwin Kernel Version 22.3.0: Mon Jan 30 20:39:46 PST 2023; root:xnu-8792.81.3~2/RELEASE_ARM64_T6020 arm64
**Describe the bug**
```
[ 60%] Building CXX object lib/CMakeFiles/jxl_extras-static.dir/extras/enc/jxl.cc.o
/Users/z/git/libjxl/lib/extras/enc/jxl.cc:120:28: error: use of undeclared identifier 'JxlEncoderSetExtraChannelDistance'; did you mean 'JxlEncoderSetExtraChannelInfo'?
JXL_ENC_SUCCESS != JxlEncoderSetExtraChannelDistance(
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
JxlEncoderSetExtraChannelInfo
/opt/homebrew/include/jxl/encode.h:912:29: note: 'JxlEncoderSetExtraChannelInfo' declared here
JXL_EXPORT JxlEncoderStatus JxlEncoderSetExtraChannelInfo(
^
/Users/z/git/libjxl/lib/extras/enc/jxl.cc:121:32: error: cannot initialize a parameter of type 'JxlEncoder *' (aka 'JxlEncoderStruct *') with an lvalue of type 'JxlEncoderFrameSettings *' (a
ka 'JxlEncoderFrameSettingsStruct *')
settings, 0, params.alpha_distance)) {
^~~~~~~~
/opt/homebrew/include/jxl/encode.h:52:16: note: 'JxlEncoderFrameSettingsStruct' is not defined, but forward declared here; conversion would be valid if it was derived from 'JxlEncoderStruct'
typedef struct JxlEncoderFrameSettingsStruct JxlEncoderFrameSettings;
^
/opt/homebrew/include/jxl/encode.h:913:17: note: passing argument to parameter 'enc' here
JxlEncoder* enc, size_t index, const JxlExtraChannelInfo* info);
^
2 errors generated.
make[2]: *** [lib/CMakeFiles/jxl_extras-static.dir/extras/enc/jxl.cc.o] Error 1
make[1]: *** [lib/CMakeFiles/jxl_extras-static.dir/all] Error 2
make: *** [all] Error 2
```
<img width="1779" alt="image" src="https://user-images.githubusercontent.com/127726508/230264975-4093433f-5093-47a0-a4eb-1df04aa378ea.png">
| compile error in mac m2 : error: use of undeclared identifier 'JxlEncoderSetExtraChannelDistance | https://api.github.com/repos/libjxl/libjxl/issues/2359/comments | 2 | 2023-04-06T03:28:35Z | 2023-04-07T04:20:55Z | https://github.com/libjxl/libjxl/issues/2359 | 1,656,606,195 | 2,359 |
[
"libjxl",
"libjxl"
] | **Is your feature request related to a problem? Please describe.**
Neither jxl-x64-windows-static.zip nor jxl-x86-windows-static.zip actually contain headers and link libraries.
Also they only contain DLLs and dynamically linked binaries. Why are you calling the zipfiles "static"?
**Describe the solution you'd like**
jxl-x64-windows-static.zip and jxl-x86-windows-static.zip should contain link libraries and headers. They should contain static link libraries (if they contain DLLs also, that's OK I guess).
**Describe alternatives you've considered**
**Additional context**
| Why doesn't jxl-x64-windows-static.zip contain static libraries and headers? Or any headers at all? | https://api.github.com/repos/libjxl/libjxl/issues/2333/comments | 5 | 2023-03-26T21:56:47Z | 2024-03-04T10:24:20Z | https://github.com/libjxl/libjxl/issues/2333 | 1,641,107,120 | 2,333 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
Trying to build vips for x86_64 macos using osxcross, I'm getting a link error, see below. Can't find where that symbol comes from.
```
Undefined symbols for architecture x86_64:
"___cpu_model", referenced from:
_JxlFastLosslessWriteOutput in libjxl.a(enc_fast_lossless.cc.o)
_JxlFastLosslessEncode in libjxl.a(enc_fast_lossless.cc.o)
_JxlFastLosslessPrepareFrame in libjxl.a(enc_fast_lossless.cc.o)
```
This does not happen when building for aarch64 macos in the same conditions.
**To Reproduce**
Generate a cross toolchain using osxcross, build static libjxl and deps, and try to link vips against them.
**Expected behavior**
Linking should succeed like it does for aarch64.
**Screenshots**
N/A
**Environment**
- OS: Arch Linux
- Compiler version: clang 15.0.7
- CPU type: x86_64
- cjxl/djxl version string: 0.8.0
**Additional context**
Happy to provide additional information if needed.
| Cross-compiling vips for x86_64 macos fails when linking against libjxl | https://api.github.com/repos/libjxl/libjxl/issues/2332/comments | 2 | 2023-03-26T15:12:12Z | 2023-06-20T16:49:42Z | https://github.com/libjxl/libjxl/issues/2332 | 1,640,950,866 | 2,332 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
When encoding losless images with the fast lossless [encoder](https://github.com/libjxl/libjxl/tree/main/experimental/fast_lossless), the encoder creates images the decoder cannot parse
**To Reproduce**
The below image
was used as the test sample
**Expected behavior**
An image encoded to` .jxl` which could be decoded back to the expected input using the fast encoder
- OS: Manjaro Linux
- Compiler version: GCC 12.2.1
- CPU type: x86_64
**Additional context**
- Removing the alpha channel makes the encoder produce a working image.
- The cjxl encoder generates a decodable image.
the first generated bytes are
|cjxl|fast_lossless|
|-----|-----------------|
|ff 0a fa 15 e0 3f 58 09 08 08 10 00 84 02 00 00| ff 0a fa 15 e0 3f 58 14 37 02 08 08 10 00 30 0b|
The only weird behaviour I could note was that jxlinfo interpreted the output as RGB+A
Command
```sh
jxlinfo -v ./a.jxl
```
output was
```text
JPEG XL image, 512x704, (possibly) lossless, 8-bit RGB+Alpha
num_color_channels: 3
num_extra_channels: 1
extra channel 0:
type: Alpha
bits_per_sample: 8
alpha_premultiplied: 0 (Non-premultiplied)
have_preview: 0
have_animation: 0
Intrinsic dimensions: 512x704
Orientation: 1 (Normal)
Color space: RGB, D65, sRGB primaries, sRGB transfer function, rendering intent: Relative
```
There is also a huge difference in size, when using cjxl and fast encoder, (53k vs 33k), is there something I'm missing?
**sample outputs**
- wrong-image: Output of what I got after encoding with fast lossless encoder
- correct-working-jxl-image: Working image
- input.ppm.txt: Input image, P7 ppm format in grayscale+alpha
[input.ppm.txt](https://github.com/libjxl/libjxl/files/11065827/a.ppm.txt)
| Fast lossless incorrectly encodes grayscale+alpha image | https://api.github.com/repos/libjxl/libjxl/issues/2331/comments | 3 | 2023-03-24T19:19:07Z | 2023-03-31T10:10:04Z | https://github.com/libjxl/libjxl/issues/2331 | 1,639,915,501 | 2,331 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
Build fails.
[ 68%] Building CXX object lib/CMakeFiles/jpegli-static.dir/jpegli/color_transform.cc.o
/home/work/jxl/libjxl/lib/jpegli/color_transform.cc: In function ‘bool jpegli::CheckColorSpaceComponents(int, J_COLOR_SPACE)’:
/home/work/jxl/libjxl/lib/jpegli/color_transform.cc:138:10: error: ‘JCS_EXT_RGB’ was not declared in this scope; did you mean ‘JCS_RGB’?
138 | case JCS_EXT_RGB:
| ^~~~~~~~~~~
| JCS_RGB
/home/work/jxl/libjxl/lib/jpegli/color_transform.cc:139:10: error: ‘JCS_EXT_BGR’ was not declared in this scope
139 | case JCS_EXT_BGR:
| ^~~~~~~~~~~
/home/work/jxl/libjxl/lib/jpegli/color_transform.cc:143:10: error: ‘JCS_EXT_RGBX’ was not declared in this scope
143 | case JCS_EXT_RGBX:
| ^~~~~~~~~~~~
/home/work/jxl/libjxl/lib/jpegli/color_transform.cc:144:10: error: ‘JCS_EXT_BGRX’ was not declared in this scope
144 | case JCS_EXT_BGRX:
| ^~~~~~~~~~~~
/home/work/jxl/libjxl/lib/jpegli/color_transform.cc:145:10: error: ‘JCS_EXT_XBGR’ was not declared in this scope
145 | case JCS_EXT_XBGR:
| ^~~~~~~~~~~~
**To Reproduce**
git clone https://github.com/libjxl/libjxl
cd libjxl
./deps.sh
mkdir build
cd build
cmake -DBUILD_TESTING=OFF ..
make
**Expected behavior**
It should compile without issues.
**Environment**
Ubuntu 20.04 clean install.
| Build fails on Ubuntu 20.04 | https://api.github.com/repos/libjxl/libjxl/issues/2320/comments | 3 | 2023-03-21T01:14:40Z | 2024-03-12T08:13:24Z | https://github.com/libjxl/libjxl/issues/2320 | 1,633,087,468 | 2,320 |
[
"libjxl",
"libjxl"
] | /libjxl-main/tools/file_io.cc line 62
Code:
if (fwrite(bytes.data(), 1, bytes.size(), file) != bytes.size()) {
fprintf(stderr,
"Could not write to file\n"
"Error: %s",
strerror(errno));
// <- may be better add fclose(file);
return false;
}
| missing fclose | https://api.github.com/repos/libjxl/libjxl/issues/2308/comments | 0 | 2023-03-17T10:18:32Z | 2023-03-20T08:20:39Z | https://github.com/libjxl/libjxl/issues/2308 | 1,629,056,565 | 2,308 |
[
"libjxl",
"libjxl"
] | libjxl-main/tools/cjxl_main.cc line 665
Code:
if (((distance_set && (args->distance != 0.0)) ||
(quality_set && (args->quality != 100))) &&
args->lossless_jpeg && args->lossless_jpeg) { // <-- identical
| identical expression | https://api.github.com/repos/libjxl/libjxl/issues/2304/comments | 1 | 2023-03-16T14:45:44Z | 2023-03-17T07:49:38Z | https://github.com/libjxl/libjxl/issues/2304 | 1,627,664,498 | 2,304 |
[
"libjxl",
"libjxl"
] | error log:
/Applications/Xcode_14.2.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/lib/clang/14.0.0/include/avx512fintrin.h:9586:1: note: previous definition is here
_mm512_cvtsi512_si32(__m512i __A) {
^
/Users/runner/work/mpv-macos-intel/mpv-macos-intel/packages/libjxl/lib/jxl/enc_fast_lossless.cc:37:1: error: redefinition of '_mm512_cvtsi512_si32'
_mm512_cvtsi512_si32(__m512i __A) {
^
/Applications/Xcode_14.2.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/lib/clang/14.0.0/include/avx512fintrin.h:9586:1: note: previous definition is here
_mm512_cvtsi512_si32(__m512i __A) {
^
3 warnings and 1 error generated.
| build failed on mac Monterey | https://api.github.com/repos/libjxl/libjxl/issues/2291/comments | 0 | 2023-03-13T06:22:07Z | 2023-04-17T09:38:51Z | https://github.com/libjxl/libjxl/issues/2291 | 1,620,830,260 | 2,291 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
I saved a 598KB jpg file to a PNG 3.2MB loslessly using photoshop. Then using CJXL I converted the PNG to jxl using --distance 0 --effort 9 and it resulted in a 2.09MB JXL file that is still bigger than the JPG orignal file.
**To Reproduce**
convert jpg to png
convert the png to jxl using --distance 0 --effort 9
**Expected behavior**
expected the jxl file to have a smaller size than the orignal JPG file
**Environment**
- OS: Windows 10
- Compiler version: [e.g. clang 11.0.1]
- CPU type: 13900k
- cjxl/djxl version string: v0.8.1
**Additional context**
I originally tried to convert the jpg to JXL but it never comes out pixel to pixel lossless and there is always a difference in pixel color using --distance 0 --effort 9
<!--
Currently github does not allow uploading files that end in `.jxl`, but when you
rename them for example as `image.jxl.jpg`, it will be possible to upload them
and also view them in browsers that are configured to support it.
See https://github.com/orgs/github-community/discussions/18139
-->
| Saved jpg to PNG, then converted to JXL resulting in bigger file size | https://api.github.com/repos/libjxl/libjxl/issues/2290/comments | 2 | 2023-03-11T21:10:58Z | 2023-03-11T21:31:33Z | https://github.com/libjxl/libjxl/issues/2290 | 1,620,146,831 | 2,290 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
When converting a particular JXL image to PNG, the PNG has noticeable color issues with patches of discoloration.
**To Reproduce**
Download [this](https://cdn.discordapp.com/attachments/794206170445119489/1083872278128885790/imagetest15-Q70.jxl)
djxl theimage.jxl out.png
Compare the png to the JXL.
**Expected behavior**
The PNG would be visually identical to the JXL, or at least close enough where I wouldn't see obvious discoloration or where the PNG is earning horrible ssimu2 scores.
**Screenshots**
If applicable, add screenshots or example input/output images to help explain your problem.
**Environment**
- OS: Linux
- Compiler version: gcc (GCC) 12.2.1 20230201
- CPU type: x86_64
- cjxl/djxl version string: djxl v0.9.0 13caf6f8 [AVX2,SSE4,SSSE3,SSE2]
**Additional context**
The problem persists with other versions of cjxl/djxl when using the source PNG (which is almost 100mb).
<!--
Currently github does not allow uploading files that end in `.jxl`, but when you
rename them for example as `image.jxl.jpg`, it will be possible to upload them
and also view them in browsers that are configured to support it.
See https://github.com/orgs/github-community/discussions/18139
--> | djxl PNG Conversion Acting Up | https://api.github.com/repos/libjxl/libjxl/issues/2289/comments | 31 | 2023-03-11T05:10:51Z | 2025-06-21T05:31:10Z | https://github.com/libjxl/libjxl/issues/2289 | 1,619,886,499 | 2,289 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
Running benchmark_xl with --debug_image_dir set to any path will result in `Error in jxl codec
D:\a\libjxl\libjxl\tools\benchmark\benchmark_xl.cc:144: JXL_CHECK: speed_stats.GetSummary(&summary)`
Tested on the current stable release of 0.8.0 and also 0.9.0
**To Reproduce**
Run `benchmark_xl --input=Test.png --debug_image_dir=(Anything)`
**Expected behavior**
A debug image to be generated inside the provided directory
**Environment**
- OS: Windows 10
- CPU type: x86_64
- cjxl/djxl version string: benchmark_xl v0.9.0 13caf6f [AVX2,SSE4,SSSE3,SSE2]
<!--
Currently github does not allow uploading files that end in `.jxl`, but when you
rename them for example as `image.jxl.jpg`, it will be possible to upload them
and also view them in browsers that are configured to support it.
See https://github.com/orgs/github-community/discussions/18139
-->
| Benchmark_xl fails on Windows when --debug_image_dir is used | https://api.github.com/repos/libjxl/libjxl/issues/2287/comments | 1 | 2023-03-10T10:09:40Z | 2025-01-18T19:24:22Z | https://github.com/libjxl/libjxl/issues/2287 | 1,618,700,113 | 2,287 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
A legacy JPEG file produced with benchmark_xl/jpegli should be recompressible with cjxl using standard JPEG transcoding. It fails.
**To Reproduce**
```
$ benchmark_xl --save_compressed --output_dir=. --codec=jpeg:enc-jpegli:xyb:d1 --input=lenna.png
```
followed by:
```
$ cjxl lenna.png.d1.xyb.enc-jpegli.jpeg lenna.jxl
JPEG XL encoder v0.9.0 13caf6f8 [AVX2]
Note: Implicit-default for JPEG is lossless-transcoding. To silence this message, set --lossless_jpeg=(1|0).
Read JPEG image with 54804 bytes.
Encoding [Container | JPEG, lossless transcode, effort: 7 | JPEG reconstruction data],
./lib/jxl/enc_frame.cc:322: JXL_FAILURE: Chroma subsampling is not supported when color transform is not YCbCr
./lib/jxl/enc_frame.cc:1275: JXL_RETURN_IF_ERROR code=1: MakeFrameHeader(cparams, passes_enc_state->progressive_splitter, frame_info, ib, frame_header.get())
./lib/jxl/encode.cc:569: Failed to encode frame
JxlEncoderProcessOutput failed.
EncodeImageJXL() failed.
```
**Expected behavior**
It should produce a recompressed legacy JPEG as a JXL file.
**Environment**
- OS: Arch Linux x86_64
- Compiler version: GCC 12.2.1
- cjxl/djxl version string: JPEG XL encoder v0.9.0 13caf6f8 [AVX2]
**Additional context**
XYB jpeg attached. | jpegli output fails to be JPEG-recompressible using cjxl | https://api.github.com/repos/libjxl/libjxl/issues/2284/comments | 11 | 2023-03-09T15:34:07Z | 2025-01-19T23:33:31Z | https://github.com/libjxl/libjxl/issues/2284 | 1,617,481,085 | 2,284 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
I've found another input that upsets the fast lossless encoder.
Encoding with `cjxl -e 1 -d 0` and decoding again, the result is:
Similar to #2118, minor variations on this image can produce JXLs that don't decode at all. e.g.
```
$ djxl b.jxl b-decoded.png
JPEG XL decoder v0.9.0 49cd0322 [AVX2,SSE4,SSSE3,SSE2]
Read 3221 compressed bytes.
./lib/jxl/modular/encoding/encoding.cc:566: JXL_FAILURE: Truncated input
./lib/jxl/modular/encoding/encoding.cc:597: JXL_RETURN_IF_ERROR code=1: dec_status
./lib/jxl/dec_modular.cc:357: JXL_RETURN_IF_ERROR code=1: status
./lib/jxl/dec_frame.cc:489: JXL_RETURN_IF_ERROR code=1: modular_frame_decoder_.DecodeGroup( mrect, br[i - pass0], minShift, maxShift, ModularStreamId::ModularAC(ac_group_id, i), false, dec_state_, &render_pipeline_input, false, &modular_pass_ready)
./lib/jxl/dec_frame.cc:716: JXL_FAILURE: Error in AC group
./lib/jxl/decode.cc:1143: frame out of bounds
Failed to decode image
DecompressJxlToPackedPixelFile failed
```
**To Reproduce**
- `cjxl a.png a.jxl -d 0 -e 1`
- Attempt to decode a.jxl
**Environment**
- OS: Linux
- Compiler version: clang 14.0.0
- CPU type: x86_64
- cjxl/djxl version string: JPEG XL encoder v0.9.0 49cd0322 [AVX2,SSE4,SSSE3,SSE2]
<!--
Currently github does not allow uploading files that end in `.jxl`, but when you
rename them for example as `image.jxl.jpg`, it will be possible to upload them
and also view them in browsers that are configured to support it.
See https://github.com/orgs/github-community/discussions/18139
-->
| Fast lossless can produce invalid/incorrect JXLs | https://api.github.com/repos/libjxl/libjxl/issues/2281/comments | 1 | 2023-03-08T19:07:08Z | 2023-03-10T21:19:31Z | https://github.com/libjxl/libjxl/issues/2281 | 1,615,777,088 | 2,281 |
[
"libjxl",
"libjxl"
] | Release builds for Windows are much slower than static Clang builds, especially for fast speeds
ProcProfile64:
clang v0.9.0 [cjxl e1 lossless]
`119,771,153 -> 52,094,805: 43.49%. Cpu 349 mb/s (0.327 sec), real 318 mb/s (0.359 sec) = 91%. ram 401144 KB, vmem 569556 KB`
git build v0.8.1 [cjxl e1 lossless]
`119,771,153 -> 51,493,245: 42.99%. Cpu 78 mb/s (1.453 sec), real 78 mb/s (1.452 sec) = 100%. ram 402340 KB, vmem 569940 KB` | The official Windows builds are much slower than the static Clang builds | https://api.github.com/repos/libjxl/libjxl/issues/2268/comments | 8 | 2023-03-06T16:10:22Z | 2024-03-05T14:54:54Z | https://github.com/libjxl/libjxl/issues/2268 | 1,611,757,354 | 2,268 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
Encoding .jpegs with the progressive option crashes cjxl.
**To Reproduce**
I'm attaching the .jpeg I used for the sake of documenting it, but this error occurs with many (all?) jpegs, on cjxl versions `v0.8.1 c27d4992` and `v0.9.0 6fb35528` (latest as of 5th March, 23). Binaries are static linux binaries from https://artifacts.lucaversari.it/libjxl/libjxl.
[turtle.zip](https://github.com/libjxl/libjxl/files/10891999/turtle.zip)
```
$: ./cjxl turtle.jpg turtle.jxl -v --progressive
JPEG XL encoder v0.9.0 6fb35528 [AVX2,SSE4,SSSE3,SSE2]
Note: Implicit-default for JPEG is lossless-transcoding. To silence this message, set --lossless_jpeg=(1|0).
Read JPEG image with 225719 bytes.
Encoding [Container | JPEG, lossless transcode, effort: 7 | JPEG reconstruction data],
./lib/jxl/enc_frame.cc:908: JXL_CHECK: enc_state_->passes.size() == 1
fish: Job 1, './cjxl turtle.jpg turtle.jxl -v…' terminated by signal SIGILL (Illegal instruction)
```
**Expected behavior**
If progressively and losslessly encoding a .jpeg is not possible, cjxl should inform the user it's not possible and exit, instead of crashing.
**Environment**
- OS: `Linux pop-os 6.1.11-76060111-generic #202302091138~1675975749~22.04~f771a7f SMP PREEMPT_DYNAMIC Thu F x86_64 GNU/Linux`
- CPU type: x86_64
- cjxl version string: `JPEG XL encoder v0.9.0 6fb35528 [AVX2,SSE4,SSSE3,SSE2]`
**Related**
- https://github.com/libjxl/libjxl/issues/1871
- https://github.com/libjxl/libjxl/pull/1814 (Suggests progressive .jpeg encoding is possible)
- https://github.com/libjxl/libjxl/pull/1820
| Progressive encoding JPEGs crashes CJXL | https://api.github.com/repos/libjxl/libjxl/issues/2265/comments | 2 | 2023-03-05T14:38:13Z | 2023-05-16T11:31:13Z | https://github.com/libjxl/libjxl/issues/2265 | 1,610,225,511 | 2,265 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
When decoding an image compressed with `--progressive_ac` and `--progressive_dc=1` the `libjxl` decoder generates the same number of `JXL_DEC_FRAME_PROGRESSION` events for both `kLastPasses` and `kPasses` progressive detail options. In both of those cases flushing the output buffer after receiving the event generates the same bytes. The `kDC` progressive detail option generates one event (as expected) and if the buffer is flushed after the event, its content will match bytes given by the first `JXL_DEC_FRAME_PROGRESSION` event when other progressive detail options are used.
**To Reproduce**
1. Use `cjxl` with `--progressive_ac` and `--progressive_dc=1` options to compress an image.
2. Use `JxlDecoderSubscribeEvent` to subscribe to the `JXL_DEC_FRAME_PROGRESSION` event.
3. Use `JxlDecoderSetProgressiveDetail` to `kPasses` or `kLastPasses`.
4. Decode the image and observe the number of `JXL_DEC_FRAME_PROGRESSION` events.
5. Optionally flush the output buffer and write it to disc.
6. Decode the image again, this time using the other progressive detail option.
7. Optionally compare the output buffers.
**Expected behavior**
Decoding an image compressed with `cjxl`'s `--progressive_ac` and `--progressive_dc=1` options using a Decoder with `kPasses` progressive detail option generates one more `JXL_DEC_FRAME_PROGRESSION` event than decoding with `kLastPasses` option.
**Screenshots**
[The sample image I used to test this](https://github.com/libjxl/libjxl/files/10889806/sample.jxl.gz)
If it is needed, I can also provide the output buffer content after each event.
**Environment**
- OS: `Linux 6.1.12-gentoo`
- Compiler version: `gcc 12.2.1`
- CPU type: `amd64`
- cjxl/djxl version string: `djxl v0.8.0 [SSSE3]`
**Additional context**
Using `--qprogressive_ac` instead of `--progressive_ac` during compression causes the decoder to generate only two `JXL_DEC_FRAME_PROGRESSION` events. I also did not get any `JXL_DEC_FRAME_PROGRESSION` events when trying to decode an image encoded using JPEG XL's modular mode, but I am not sure if the progressive decoding of modular images is supported at all. | Unexpected number of `JXL_DEC_FRAME_PROGRESSION` events generated when using `kPasses` progressive detail decoder option | https://api.github.com/repos/libjxl/libjxl/issues/2264/comments | 0 | 2023-03-04T22:49:59Z | 2023-05-10T11:54:26Z | https://github.com/libjxl/libjxl/issues/2264 | 1,609,975,841 | 2,264 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
Compilation error at enc_fast_lossless.cc, preventing installation with brew on MacOS 10.13.6.
[ 55%] Building CXX object lib/CMakeFiles/jxl_enc-obj.dir/jxl/enc_noise.cc.o
cd /tmp/jpeg-xl-20230303-35201-11dm0kk/libjxl-0.8.1/build/lib && /usr/local/Homebrew/Library/Homebrew/shims/mac/super/clang++ -DHWY_DISABLED_TARGETS="(HWY_SVE|HWY_SVE2|HWY_SVE_256|HWY_SVE2_128|HWY_RVV)" -DJPEGXL_MAJOR_VERSION=0 -DJPEGXL_MINOR_VERSION=8 -DJPEGXL_PATCH_VERSION=1 -DJXL_INTERNAL_LIBRARY_BUILD -D__DATE__=\"redacted\" -D__TIMESTAMP__=\"redacted\" -D__TIME__=\"redacted\" -I/tmp/jpeg-xl-20230303-35201-11dm0kk/libjxl-0.8.1 -I/tmp/jpeg-xl-20230303-35201-11dm0kk/libjxl-0.8.1/lib/include -I/tmp/jpeg-xl-20230303-35201-11dm0kk/libjxl-0.8.1/third_party/skcms -I/tmp/jpeg-xl-20230303-35201-11dm0kk/libjxl-0.8.1/build/lib/include -fno-rtti -funwind-tables -Xclang -mrelax-all -Xclang -mconstructor-aliases -fno-omit-frame-pointer -O3 -DNDEBUG -O2 -fPIC -fvisibility=hidden -fvisibility-inlines-hidden -Wno-builtin-macro-redefined -Wall -fmerge-all-constants -fno-builtin-fwrite -fno-builtin-fread -Wextra -Wc++11-compat -Warray-bounds -Wformat-security -Wimplicit-fallthrough -Wno-register -Wno-unused-function -Wno-unused-parameter -Wnon-virtual-dtor -Woverloaded-virtual -Wvla -Wdeprecated-increment-bool -Wfloat-overflow-conversion -Wfloat-zero-conversion -Wfor-loop-analysis -Wgnu-redeclared-enum -Winfinite-recursion -Wliteral-conversion -Wno-c++98-compat -Wno-unused-command-line-argument -Wprivate-header -Wself-assign -Wstring-conversion -Wtautological-overlap-compare -Wthread-safety-analysis -Wundefined-func-template -Wunreachable-code -Wunused-comparison -fsized-deallocation -fno-exceptions -fmath-errno -fnew-alignment=8 -fno-cxx-exceptions -fno-slp-vectorize -fno-vectorize -disable-free -disable-llvm-verifier -DJPEGXL_ENABLE_SKCMS=1 -DJPEGXL_BUNDLE_SKCMS=1 -DJPEGXL_ENABLE_TRANSCODE_JPEG=1 -DJPEGXL_ENABLE_BOXES=1 -std=c++11 -MD -MT lib/CMakeFiles/jxl_enc-obj.dir/jxl/enc_noise.cc.o -MF CMakeFiles/jxl_enc-obj.dir/jxl/enc_noise.cc.o.d -o CMakeFiles/jxl_enc-obj.dir/jxl/enc_noise.cc.o -c /tmp/jpeg-xl-20230303-35201-11dm0kk/libjxl-0.8.1/lib/jxl/enc_noise.cc
In file included from /tmp/jpeg-xl-20230303-35201-11dm0kk/libjxl-0.8.1/lib/jxl/enc_fast_lossless.cc:3768:
/tmp/jpeg-xl-20230303-35201-11dm0kk/libjxl-0.8.1/lib/jxl/enc_fast_lossless.cc:768:33: error: use of undeclared identifier '_cvtmask16_u32'
return CtzNonZero(~uint64_t{_cvtmask16_u32(mask)});
Suppose this issue was caused by "~uint64_t", but changing it caused next issue, possible not related to this one:
[ 55%] Building CXX object lib/CMakeFiles/jxl_enc-obj.dir/jxl/enc_xyb.cc.o
cd /tmp/jpeg-xl-20230303-35201-11dm0kk/libjxl-0.8.1/build/lib && /usr/local/Homebrew/Library/Homebrew/shims/mac/super/clang++ -DHWY_DISABLED_TARGETS="(HWY_SVE|HWY_SVE2|HWY_SVE_256|HWY_SVE2_128|HWY_RVV)" -DJPEGXL_MAJOR_VERSION=0 -DJPEGXL_MINOR_VERSION=8 -DJPEGXL_PATCH_VERSION=1 -DJXL_INTERNAL_LIBRARY_BUILD -D__DATE__=\"redacted\" -D__TIMESTAMP__=\"redacted\" -D__TIME__=\"redacted\" -I/tmp/jpeg-xl-20230303-35201-11dm0kk/libjxl-0.8.1 -I/tmp/jpeg-xl-20230303-35201-11dm0kk/libjxl-0.8.1/lib/include -I/tmp/jpeg-xl-20230303-35201-11dm0kk/libjxl-0.8.1/third_party/skcms -I/tmp/jpeg-xl-20230303-35201-11dm0kk/libjxl-0.8.1/build/lib/include -fno-rtti -funwind-tables -Xclang -mrelax-all -Xclang -mconstructor-aliases -fno-omit-frame-pointer -O3 -DNDEBUG -O2 -fPIC -fvisibility=hidden -fvisibility-inlines-hidden -Wno-builtin-macro-redefined -Wall -fmerge-all-constants -fno-builtin-fwrite -fno-builtin-fread -Wextra -Wc++11-compat -Warray-bounds -Wformat-security -Wimplicit-fallthrough -Wno-register -Wno-unused-function -Wno-unused-parameter -Wnon-virtual-dtor -Woverloaded-virtual -Wvla -Wdeprecated-increment-bool -Wfloat-overflow-conversion -Wfloat-zero-conversion -Wfor-loop-analysis -Wgnu-redeclared-enum -Winfinite-recursion -Wliteral-conversion -Wno-c++98-compat -Wno-unused-command-line-argument -Wprivate-header -Wself-assign -Wstring-conversion -Wtautological-overlap-compare -Wthread-safety-analysis -Wundefined-func-template -Wunreachable-code -Wunused-comparison -fsized-deallocation -fno-exceptions -fmath-errno -fnew-alignment=8 -fno-cxx-exceptions -fno-slp-vectorize -fno-vectorize -disable-free -disable-llvm-verifier -DJPEGXL_ENABLE_SKCMS=1 -DJPEGXL_BUNDLE_SKCMS=1 -DJPEGXL_ENABLE_TRANSCODE_JPEG=1 -DJPEGXL_ENABLE_BOXES=1 -std=c++11 -MD -MT lib/CMakeFiles/jxl_enc-obj.dir/jxl/enc_xyb.cc.o -MF CMakeFiles/jxl_enc-obj.dir/jxl/enc_xyb.cc.o.d -o CMakeFiles/jxl_enc-obj.dir/jxl/enc_xyb.cc.o -c /tmp/jpeg-xl-20230303-35201-11dm0kk/libjxl-0.8.1/lib/jxl/enc_xyb.cc
/tmp/jpeg-xl-20230303-35201-11dm0kk/libjxl-0.8.1/lib/jxl/enc_fast_lossless.cc:369:7: error: invalid cpu feature string for builtin
if (__builtin_cpu_supports("avx512vbmi2")) {
**To Reproduce**
Steps to reproduce the behavior:
brew install --debug --verbose jpeg-xl
**Expected behavior**
Successful installation
**Environment**
- OS: MacOS 10.13.6
- Compiler version: clang 10.0.1
- CPU type: x86_64
- cjxl/djxl version string: N/A
| Compilation errror enc_fast_lossless.cc | https://api.github.com/repos/libjxl/libjxl/issues/2260/comments | 0 | 2023-03-03T11:47:12Z | 2023-04-17T13:30:49Z | https://github.com/libjxl/libjxl/issues/2260 | 1,608,490,628 | 2,260 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
HDR lossy XYB-encoded JXL images are too dark when sRGB is requested from libjxl.
**To Reproduce**
Take an HDR image, XYB-encoded (example attached). Decode it to sRGB with:
```
$ djxl --color_space=RGB_D65_SRG_Rel_SRG hdr.jxl sdr.png
```
and view the resulting sRGB PNG with any PNG viewer. The image will be too dark.
Then, decode it to a PQ PNG with:
```
$ djxl hdr.jxl hdr.png
```
(this image is tagged as BT.2100/PQ so the explicit color space does not need to be requested).
View that PQ png in Chromium, mpv, Preview, or any ICC-aware viewer that tonemaps correctly. The colors will be much more accurately bright and not dark.
**Expected behavior**
The SDR should look like an SDR approximation of the HDR image. Some tradeoffs need to be made, but darkening the whole image is definitely undesirable.
Changing the rendering intent from Rel to `Abs`, `Sat`, or `Per` didn't fix the issue.
**Screenshots**
^ the above `.jpg` is a JXL file
`sdr.png` is 34 MiB but you can view it here: https://files.catbox.moe/9hjw4u.png
**Environment**
djxl 60ab29d7
```
- OS: Arch Linux x86_64
- Compiler version: gcc 12.2.1
- cjxl/djxl version string: djxl v0.9.0 60ab29d7 [AVX2]
``
| libjxl produces too-dark HDR images when sRGB is requested | https://api.github.com/repos/libjxl/libjxl/issues/2251/comments | 4 | 2023-03-01T16:00:24Z | 2025-01-18T19:16:24Z | https://github.com/libjxl/libjxl/issues/2251 | 1,605,240,560 | 2,251 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
running tests from 0.8.1 on OpenBSD, a single test timeouts and is killed by cmake. This is 100% reproducible.
**To Reproduce**
running make test in the wip port for libjxl 0.8.1
**Screenshots**
```
test 740
Start 740: ConvolveTestGroup/ConvolveTest.TestConvolve/EMU128 # GetParam() = 2305843009213693952
740: Test command: /usr/obj/ports/libjxl-0.8.1/build-amd64/lib/tests/convolve_test "--gtest_filter=ConvolveTestGroup/ConvolveTest.TestConvolve/EMU128" "--gtest_also_run_disabled_tests"
740: Working Directory: /usr/obj/ports/libjxl-0.8.1/build-amd64/lib
740: Test timeout computed to be: 1500
740: Running main() from /usr/obj/ports/gtest-1.11.0pl20220208/googletest-06519cedc3159de8b36a504766ad6b7966555f10/googletest/src/gtest_main.cc
740: Note: Google Test filter = ConvolveTestGroup/ConvolveTest.TestConvolve/EMU128
740: [==========] Running 1 test from 1 test suite.
740: [----------] Global test environment set-up.
740: [----------] 1 test from ConvolveTestGroup/ConvolveTest
740: [ RUN ] ConvolveTestGroup/ConvolveTest.TestConvolve/EMU128
740/3489 Test #740: ConvolveTestGroup/ConvolveTest.TestConvolve/EMU128 # GetParam() = 2305843009213693952 .................................................................................................................................................***Timeout 1500.03 sec
```
**Environment**
- OS: OpenBSD
- Compiler version: clang 13
- CPU type: amd64, arm64
- cjxl/djxl version string: 0.8.1 | ConvolveTestGroup/ConvolveTest.TestConvolve/EMU128 timeouts on OpenBSD | https://api.github.com/repos/libjxl/libjxl/issues/2243/comments | 1 | 2023-03-01T08:55:03Z | 2023-03-01T11:01:16Z | https://github.com/libjxl/libjxl/issues/2243 | 1,604,539,287 | 2,243 |
[
"libjxl",
"libjxl"
] | People seem to be talking about this new higher-quality JPG thing, but the README barely talk about what it is, what it does, how well it does, and how it achieves everything.
Might… be a good idea to put that on the roadmap. Gotta have something to cite for Wikipedia! | [REQ] Want a stronger README of jpegli | https://api.github.com/repos/libjxl/libjxl/issues/2241/comments | 3 | 2023-02-28T21:18:32Z | 2025-01-19T04:43:13Z | https://github.com/libjxl/libjxl/issues/2241 | 1,603,829,837 | 2,241 |
[
"libjxl",
"libjxl"
] | Can jpeg xl support the whole platform？as ios 、android、linux、windows、mac 。 | Which platforms are supported？ | https://api.github.com/repos/libjxl/libjxl/issues/2228/comments | 1 | 2023-02-24T10:17:12Z | 2024-04-17T08:43:44Z | https://github.com/libjxl/libjxl/issues/2228 | 1,598,331,832 | 2,228 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
Remove 1.0-snapshot tag, it confuses naive reader, and Debian bot into thinking that 1.0-snapshot tag is most recent that 0.8.1 release.
**To Reproduce**
* https://github.com/libjxl/libjxl/releases/tag/v0.8.1
* https://github.com/libjxl/libjxl/releases/tag/v1.0-snapshot
**Expected behavior**
v0.8.1 should appear as latest release on page:
* https://tracker.debian.org/pkg/jpeg-xl
**Screenshots**
| Remove 1.0-snapshot | https://api.github.com/repos/libjxl/libjxl/issues/2216/comments | 5 | 2023-02-23T09:04:48Z | 2025-01-23T20:37:47Z | https://github.com/libjxl/libjxl/issues/2216 | 1,596,471,804 | 2,216 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
When you compile `benchmark_xl` with `-D_GLIBCXX_ASSERTIONS`, as is done by default on Arch, it often fails to encode with jpegli.
Error is listed as below:
```
❯ ~/dev/libjxl/build/tools/benchmark_xl --input=jpegxl-home.jpg --codec=jpeg:enc-jpegli:xyb:q87 --save_compressed --output_dir=jpegxl-home
benchmark_xl v0.9.0 7566e4c8 [AVX2,SSE4,SSSE3,Unknown]
8 total threads, 1 tasks, 0 threads, 8 inner threads
/usr/sbin/../lib64/gcc/x86_64-pc-linux-gnu/12.2.1/../../../../include/c++/12.2.1/bits/stl_vector.h:1123: std::vector::reference std::vector<jpegli::JPEGHuffmanCode>::operator[](std::vector::size_type) [_Tp = jpegli::JPEGHuffmanCode, _Alloc = std::allocator<jpegli::JPEGHuffmanCode>]: Assertion '__n < this->size()' failed.
[1] 11439 IOT instruction ~/dev/libjxl/build/tools/benchmark_xl --input=jpegxl-home.jpg
```
**To Reproduce**
Use any image, for exact reproduction use the image at https://jpeg.org/images/jpegxl-home.jpg (JPEG XL info page at JPEG website).
Compile `benchmark_xl` with `-D_GLIBCXX_ASSERTIONS`, for example with `-DCMAKE_CXX_FLAGS="-D_GLIBCXX_ASSERTIONS"`.
Try to encode using `benchmark_xl` with codec `jpeg:enc-jpegli:xyb:q87`, changing the quality if you want. Most of the time (but not all the time, for example if you change the quality), the encoding will error out.
**Expected behavior**
`benchmark_xl` encodes successfully whether `_GLIBCXX_ASSERTIONS` is defined or not.
**Environment**
- OS: Arch on WSL 2
- Compiler version: clang 15.0.7
- CPU type: x86_64
- cjxl/djxl version string: JPEG XL encoder v0.9.0 7566e4c8 [AVX2,SSE4,SSSE3,Unknown]
<!--
Currently github does not allow uploading files that end in `.jxl`, but when you
rename them for example as `image.jxl.jpg`, it will be possible to upload them
and also view them in browsers that are configured to support it.
See https://github.com/orgs/github-community/discussions/18139
-->
| benchmark_xl fails to encode with jpegli when compiled with `-D_GLIBCXX_ASSERTIONS` | https://api.github.com/repos/libjxl/libjxl/issues/2207/comments | 3 | 2023-02-21T05:38:37Z | 2023-02-22T23:58:57Z | https://github.com/libjxl/libjxl/issues/2207 | 1,592,844,641 | 2,207 |
[
"libjxl",
"libjxl"
] | Both `libbrotlidec` and `libbrotlienc` already require `libbrotlicommon` for static builds in their respective .pc files and pkg-config will insert it automatically, so `libbrotlicommon` should be removed here to avoid overlinking or link order problems:
https://github.com/libjxl/libjxl/blob/0eebc1746afd51b6ea092007395765afdbfa0633/lib/jxl.cmake#L291
| Overlinking libbrotlicommon in pkg-config file | https://api.github.com/repos/libjxl/libjxl/issues/2206/comments | 0 | 2023-02-20T15:25:41Z | 2023-02-24T07:14:13Z | https://github.com/libjxl/libjxl/issues/2206 | 1,592,063,592 | 2,206 |
[
"libjxl",
"libjxl"
] | **Is your feature request related to a problem? Please describe.**
There are no readily-available Wasm builds of libjxl available except for [the really old one in Squoosh](https://github.com/GoogleChromeLabs/squoosh/tree/dev/codecs/jxl), last updated over a year ago!
**Describe the solution you'd like**
Wasm builds are already [built and tested](https://github.com/libjxl/libjxl/blob/main/.github/workflows/build_test.yml#L348) through GitHub Actions, but are not added to the releases, so I'd like for them to be added.
**Describe alternatives you've considered**
I'm currently using the aformentioned version in Squoosh but it's extremely out of date, so much worse performance and probably also security concerns with that.
| Add WebAssembly builds to releases section | https://api.github.com/repos/libjxl/libjxl/issues/2200/comments | 2 | 2023-02-20T09:36:27Z | 2023-05-10T11:45:04Z | https://github.com/libjxl/libjxl/issues/2200 | 1,591,484,964 | 2,200 |
[
"libjxl",
"libjxl"
] | On my Linux system, the `gdk-pixbuf-2.0.pc` file contains the following:
```
prefix=/usr
libdir=${prefix}/x86_64-pc-linux-gnu/lib
gdk_pixbuf_binarydir=${libdir}/gdk-pixbuf-2.0/2.10.0
gdk_pixbuf_moduledir=${gdk_pixbuf_binarydir}/loaders
```
so the expected final value for `gdk_pixbuf_moduledir` is `/usr/x86_64-pc-linux-gnu/lib/gdk-pixbuf-2.0/2.10.0/loaders`. However, when compiling libjxl using `cmake -DCMAKE_INSTALL_PREFIX:PATH=/usr/x86_64-pc-linux-gnu`, the plugin is instead installed to `/usr/x86_64-pc-linux-gnu/x86_64-pc-linux-gnu/lib/gdk-pixbuf-2.0/2.10.0/loaders` and is not found by gdk-pixbuf.
The reason for this is that when the libjxl build looks up the `gdk_pixbuf_moduledir` variable, it substitutes `CMAKE_INSTALL_PREFIX` (`/usr/x86_64-pc-linux-gnu`) in place of the prefix stored in the pkg-config file (`/usr`) using the `--define-variable=` pkg-config option:
https://github.com/libjxl/libjxl/blob/eb8a564d66a6eccb2d07c3942de524194f779bd5/plugins/gdk-pixbuf/CMakeLists.txt#L27-L30
The two variables come from different build systems and don't have the same meaning, so it doesn't make sense to replace one with the other. And even if they did mean the same thing, substituting in a different prefix would result in the gdk-pixbuf plugin being installed to a different directory from where gdk-pixbuf is looking for image loaders - so the plugin won't work.
I think the variable substitution should be dropped, so that the gdk-pixbuf plugin gets installed to the path as specified in the pkg-config file, even if it would be in a different prefix from libjxl.
This also would allow simplifying the cmake file - the complex `execute_process` call can be replaced with the much simpler `pkg_get_variable` call. | gdk-pixbuf plugin installs to incorrect prefix | https://api.github.com/repos/libjxl/libjxl/issues/2199/comments | 5 | 2023-02-19T16:00:26Z | 2023-02-20T14:49:51Z | https://github.com/libjxl/libjxl/issues/2199 | 1,590,742,330 | 2,199 |
[
"libjxl",
"libjxl"
] | I just compiled a recent version of cjxl (JPEG XL encoder v0.9.0 ed027ed5 [AVX3], this commit 5e222241cb6a2630fc2f4356b21b448088125bde), and the recently added encode effort level 10 isn't available. Has it been removed? It's still [referenced in the docs](https://github.com/libjxl/libjxl/blob/main/doc/encode_effort.md) as a setting in the encoder. | Encode effort 10 missing from encoder | https://api.github.com/repos/libjxl/libjxl/issues/2196/comments | 1 | 2023-02-19T03:55:40Z | 2023-02-19T14:07:13Z | https://github.com/libjxl/libjxl/issues/2196 | 1,590,549,617 | 2,196 |
[
"libjxl",
"libjxl"
] | Hi, There is any chance to install it on High Sierra? (homebrew or anything) | Install on High Sierra | https://api.github.com/repos/libjxl/libjxl/issues/2195/comments | 1 | 2023-02-18T19:29:06Z | 2023-04-17T05:27:32Z | https://github.com/libjxl/libjxl/issues/2195 | 1,590,447,086 | 2,195 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
I have converted a PNG into a lossless JXL. When I try to convert this JXL into a lossy one I get an error:
`Getting pixel data failed.`
**To Reproduce**
1. save the attached image and name it "logs.png"
2. run `cjxl logs.png logs_0.0.jxl -d 0.0` - you get lossless JXL (it is viewable via XnView on my system)
3. run `cjxl logs_0.0.jxl logs_1.0.jxl -d 1.0` - you get error: **Getting pixel data failed.**
**Expected behavior**
I expected no read error. :-)
**Screenshots**
If applicable, add screenshots or example input/output images to help explain your problem.
**Environment**
- OS: Windows 10 Pro 22H2 x64
- CPU type: AMD Ryzen 5 PRO 3500U
- cjxl/djxl version string: JPEG XL encoder v0.8.1 c27d499 [AVX2,SSE4,SSSE3,Unknown]
<!--
Currently github does not allow uploading files that end in `.jxl`, but when you
rename them for example as `image.jxl.jpg`, it will be possible to upload them
and also view them in browsers that are configured to support it.
See https://github.com/orgs/github-community/discussions/18139
-->
| cjxl can't read JXL it has just generated | https://api.github.com/repos/libjxl/libjxl/issues/2193/comments | 10 | 2023-02-17T16:38:24Z | 2025-01-24T22:56:51Z | https://github.com/libjxl/libjxl/issues/2193 | 1,589,644,247 | 2,193 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
Brotli libraries(.dll and .lib) are not copied to the installation directory specified by -DCMAKE_INSTALL_PREFIX.
They exist in the build directory, so they are simply not copied.
**To Reproduce**
Execute cmake.
**Expected behavior**
Brotli libraries must be copied into the installation directory.
**Screenshots**
None.
**Environment**
- OS: Windows 10 Pro x64 22H2 build 19045.2546
- Compiler version: Microsoft(R) C/C++ Optimizing Compiler Version 19.34.31937 for x64
- CPU type: Intel x86_64
- cjxl/djxl version string: JPEG XL encoder v0.9.0 6280645e [AVX2,SSE4,SSSE3,Unknown]
**Additional context**
None.
<!--
Currently github does not allow uploading files that end in `.jxl`, but when you
rename them for example as `image.jxl.jpg`, it will be possible to upload them
and also view them in browsers that are configured to support it.
See https://github.com/orgs/github-community/discussions/18139
-->
| Brotli libraries are not copied to the installation directory | https://api.github.com/repos/libjxl/libjxl/issues/2185/comments | 4 | 2023-02-12T22:16:28Z | 2023-03-02T09:19:04Z | https://github.com/libjxl/libjxl/issues/2185 | 1,581,429,591 | 2,185 |
[
"libjxl",
"libjxl"
] | If the build is done on a Linux distribution where the DATADIR is in a location other than `${CMAKE_INSTALL_PREFIX}/share`, then the data files for plugins - the mime `packages/image-jxl.xml` file, and thumbnailers `jxl.thumbnailer` file in particular - will be installed in the wrong location.
An example is in [Exherbo](https://www.exherbo.org/), where our standard cmake path setup uses `-DCMAKE_INSTALL_PREFIX:PATH=/usr/x86_64-pc-linux-gnu -DCMAKE_INSTALL_DATAROOTDIR:PATH=/usr/share/`. The data files get installed to `/usr/x86_64-pc-linux-gnu/share/` instead of the desired `/usr/share/`.
The fix is that these files should be installed using the [GNUInstallDirs](https://cmake.org/cmake/help/latest/module/GNUInstallDirs.html#module:GNUInstallDirs) `${CMAKE_INSTALL_DATADIR}` variable instead of a hardcoded `share` path. | Installation of plugin mime and thumbnailer files doesn't respect DATAROOTDIR | https://api.github.com/repos/libjxl/libjxl/issues/2183/comments | 3 | 2023-02-11T20:59:58Z | 2024-02-06T11:44:44Z | https://github.com/libjxl/libjxl/issues/2183 | 1,581,013,881 | 2,183 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
Can't decode [2021-08_totocaca.jxl](https://jpegxl.info/art/2021-08_totocaca.jxl) from https://jpegxl.info/art
**To Reproduce**
libjxl version:
b28f9ffa62a7d469915ab8c37c61ad4e93343812
submodule versions:
1405a214a8232991c585dc2164b5dec5bd559a33 testdata (1405a21)
36533a866ed1ca4b75cf049f4521e4ec5fe24727 third_party/brotli (36533a8)
58d77fa8070e8cec2dc1ed015d66b454c8d78850 third_party/googletest (release-1.12.1)
58746ca5b9f9444a2a3549704602ecc6239f8f41 third_party/highway (1.0.3)
65c63bf549d78253c14b30b3d62cb668bbbe612c third_party/lcms (65c63bf)
a40189cf881e9f0db80511c382292a5604c3c3d1 third_party/libpng (v1.6.37)
868ab558fad70fcbe8863ba4e85179eeb81cc840 third_party/sjpeg (868ab55)
b25b07b4b07990811de121c0356155b2ba0f4318 third_party/skcms (b25b07b)
cacf7f1d4e3d44d871b605da3b647f07d718623f third_party/zlib (v1.2.11)
Compile a checkout of those commit hashes with this command line:
```console
$ cmake -DCMAKE_BUILD_TYPE=Release -DBUILD_TESTING=OFF -DJPEGXL_ENABLE_JPEGLI=OFF -DJPEGXL_ENABLE_JNI=OFF -DJPEGXL_ENABLE_TCMALLOC=OFF -B build -G Ninja && cmake --build build
```
Then try decode [2021-08_totocaca.jxl](https://jpegxl.info/art/2021-08_totocaca.jxl) from the jxl art web page:
```console
$ djxl 2021-08_totocaca.jxl test.jpg
```
You will get error output like this:
```plain
JPEG XL decoder v0.9.0 b28f9ffa [AVX2,SSE4,SSSE3,Unknown]
Read 81 compressed bytes.
Warning: could not decode losslessly to JPEG. Retrying with --pixels_to_jpeg...
./lib/extras/enc/encode.cc:135: encoding 10-bit original to 8 bits
./lib/jxl/splines.cc:645: JXL_FAILURE: Too many pixels covered with splines
./lib/jxl/dec_frame.cc:290: JXL_RETURN_IF_ERROR code=1: shared.image_features.splines.InitializeDrawCache( frame_dim_.xsize_upsampled, frame_dim_.ysize_upsampled, dec_state_->shared->cmap)
./lib/jxl/decode.cc:1149: frame processing failed
Failed to decode image
DecompressJxlToPackedPixelFile failed
```
The instructions try to perfectly replicate my setup, but it shouldn't matter, the errors are explicit checks in the code failing.
**Screenshots**
**Environment**
- OS: OpenBSD 7.2 amd64
- Compiler version: Clang 13.0.0 from the OpenBSD base system
- CPU type: Intel Core i7 4770, i.e. Haswell
- cjxl/djxl version string: JPEG XL decoder v0.9.0 b28f9ffa [AVX2,SSE4,SSSE3,Unknown] | 2021-08_totocaca.jxl: Failed to decode image | https://api.github.com/repos/libjxl/libjxl/issues/2171/comments | 2 | 2023-02-09T12:35:31Z | 2023-04-25T06:20:27Z | https://github.com/libjxl/libjxl/issues/2171 | 1,577,817,780 | 2,171 |
[
"libjxl",
"libjxl"
] | jpegli doesn't compile on OpenBSD because it includes `<jpeglib.h>` which is not in the default include path. In OpenBSD, the default include path is _/usr/include_, but package headers are installed to _/usr/local/include_.
The other BSDs have the same detail, NetBSD in particular uses _/usr/pkg/include_, so it's very likely it doesn't work on the other BSDs too, but I haven't tried it out.
The solution would be to use pkg-config to get the CFLAGS but not the LDFLAGS needed to use libjpeg. This might have to be propagated to libjxl's own pkg-config files.
| jpegli doesn't compile on OpenBSD | https://api.github.com/repos/libjxl/libjxl/issues/2167/comments | 2 | 2023-02-09T09:24:21Z | 2023-02-18T09:36:24Z | https://github.com/libjxl/libjxl/issues/2167 | 1,577,532,266 | 2,167 |
[
"libjxl",
"libjxl"
] | Hello,
I am experimenting with including `libjxl` into Android application. I am building `libjx` on Windows in Android Studio and the app builds via Gradle.
I observed that **by default**, all cmake targets are built there (unless I manually specify the exact target I want).
I was getting following errors:
`C:\Devel\Camera\PhotonCamera\app\src\main\cpp\deps\libjxl\tools\fuzzer_stub.cc:11:10: fatal error: 'jxl/thread_parallel_runner.h' file not found`
`C:\Devel\Camera\PhotonCamera\app\src\main\cpp\deps\libjxl\tools\cjxl_fuzzer.cc:18:10: fatal error: 'jxl/encode.h' file not found`
The problem is, that majority of fuzzer runners fail to build when **JPEGXL_ENABLE_TOOLS** is not enabled, because they need **jxl_extras-static** target, which depends on that option:
```
if(JPEGXL_ENABLE_TOOLS)
include(jxl_extras.cmake)
endif()
```
I am able to reproduce the build problem under Linux too, when attempting to build `_fuzzer_runner` target explicitly:
```
make fields_fuzzer_runner
[ 0%] Built target tool_version_git
[ 40%] Built target hwy
[ 80%] Built target jxl_tool
[ 80%] Building CXX object tools/CMakeFiles/fields_fuzzer_runner.dir/fuzzer_stub.cc.o
[ 80%] Building CXX object tools/CMakeFiles/fields_fuzzer_runner.dir/fields_fuzzer.cc.o
[100%] Linking CXX executable fields_fuzzer_runner
/usr/lib/gcc/x86_64-pc-linux-gnu/12/../../../../x86_64-pc-linux-gnu/bin/ld: cannot find -ljxl_extras-static: Adresár alebo súbor neexistuje
collect2: error: ld returned 1 exit status
make[3]: *** [tools/CMakeFiles/fields_fuzzer_runner.dir/build.make:124: tools/fields_fuzzer_runner] Error 1
make[2]: *** [CMakeFiles/Makefile2:873: tools/CMakeFiles/fields_fuzzer_runner.dir/all] Error 2
make[1]: *** [CMakeFiles/Makefile2:880: tools/CMakeFiles/fields_fuzzer_runner.dir/rule] Error 2
make: *** [Makefile:478: fields_fuzzer_runner] Error 2
```
```
make color_encoding_fuzzer_runner
[ 0%] Built target tool_version_git
[ 40%] Built target hwy
[ 80%] Built target jxl_tool
[ 80%] Linking CXX executable color_encoding_fuzzer_runner
/usr/lib/gcc/x86_64-pc-linux-gnu/12/../../../../x86_64-pc-linux-gnu/bin/ld: cannot find -ljxl_extras-static: Adresár alebo súbor neexistuje
collect2: error: ld returned 1 exit status
make[3]: *** [tools/CMakeFiles/color_encoding_fuzzer_runner.dir/build.make:124: tools/color_encoding_fuzzer_runner] Error 1
make[2]: *** [CMakeFiles/Makefile2:735: tools/CMakeFiles/color_encoding_fuzzer_runner.dir/all] Error 2
make[1]: *** [CMakeFiles/Makefile2:742: tools/CMakeFiles/color_encoding_fuzzer_runner.dir/rule] Error 2
make: *** [Makefile:413: color_encoding_fuzzer_runner] Error 2
```
Only `djxl_fuzzer_runner` builds because it used different target_link_libraries.
I suggest that instead of https://github.com/libjxl/libjxl/blob/9b367dcf5f01d72b7d0159597613e789e10036e3/tools/CMakeLists.txt#L310
to use something like:
```
set(FUZZER_BINARIES djxl_fuzzer)
if(JPEGXL_ENABLE_TOOLS)
list(APPEND FUZZER_BINARIES
color_encoding_fuzzer
decode_basic_info_fuzzer
cjxl_fuzzer
icc_codec_fuzzer
fields_fuzzer
rans_fuzzer
set_from_bytes_fuzzer
transforms_fuzzer
)
endif()
``` | Fuzzers fail to build without jxl_extras-static | https://api.github.com/repos/libjxl/libjxl/issues/2160/comments | 1 | 2023-02-08T11:19:43Z | 2023-02-10T10:33:36Z | https://github.com/libjxl/libjxl/issues/2160 | 1,575,938,476 | 2,160 |
[
"libjxl",
"libjxl"
] | **Is your feature request related to a problem? Please describe.**
Especially in the case of images that are not photos, but rather textures that will be post-processed (eg. warped), there might be regions of interest, or more generally we can consider a map of lossy-ness we can afford, so that they are less messed-up than others in the compression process. With a uniform quality setting, one can use a workaround consisting in pre-warping an image prior to performing lossy compression, but this is kind of ugly.
**Describe the solution you'd like**
The ability of specifying a spatially varying max. distance (or something more general in the case of "multi-modal" images) would be pretty cool. In case the image is uint8 we might want to consider a floating point multiplier for the values...
If the distance value is zero, then the corresponding areas shall be losslessly compressed.
**Describe alternatives you've considered**
I have not thought about this a lot, and maybe something like this is already possible or documented, but I didn't find it after a cursory search.
| Encoding: specifying spatially variable compression quality using a non-uniform max. distance or similar | https://api.github.com/repos/libjxl/libjxl/issues/2157/comments | 0 | 2023-02-07T13:42:19Z | 2023-02-08T13:48:05Z | https://github.com/libjxl/libjxl/issues/2157 | 1,574,358,167 | 2,157 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
I was doing random lossless compression benchmarks of libjxl using cjxl, on a collection of screenshots, measuring various things, in particular compressed file size, and noticed that quite frequently, an increase of compression effort doesn't result in an increased (or equal) compression ratio.
**Expected behavior**
When the effort increases, the compression ratio shall only improve, or stay the same, but not get reduced.
libjxl documentation does say, though:
> For lossless compression, higher effort should result in smaller filesizes, although this is not guaranteed
**To Reproduce**
Here's an example file:
Giving:
```sh
for effort in {1..9}; do
cjxl 217253324-c27754ef-6a86-457a-ac1e-e40c0400f5f4.png --effort=${effort} --disable_output 2>&1 | grep "to ";
done
```
```
Compressed to 23381 bytes (2.205 bpp).
Compressed to 23381 bytes (2.205 bpp).
Compressed to 22058 bytes (2.080 bpp).
Compressed to 22037 bytes (2.078 bpp).
Compressed to 18346 bytes (1.730 bpp).
Compressed to 18329 bytes (1.729 bpp).
Compressed to 9626 bytes (0.908 bpp).
Compressed to 18417 bytes (1.737 bpp).
Compressed to 18485 bytes (1.743 bpp).
```
**Environment**
- OS: Linux
- Compiler version: gcc-12
- CPU type: x86_64
- cjxl/djxl version string: JPEG XL encoder v0.9.0 195c36a0 [AVX2]
| When compression effort is increased, file size is not always reduced or equal | https://api.github.com/repos/libjxl/libjxl/issues/2156/comments | 3 | 2023-02-07T13:14:48Z | 2023-02-15T12:42:50Z | https://github.com/libjxl/libjxl/issues/2156 | 1,574,312,967 | 2,156 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
According to the built in help in cjxl (`-v -h`), the `--photon_noise` parameter works as follows:
> --photon_noise=ISO3200
Adds noise to the image emulating photographic film noise. The higher the given number, the grainier the image will be. As an example, a value of 100 gives low noise whereas a value of 3200 gives a lot of noise. The default value is 0.
However, if this is set to the 'default' of `0`, the following error appears:
> Error parsing flag --photon_noise=ISO0
**To Reproduce**
Run `cjxl` like so: `cjxl foo.png foo.jxl --photon_noise=ISO0`
**Expected behavior**
Ideally, the image would still be encoded, just with no photon noise at all, even though 'ISO0' doesn't make a whole lot of sense naming-wise.
**Environment**
- OS: Linux
- Compiler version: `clang version 14.0.6`
- CPU type: x86_64
- cjxl/djxl version string: `cjxl v0.8.1 [AVX2,SSE4,SSSE3,Unknown]` | 'photon_noise' cannot be set to 'ISO0' despite it being the default. | https://api.github.com/repos/libjxl/libjxl/issues/2149/comments | 0 | 2023-02-04T07:56:31Z | 2023-02-09T10:45:44Z | https://github.com/libjxl/libjxl/issues/2149 | 1,570,823,072 | 2,149 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
As stated in the [docs](https://libjxl.readthedocs.io/en/latest/api_common.html#_CPPv4N12JxlBasicInfo21uses_original_profileE):
`If the original profile is not used, the decoder only outputs the data as sRGB (linear if outputting to floating point, nonlinear with standard sRGB transfer function if outputting to unsigned integers) but will not convert it to to the original color profile.`
The decoder should returns with sRGB TRC when pixel format is uint, but this is not the case and instead it outputs as a linear TRC instead: `RGB_D65_SRG_rel_Lin`,
**To Reproduce**
Steps to reproduce the behavior:
Decode an uint JXL image that was converted from an color profile / ICC that's not currently supported by libjxl's enum. For example this image:
https://kampidh.com/jxlcoloranalysis/CC-proph-boost-xyb-icc-integer.jxl
**Expected behavior**
According to the docs, it should return as sRGB TRC, not linear.
**Environment**
- OS: Windows 10
- Compiler version: Clang 15
- CPU type: x86_64
- cjxl/djxl version string: 0.8.0
**Additional context**
This is possibly related to another reports that shows the image results were darker, since on non-color managed viewer / browser, linear TRC always looks dark.
<!--
Currently github does not allow uploading files that end in `.jxl`, but when you
rename them for example as `image.jxl.jpg`, it will be possible to upload them
and also view them in browsers that are configured to support it.
See https://github.com/orgs/github-community/discussions/18139
-->
| Decoding from XYB with ICC always returns linear sRGB regardless of pixel format | https://api.github.com/repos/libjxl/libjxl/issues/2147/comments | 0 | 2023-02-03T22:01:39Z | 2024-04-22T12:31:37Z | https://github.com/libjxl/libjxl/issues/2147 | 1,570,534,196 | 2,147 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
Installation following the https://github.com/libjxl/libjxl/blob/main/README.md does not seem to be compiling/installing JPEG XL GDK Pixbuf (libpixbufloader-jxl.so) neither jxl.thumbnailer.
The libpixbufloader-jxl.so file is also not present in the build folder to be manually copied as described at https://github.com/libjxl/libjxl/tree/main/plugins/gdk-pixbuf#jpeg-xl-gdk-pixbuf.
The jxl.thumbnailer is already present when you clone the project, so is possible to install it manually.
**To Reproduce**
Just follow the installation steps at https://github.com/libjxl/libjxl/blob/main/README.md#compilation
**Expected behavior**
Have the files installed automatically as cited at https://github.com/libjxl/libjxl/tree/main/plugins/gdk-pixbuf#jpeg-xl-gdk-pixbuf or have the same available to be copied manually.
**Environment**
- OS: Pop!_OS 22.04 LTS
- Compiler version: clang 1:14.0-55~exp2
- CPU type: x86_64
- cjxl/djxl version string: JPEG XL encoder v0.9.0 4112b8a9 [AVX2,SSE4,SSSE3,Unknown] | JPEG XL decoder v0.9.0 4112b8a9 [AVX2,SSE4,SSSE3,Unknown]
**Additional context**
Personal context, I'm not a new Linux user, but I have very little experience compiling projects. | Can't seem to find JPEG XL GDK Pixbuf | https://api.github.com/repos/libjxl/libjxl/issues/2139/comments | 5 | 2023-02-03T01:53:15Z | 2023-02-04T12:33:00Z | https://github.com/libjxl/libjxl/issues/2139 | 1,569,071,565 | 2,139 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
`cjxl` fails to decode a CMYK JPG to pixels (`-j 0`, i.e. JPEG lossless transcode is turned off) and encode it to JXL.
**To Reproduce**
1. Download the [CMYK JPG from the testdata repo](https://raw.githubusercontent.com/libjxl/testdata/main/jxl/flower/flower_small.cmyk.jpg).
2. `./cjxl -j 0 flower_small.cmyk.jpg cmyk.jxl`
**Actual behavior**
```
JPEG XL encoder v0.9.0 9a7feac1 [AVX2,SSE4,SSSE3,Unknown]
Getting pixel data failed.
```
**Expected behavior**
A CMYK JXL is produced.
**Environment**
- OS: Linux
- CPU type: x86_64
- [Binary used](https://artifacts.lucaversari.it/libjxl/libjxl/latest/jxl-linux-x86_64-static.zip) | cjxl fails to encode CMYK image | https://api.github.com/repos/libjxl/libjxl/issues/2126/comments | 1 | 2023-02-01T05:33:53Z | 2024-04-22T09:49:00Z | https://github.com/libjxl/libjxl/issues/2126 | 1,565,464,684 | 2,126 |
[
"libjxl",
"libjxl"
] | cjxl in fast lossless mode produces invalid JXLs from the attached PNGs.
**To Reproduce**
Encode the attached PNGs with `cjxl -e 1 -d 0` and attempt to decode.
[failure.png](https://user-images.githubusercontent.com/16582685/215329162-c94173e3-8f5e-4050-a95e-726a16ba997f.png)
```
$ cjxl failure.png failure.jxl -d 0 -e 1 && djxl failure.jxl --disable_output
JPEG XL encoder v0.9.0 20ca234f [AVX2,SSE4,SSSE3,Unknown]
./lib/extras/dec/color_hints.cc:54: No color_space/icc_pathname given, assuming sRGB
Read 2x2 image, 84 bytes, 0.3 MP/s
Encoding [Modular, lossless, effort: 1],
Compressed to 183 bytes (366.000 bpp).
2 x 2, 0.00 MP/s [0.00, 0.00], 1 reps, 8 threads.
JPEG XL decoder v0.9.0 20ca234f [AVX2,SSE4,SSSE3,Unknown]
Read 183 compressed bytes.
./lib/jxl/modular/encoding/encoding.cc:566: JXL_FAILURE: Truncated input
./lib/jxl/modular/encoding/encoding.cc:597: JXL_RETURN_IF_ERROR code=1: dec_status
./lib/jxl/dec_modular.cc:257: JXL_RETURN_IF_ERROR code=1: dec_status
./lib/jxl/decode.cc:1139: frame out of bounds
Failed to decode image
DecompressJxlToPackedPixelFile failed
```
Decoder backtrace to the first failure:
```
#0 jxl::ModularDecode (br=0x5555555f4760, image=..., header=..., group_id=0, options=0x7fffffffd1e0, global_tree=0x5555555fb748, global_code=0x5555555fb760, global_ctx_map=0x5555555fb7f0, allow_truncated_group=false)
at ./lib/jxl/modular/encoding/encoding.cc:566
#1 0x00007ffff7b5b9bd in jxl::ModularGenericDecompress (br=0x5555555f4760, image=..., header=0x5555555fb808, group_id=0, options=0x7fffffffd1e0, undo_transforms=false, tree=0x5555555fb748, code=0x5555555fb760, ctx_map=0x5555555fb7f0, allow_truncated_group=false)
at ./lib/jxl/modular/encoding/encoding.cc:595
#2 0x00007ffff7adba31 in jxl::ModularFrameDecoder::DecodeGlobalInfo (this=0x5555555fb640, reader=0x5555555f4760, frame_header=..., allow_truncated_group=false) at ./lib/jxl/dec_modular.cc:252
#3 0x00007ffff7a4c3d0 in jxl::FrameDecoder::ProcessDCGlobal (this=0x5555555fb330, br=0x5555555f4760) at ./lib/jxl/dec_frame.cc:292
#4 0x00007ffff7a4f14e in jxl::FrameDecoder::ProcessSections (this=0x5555555fb330, sections=0x5555555fbf70, num=1, section_status=0x5555555fbf90) at ./lib/jxl/dec_frame.cc:619
#5 0x00007ffff7afa4e1 in jxl::(anonymous namespace)::JxlDecoderProcessSections (dec=0x5555555f4bd0) at ./lib/jxl/decode.cc:1123
#6 0x00007ffff7afbc1d in jxl::(anonymous namespace)::JxlDecoderProcessCodestream (dec=0x5555555f4bd0) at ./lib/jxl/decode.cc:1438
#7 0x00007ffff7afd682 in HandleBoxes (dec=0x5555555f4bd0) at ./lib/jxl/decode.cc:1898
#8 0x00007ffff7afdd42 in JxlDecoderProcessInput (dec=0x5555555f4bd0) at ./lib/jxl/decode.cc:2049
#9 0x0000555555570546 in jxl::extras::DecodeImageJXL (bytes=0x5555555f4af0 "\377\nA\300J\b\b\020", bytes_size=201, dparams=..., decoded_bytes=0x7fffffffdd40, ppf=0x7fffffffe230, jpeg_bytes=0x0) at ./lib/extras/dec/jxl.cc:205
#10 0x000055555555e7f3 in (anonymous namespace)::DecompressJxlToPackedPixelFile (args=..., compressed=std::vector of length 201, capacity 201 = {...}, accepted_formats=std::vector of length 0, capacity 0, runner=0x5555555f3ae0, ppf=0x7fffffffe230, decoded_bytes=0x7fffffffdd40,
stats=0x7fffffffde30) at ./tools/djxl_main.cc:283
#11 0x000055555555f378 in main (argc=3, argv=0x7fffffffe568) at ./tools/djxl_main.cc:437
```
The second PNG attached here is a crop of the first one - it appears to work correctly, but decodes to the wrong pixels.
[wrongpixels.png](https://user-images.githubusercontent.com/16582685/215329196-362ccb34-8553-45ac-92f5-4683eb492a31.png)
```
$ cjxl wrongpixels.png wrongpixels.jxl -d 0 -e 1
JPEG XL encoder v0.9.0 20ca234f [AVX2,SSE4,SSSE3,Unknown]
./lib/extras/dec/color_hints.cc:54: No color_space/icc_pathname given, assuming sRGB
Read 1x2 image, 75 bytes, 0.1 MP/s
Encoding [Modular, lossless, effort: 1],
Compressed to 182 bytes (728.000 bpp).
1 x 2, 0.00 MP/s [0.00, 0.00], 1 reps, 8 threads.
$ magick identify -format '%#\n' wrongpixels.png wrongpixels.jxl
3674e32a54f125bc23b069ab31af8043f8f17a02ae27f6f22ff19ffdb0961318
c31c8fa31580112a8f7f8aaca994ec1ef63511a32600dca28e2ec02aebfe6a36
$ magick identify -verbose wrongpixels.png wrongpixels.jxl | grep Colors:
Colors: 2
Colors: 1
```
In this case the result is fully transparent - a slightly wider crop around the same pixels resulted in some strange artifacts, as if a section of the image had been rotated 90°.
I haven't been able to reproduce it with inputs larger than a few pixels.
**Expected behavior**
Successful encoding and decoding.
**Environment**
- OS: Linux
- Compiler version: clang 14.0.0
- CPU type: x86_64
- cjxl/djxl version string: v0.9.0 20ca234f [AVX2,SSE4,SSSE3,Unknown]
**Additional context**
The symptoms are similar to #2039 but the example PNG from that bug doesn't cause a problem now.
| Fast lossless can produce invalid/incorrect JXLs | https://api.github.com/repos/libjxl/libjxl/issues/2118/comments | 2 | 2023-01-29T13:31:04Z | 2023-01-31T11:19:32Z | https://github.com/libjxl/libjxl/issues/2118 | 1,561,290,074 | 2,118 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
Commit https://github.com/libjxl/libjxl/commit/c3779422fd5bbae2c27e72761af6a5a51f64fa81 from https://github.com/libjxl/libjxl/pull/2110 breaks compilation on Ubuntu 20.04 with
```
CMake Error at lib/jxl.cmake:426 (target_include_directories):
Error evaluating generator expression:
$<TARGET_PROPERTY:brotlicommon-static,INTERFACE_INCLUDE_DIRECTORIES>
Target "brotlicommon-static" not found.
```
**To Reproduce**
```
docker run --rm -it ubuntu:20.04
# inside Docker
apt update
apt install git g++ cmake
git clone https://github.com/libjxl/libjxl.git --recursive --shallow-submodules
cd libjxl
mkdir build
cd build
cmake ..
```
CC @eustas | third_party/brotli upgrade breaks compilation on Ubuntu 20.04 | https://api.github.com/repos/libjxl/libjxl/issues/2112/comments | 2 | 2023-01-26T19:13:52Z | 2023-01-27T19:59:34Z | https://github.com/libjxl/libjxl/issues/2112 | 1,558,638,448 | 2,112 |
[
"libjxl",
"libjxl"
] | When encoding lossy JXL, distance/quality of alpha channel cannot be adjusted separately and depends on the overall distance/quality setting. This may result in inefficient compression of alpha channel and larger file size.
In the example bellow, there are three JXL files. All of them were cretaed using the recent version of cjxl.
1. The lone alpha channel encoded as grayscale image, arguments: "-d 0 -e 9". The file size is **1091 bytes**.
2. The RGB image, arguments: "-e 9". The file size is **14861 bytes**.
3. The RGB+Alpha image combinng the previous images, arguments: "-e 9". The file size is **19594 bytes**.
Apparetly the image 3 is notieably larger than images 1 and 2 added together. The reason is the lossless encoding, in this case, compresses the alpha channel much more efficiently than lossy encoding.
**My suggested solution is to allow specifying independent distance/quality for alpha channel.** Perhaps the default could be set to lossless alpha. Lossless compression is very efficient for alpha consisting mostly of fully transparent and fully opaque areas.
| Inefficient alpha compression in lossy JXL | https://api.github.com/repos/libjxl/libjxl/issues/2111/comments | 5 | 2023-01-26T18:19:24Z | 2023-05-10T10:51:06Z | https://github.com/libjxl/libjxl/issues/2111 | 1,558,570,284 | 2,111 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
I have a jpg that was saved from Photoshop. I converted it to jxl using `cjxl image.jpg image.jxl`
When viewing the two images in ImageGlass (8.7.11.6), there is a distinct difference in the color output in the jxl vs the jpg.
The good news is that running the opposite `djxl image.jxl image2.jpg` produces an output jpg that is identical to the original.
From my tests, this only seems to happen with images saved from photoshop. I assume it has something to do with color profiles, but I'm not quite well versed enough in that to understand the difference.
**To Reproduce**
Take jxltest.jpg and run `cjxl jxltest.jpg jxltest.jxl`
compare visually in ImageGlass
**Expected behavior**
I expected the jxl to appear visually identical to the original jpeg.
**Screenshots**
original (crop):
jxl:
viewing both in ImageGlass (I'm not sure why my screencap shows less saturated for both, on my screen the left is identical to the original. however, you can still see the saturation difference)
**Environment**
- OS: **Ubuntu 22.04.1 [wsl under Windows 10]**
- Compiler version: **clang 14.0.0-1ubunut1**
- CPU type: **x86_64**
- cjxl/djxl version string: **cjxl v0.9.0 aa99c8d8 [AVX2,SSE4,SSSE3,Unknown]**
| image saturation difference between jpg and lossless jxl | https://api.github.com/repos/libjxl/libjxl/issues/2105/comments | 2 | 2023-01-25T19:12:23Z | 2023-01-26T17:45:04Z | https://github.com/libjxl/libjxl/issues/2105 | 1,557,145,549 | 2,105 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
For verification, used the synthetic examples, which, however, is in appearance a regular files and can be opened through image viewer. After run, binary file aborted with Segfault.
**To Reproduce**
Steps to reproduce the behavior:
```bash
/run/media/user/b05c56b9-304e-42f6-af2b-883e9de3e6de/target/libjxl/sub/libjxl/tools/cjxl sig_11_src_000122_poc --disable_output
```
Output:
```bash
JPEG XL encoder v0.9.0 aa99c8d8 [AVX2,SSE4,SSSE3,Unknown]
Note: Implicit-default for JPEG is lossless-transcoding. To silence this message, set --lossless_jpeg=(1|0).
Read JPEG image with 94120 bytes.
Encoding [Container | JPEG, lossless transcode, effort: 7 | JPEG reconstruction data],
Segmentation fault (core dumped)
```
```bash
Program terminated with signal SIGSEGV, Segmentation fault.
#0 jxl::FindExifTagPosition (exif=..., tagname=274) at /run/media/user/b05c56b9-304e-42f6-af2b-883e9de3e6de/target/libjxl/sub/libjxl_gdb/lib/jxl/exif.h:44
44 uint16_t nb_tags = (bigendian ? LoadBE16(t) : LoadLE16(t));
[Current thread is 1 (Thread 0x7fd2f485a240 (LWP 3344243))]
(gdb) backtrace
#0 jxl::FindExifTagPosition (exif=..., tagname=274)
at /run/media/user/b05c56b9-304e-42f6-af2b-883e9de3e6de/target/libjxl/sub/libjxl_gdb/lib/jxl/exif.h:44
#1 jxl::InterpretExif (exif=std::vector of length 94, capacity 94 = {...},
orientation=orientation@entry=0x7ffe000b4ee0)
at /run/media/user/b05c56b9-304e-42f6-af2b-883e9de3e6de/target/libjxl/sub/libjxl_gdb/lib/jxl/exif.h:68
#2 0x00007fd2f6401e6e in JxlEncoderAddJPEGFrame (frame_settings=0x55d68f8664f0, buffer=<optimized out>,
size=94120)
at /run/media/user/b05c56b9-304e-42f6-af2b-883e9de3e6de/target/libjxl/sub/libjxl_gdb/lib/jxl/encode.cc:1562
#3 0x000055d68f829c05 in jxl::extras::EncodeImageJXL (params=..., ppf=..., jpeg_bytes=0x7ffe000b59c0,
compressed=0x7ffe000b5980)
at /run/media/user/b05c56b9-304e-42f6-af2b-883e9de3e6de/target/libjxl/sub/libjxl_gdb/lib/extras/enc/jxl.cc:82
#4 0x000055d68f81999c in main (argc=<optimized out>, argv=<optimized out>)
at /run/media/user/b05c56b9-304e-42f6-af2b-883e9de3e6de/target/libjxl/sub/libjxl_gdb/tools/cjxl_main.cc:1041
```
**Expected behavior**
No assertion failed.
**Screenshots**
before patch:
affter patch:
**Environment**
- OS: ```tested at 5.15.78-2.el7.3.x86_64 (RED OS release MUROM 7.3.2 - Centos based system)``` ;
- Compiler version: ```clang version 12.0.1 (RED OS 12.0.1-1.el7)``` ;
- CPU type: ```x86_64``` ;
- cjxl version string: ```v0.9.0 aa99c8d8```
**Additional context**
For solution this problem added additional error checking to ensure that the t pointer is pointing to a valid location within the exif buffer before trying to read the value of nb_tags.
```cpp
if(t + 2 >= exif.data() + exif.size()) return 0;
```
(also added in a patch)
link to a sample (github-url):
[sig_11_src_000122_poc](https://github.com/a-shvedov/fuzzing_investigation/blob/main/res/libjxl/2022_01_25/sig_11_src_000122_poc)
<!--
Currently github does not allow uploading files that end in `.jxl`, but when you
rename them for example as `image.jxl.jpg`, it will be possible to upload them
and also view them in browsers that are configured to support it.
See https://github.com/orgs/github-community/discussions/18139
-->
| jxl::FindExifTagPosition - pointing to memory that is out of bounds (Segfault) | https://api.github.com/repos/libjxl/libjxl/issues/2100/comments | 0 | 2023-01-24T23:10:51Z | 2023-02-01T09:17:24Z | https://github.com/libjxl/libjxl/issues/2100 | 1,555,802,383 | 2,100 |
[
"libjxl",
"libjxl"
] | `benchmark_xl v0.9.0 00773081 [AVX2] Windows 11`
https://unsplash.com/photos/zHcI7_TA5Yg
[Resized source](https://user-images.githubusercontent.com/119183190/214198427-4fb712ca-b0ed-4b5c-a45c-eba6bbb72c1f.png)
```
Encoding kPixels Bytes BPP Max norm SSIMULACRA2 pnorm
--------------------------------------------------------------------------------------------------------
jpeg:enc-jpegli:q77:dec-jpegli:bd16 4270 506467 0.9487932 136.28828430 -19.66365224 28.69669831
jpeg:enc-jpegli:q78:dec-jpegli:bd16 4270 517770 0.9699678 136.17242432 -19.65461055 28.69878841
jpeg:enc-jpegli:q79:dec-jpegli:bd16 4270 539942 1.0115038 136.35876465 -19.51233277 28.69276517
jpeg:enc-jpegli:q80:dec-jpegli:bd16 4270 558322 1.0459361 136.31909180 -18.50234908 29.04873686
jpeg:enc-jpegli:q81:dec-jpegli:bd16 4270 584871 1.0956718 135.07215881 -13.64525888 27.86054494
jpeg:enc-jpegli:q82:dec-jpegli:bd16 4270 605818 1.1349130 135.13157654 -13.09396603 27.83956163
jpeg:enc-jpegli:q83:dec-jpegli:bd16 4270 626056 1.1728260 135.04438782 -14.04418433 27.88352163
jpeg:enc-jpegli:q84:dec-jpegli:bd16 4270 658955 1.2344576 135.62678528 -17.32125784 28.63664983
jpeg:enc-jpegli:q85:dec-jpegli:bd16 4270 676199 1.2667617 135.75309753 -17.21376382 28.63755474
jpeg:enc-jpegli:q86:dec-jpegli:bd16 4270 727943 1.3636967 135.72795105 -22.43448568 28.95907792
jpeg:enc-jpegli:q87:dec-jpegli:bd16 4270 750174 1.4053433 131.95985413 4.48420770 24.67779789
jpeg:enc-jpegli:q88:dec-jpegli:bd16 4270 797480 1.4939643 156.63969421 -5.87667081 27.41512382
jpeg:enc-jpegli:q89:dec-jpegli:bd16 4270 833215 1.5609087 156.64601135 -8.21494064 27.60544189
jpeg:enc-jpegli:q90:dec-jpegli:bd16 4270 858588 1.6084413 134.41885376 -0.62536848 25.45361695
jpeg:enc-jpegli:q91:dec-jpegli:bd16 4270 915529 1.7151121 158.59341431 -11.43010326 29.72990926
jpeg:enc-jpegli:q92:dec-jpegli:bd16 4270 990559 1.8556701 158.30972290 -14.02209638 29.97293925
jpeg:enc-jpegli:q93:dec-jpegli:bd16 4270 1028262 1.9263012 159.12585449 -4.53117767 28.32048932
Encoding kPixels Bytes BPP Max norm SSIMULACRA2 pnorm
---------------------------------------------------------------------------------------------------
jpeg:enc-jpegli:q77:dec-jpegli 4270 506467 0.9487932 136.29013062 -19.69533990 28.69861848
jpeg:enc-jpegli:q78:dec-jpegli 4270 517770 0.9699678 136.17773438 -19.67940353 28.70026041
jpeg:enc-jpegli:q79:dec-jpegli 4270 539942 1.0115038 136.36407471 -19.52479557 28.69409251
jpeg:enc-jpegli:q80:dec-jpegli 4270 558322 1.0459361 136.31912231 -18.50531617 29.04740624
jpeg:enc-jpegli:q81:dec-jpegli 4270 584871 1.0956718 135.07145691 -13.65910244 27.85997853
jpeg:enc-jpegli:q82:dec-jpegli 4270 605818 1.1349130 135.12487793 -13.10358663 27.84026828
jpeg:enc-jpegli:q83:dec-jpegli 4270 626056 1.1728260 135.03981018 -14.06722971 27.88398708
jpeg:enc-jpegli:q84:dec-jpegli 4270 658955 1.2344576 135.69905090 -17.34370390 28.64071421
jpeg:enc-jpegli:q85:dec-jpegli 4270 676199 1.2667617 135.75822449 -17.23259798 28.63913671
jpeg:enc-jpegli:q86:dec-jpegli 4270 727943 1.3636967 135.72116089 -22.44112236 28.95814069
jpeg:enc-jpegli:q87:dec-jpegli 4270 750174 1.4053433 131.96627808 4.44193312 24.67877867
jpeg:enc-jpegli:q88:dec-jpegli 4270 797480 1.4939643 156.65722656 -5.88161609 27.41764087
jpeg:enc-jpegli:q89:dec-jpegli 4270 833215 1.5609087 156.65017700 -8.19164424 27.60601398
jpeg:enc-jpegli:q90:dec-jpegli 4270 858588 1.6084413 134.41241455 -0.65812658 25.45297143
jpeg:enc-jpegli:q91:dec-jpegli 4270 915529 1.7151121 158.59745789 -11.44213921 29.73001177
jpeg:enc-jpegli:q92:dec-jpegli 4270 990559 1.8556701 158.30667114 -14.02800417 29.97375086
jpeg:enc-jpegli:q93:dec-jpegli 4270 1028262 1.9263012 159.13714600 -4.52964354 28.32278880
Encoding kPixels Bytes BPP Max norm SSIMULACRA2 pnorm
------------------------------------------------------------------------------------------------------
jpeg:enc-jpegli:q77:xyb:dec-jpegli 4270 440755 0.8256912 2.57653522 73.83731587 1.08323953
jpeg:enc-jpegli:q78:xyb:dec-jpegli 4270 448209 0.8396552 2.46053791 74.12077303 1.07571795
jpeg:enc-jpegli:q79:xyb:dec-jpegli 4270 469636 0.8797956 2.31674862 75.04041524 1.02801235
jpeg:enc-jpegli:q80:xyb:dec-jpegli 4270 481438 0.9019050 2.27511597 75.40451765 1.00002946
jpeg:enc-jpegli:q81:xyb:dec-jpegli 4270 511347 0.9579352 2.22766042 76.61113140 0.94573886
jpeg:enc-jpegli:q82:xyb:dec-jpegli 4270 520716 0.9754867 2.11857700 77.09132426 0.94801233
jpeg:enc-jpegli:q83:xyb:dec-jpegli 4270 546882 1.0245049 2.13817835 78.24623401 0.90216890
jpeg:enc-jpegli:q84:xyb:dec-jpegli 4270 573754 1.0748457 2.13797522 78.86015560 0.87542524
jpeg:enc-jpegli:q85:xyb:dec-jpegli 4270 593800 1.1123990 1.99826527 79.57994231 0.82144321
jpeg:enc-jpegli:q86:xyb:dec-jpegli 4270 626401 1.1734723 1.80742538 80.06207563 0.78408051
jpeg:enc-jpegli:q87:xyb:dec-jpegli 4270 665462 1.2466475 1.79822123 81.04907356 0.73527516
jpeg:enc-jpegli:q88:xyb:dec-jpegli 4270 686695 1.2864245 1.62478518 81.31783304 0.70662563
jpeg:enc-jpegli:q89:xyb:dec-jpegli 4270 734165 1.3753527 1.70427763 82.28476303 0.81580474
jpeg:enc-jpegli:q90:xyb:dec-jpegli 4270 778534 1.4584717 1.46675730 82.80288597 0.62502899
jpeg:enc-jpegli:q91:xyb:dec-jpegli 4270 830618 1.5560436 1.73360562 83.61609663 0.86297484
jpeg:enc-jpegli:q92:xyb:dec-jpegli 4270 891308 1.6697376 1.56471860 84.36080824 0.71326347
jpeg:enc-jpegli:q93:xyb:dec-jpegli 4270 959696 1.7978527 1.55117285 84.59052930 0.70106243
```
Encoded image with jpegli decoding (jpeg:enc-jpegli:q88:dec-jpegli:bd16):
https://cdn.discordapp.com/attachments/803645746661425173/1067261468745154681/al-elmes-zHcI7_TA5Yg-unsplash.png.bd16.dec-jpegli.q88.enc-jpegli.jpeg.png
Another example:
https://unsplash.com/photos/7Aj3WuHOgtM
[Resized source](https://user-images.githubusercontent.com/119183190/214200157-49144a63-7452-4527-94ed-5a4095d163f4.png)
```
Encoding kPixels Bytes BPP Max norm SSIMULACRA2 pnorm
----------------------------------------------------------------------------------------------------
jpeg:enc-jpegli:q77:dec-jpegli 5532 486612 0.7036439 133.04379272 -14.49615687 22.28802677
jpeg:enc-jpegli:q78:dec-jpegli 5532 501600 0.7253167 133.05514526 -12.27793198 22.24198255
jpeg:enc-jpegli:q79:dec-jpegli 5532 514678 0.7442275 135.58435059 -19.15237382 23.69608168
jpeg:enc-jpegli:q80:dec-jpegli 5532 529287 0.7653522 135.58409119 -16.72918914 22.48391403
jpeg:enc-jpegli:q81:dec-jpegli 5532 549683 0.7948450 127.24156189 -2.72853989 20.07771740
jpeg:enc-jpegli:q82:dec-jpegli 5532 568713 0.8223625 127.27835083 -4.17448154 20.07561954
jpeg:enc-jpegli:q83:dec-jpegli 5532 591524 0.8553473 125.41096497 -5.51799462 19.81012298
jpeg:enc-jpegli:q84:dec-jpegli 5532 611947 0.8848791 125.40847015 -7.13729803 20.03092070
jpeg:enc-jpegli:q85:dec-jpegli 5532 643516 0.9305281 137.93171692 -11.79842853 22.90897979
jpeg:enc-jpegli:q86:dec-jpegli 5532 669916 0.9687026 137.95727539 -15.22964194 22.74869022
jpeg:enc-jpegli:q87:dec-jpegli 5532 714181 1.0327101 68.47742462 -9.79819919 13.67024590
jpeg:enc-jpegli:q88:dec-jpegli 5532 745805 1.0784386 68.33783722 -19.76422160 13.83092302
jpeg:enc-jpegli:q89:dec-jpegli 5532 799633 1.1562742 73.66311646 -18.14493015 12.63958809
jpeg:enc-jpegli:q90:dec-jpegli 5532 812859 1.1753991 66.94897461 -19.09748908 12.06024470
jpeg:enc-jpegli:q91:dec-jpegli 5532 881885 1.2752111 107.37985992 -18.71724947 16.67591801
jpeg:enc-jpegli:q92:dec-jpegli 5532 956604 1.3832552 107.28179169 -16.90508134 16.83103876
jpeg:enc-jpegli:q93:dec-jpegli 5532 1031530 1.4915987 78.80941772 -22.08231051 14.14055037
Encoding kPixels Bytes BPP Max norm SSIMULACRA2 pnorm
------------------------------------------------------------------------------------------------------
jpeg:enc-jpegli:q77:xyb:dec-jpegli 5532 435254 0.6293800 9.88808060 67.96276485 1.91975512
jpeg:enc-jpegli:q78:xyb:dec-jpegli 5532 445074 0.6435797 9.54435825 68.38660486 1.86519500
jpeg:enc-jpegli:q79:xyb:dec-jpegli 5532 460681 0.6661476 9.59946251 68.92551213 1.86795871
jpeg:enc-jpegli:q80:xyb:dec-jpegli 5532 472928 0.6838568 2.48481417 70.67981377 1.04270666
jpeg:enc-jpegli:q81:xyb:dec-jpegli 5532 492416 0.7120366 6.40297413 70.70687798 1.35580168
jpeg:enc-jpegli:q82:xyb:dec-jpegli 5532 512724 0.7414020 6.45031261 71.31698220 1.34709810
jpeg:enc-jpegli:q83:xyb:dec-jpegli 5532 530277 0.7667838 10.81851101 70.26461499 1.90298127
jpeg:enc-jpegli:q84:xyb:dec-jpegli 5532 549742 0.7949303 11.13633347 69.58965599 2.12506658
jpeg:enc-jpegli:q85:xyb:dec-jpegli 5532 570472 0.8249060 11.09750748 70.08839280 2.11339572
jpeg:enc-jpegli:q86:xyb:dec-jpegli 5532 601213 0.8693577 6.73269033 73.36321044 1.32186981
jpeg:enc-jpegli:q87:xyb:dec-jpegli 5532 626568 0.9060212 1.93637526 75.24245529 0.86236665
jpeg:enc-jpegli:q88:xyb:dec-jpegli 5532 661893 0.9571013 12.80215454 72.66036916 2.16158320
jpeg:enc-jpegli:q89:xyb:dec-jpegli 5532 700917 1.0135303 12.79846287 73.14141777 2.15291847
jpeg:enc-jpegli:q90:xyb:dec-jpegli 5532 735382 1.0633669 12.82965851 73.57953140 2.14843630
jpeg:enc-jpegli:q91:xyb:dec-jpegli 5532 779395 1.1270099 1.62044144 77.40647337 0.76951391
jpeg:enc-jpegli:q92:xyb:dec-jpegli 5532 852897 1.2332943 1.54923189 78.33942532 0.72773519
jpeg:enc-jpegli:q93:xyb:dec-jpegli 5532 913782 1.3213344 1.47751951 79.08779795 0.68925159
```
Encoded image with jpegli decoding (jpeg:enc-jpegli:q79:dec-jpegli):
https://cdn.discordapp.com/attachments/803645746661425173/1067261875961724988/ali-kazal-7Aj3WuHOgtM-unsplash.png.dec-jpegli.q79.enc-jpegli.jpeg.png
even jpegli-xyb (jpeg:enc-jpegli:q85:xyb:dec-jpegli):
https://cdn.discordapp.com/attachments/803645746661425173/1067261943603273739/ali-kazal-7Aj3WuHOgtM-unsplash.png.dec-jpegli.xyb.q85.enc-jpegli.jpeg.png | Some broken blocks with jpegli decoding | https://api.github.com/repos/libjxl/libjxl/issues/2094/comments | 3 | 2023-01-24T02:43:07Z | 2023-01-25T07:50:04Z | https://github.com/libjxl/libjxl/issues/2094 | 1,554,203,001 | 2,094 |
[
"libjxl",
"libjxl"
] | Hello,
I received a QA Notice when building on Gentoo Linux:
```
* QA Notice: Package triggers severe warnings which indicate that it
* may exhibit random runtime failures.
* /var/tmp/portage/media-libs/libjxl-0.8.0/work/libjxl-0.8.0/lib/jxl/enc_fast_lossless.cc:3432:28: warning: array subscript -1 is outside array bounds of ‘unsigned int [65536]’ [-Warray-bounds]
* Please do not file a Gentoo bug and instead report the above QA
* issues directly to the upstream developers of this software.
* Homepage: https://github.com/libjxl/libjxl
``` | warning: array subscript -1 is outside array bounds | https://api.github.com/repos/libjxl/libjxl/issues/2090/comments | 1 | 2023-01-19T16:46:00Z | 2023-01-26T08:40:45Z | https://github.com/libjxl/libjxl/issues/2090 | 1,549,542,238 | 2,090 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
I can no longer compiler libjxl from git HEAD. It fails with:
```
[281/406] : && /usr/bin/c++ -g -O2 -ffile-prefix-map=/<<PKGBUILDDIR>>=. -fstack-protector-strong -Wformat -Werror=format-security -DJXL_DEBUG_WARNING -DJXL_DEBUG_ON_ERROR -Wdate-time -D_FORTIFY_SOURCE=2 -fno-rtti -funwind-tables -fno-omit-frame-pointer -Wl,-z,relro -fPIE -pie tools/CMakeFiles/cjpeg_hdr.dir/cjpeg_hdr.cc.o -o tools/cjpeg_hdr lib/libjxl_extras-static.a tools/libjxl_tool.a lib/libjxl.a /usr/lib/x86_64-linux-gnu/libbrotlidec.so /usr/lib/x86_64-linux-gnu/libbrotlicommon.so /usr/lib/x86_64-linux-gnu/libbrotlienc.so /usr/lib/x86_64-linux-gnu/liblcms2.so /usr/lib/x86_64-linux-gnu/libtcmalloc_minimal.so lib/libjxl_threads.a /usr/lib/x86_64-linux-gnu/libjpeg.so -ljpegli-static /usr/lib/x86_64-linux-gnu/libpng.so /usr/lib/x86_64-linux-gnu/libz.so /usr/lib/x86_64-linux-gnu/libOpenEXR-3_1.so /usr/lib/x86_64-linux-gnu/libOpenEXRUtil-3_1.so /usr/lib/x86_64-linux-gnu/libOpenEXRCore-3_1.so /usr/lib/x86_64-linux-gnu/libIex-3_1.so /usr/lib/x86_64-linux-gnu/libIlmThread-3_1.so /usr/lib/x86_64-linux-gnu/libImath-3_1.so /usr/lib/x86_64-linux-gnu/libhwy.so && :
FAILED: tools/cjpeg_hdr
: && /usr/bin/c++ -g -O2 -ffile-prefix-map=/<<PKGBUILDDIR>>=. -fstack-protector-strong -Wformat -Werror=format-security -DJXL_DEBUG_WARNING -DJXL_DEBUG_ON_ERROR -Wdate-time -D_FORTIFY_SOURCE=2 -fno-rtti -funwind-tables -fno-omit-frame-pointer -Wl,-z,relro -fPIE -pie tools/CMakeFiles/cjpeg_hdr.dir/cjpeg_hdr.cc.o -o tools/cjpeg_hdr lib/libjxl_extras-static.a tools/libjxl_tool.a lib/libjxl.a /usr/lib/x86_64-linux-gnu/libbrotlidec.so /usr/lib/x86_64-linux-gnu/libbrotlicommon.so /usr/lib/x86_64-linux-gnu/libbrotlienc.so /usr/lib/x86_64-linux-gnu/liblcms2.so /usr/lib/x86_64-linux-gnu/libtcmalloc_minimal.so lib/libjxl_threads.a /usr/lib/x86_64-linux-gnu/libjpeg.so -ljpegli-static /usr/lib/x86_64-linux-gnu/libpng.so /usr/lib/x86_64-linux-gnu/libz.so /usr/lib/x86_64-linux-gnu/libOpenEXR-3_1.so /usr/lib/x86_64-linux-gnu/libOpenEXRUtil-3_1.so /usr/lib/x86_64-linux-gnu/libOpenEXRCore-3_1.so /usr/lib/x86_64-linux-gnu/libIex-3_1.so /usr/lib/x86_64-linux-gnu/libIlmThread-3_1.so /usr/lib/x86_64-linux-gnu/libImath-3_1.so /usr/lib/x86_64-linux-gnu/libhwy.so && :
/usr/bin/ld: cannot find -ljpegli-static: No such file or directory
collect2: error: ld returned 1 exit status
```
**To Reproduce**
cmake options:
```
-DBUILD_TESTING:BOOL=ON -DCMAKE_BUILD_RPATH_USE_ORIGIN=ON -DJPEGXL_BUNDLE_SKCMS:BOOL=OFF -DJPEGXL_ENABLE_BENCHMARK:BOOL=ON -DJPEGXL_ENABLE_DEVTOOLS:BOOL=ON -DJPEGXL_ENABLE_DOXYGEN:BOOL=OFF -DJPEGXL_ENABLE_EXAMPLES:BOOL=ON -DJPEGXL_ENABLE_JNI:BOOL=ON -DJPEGXL_ENABLE_JPEGLI:BOOL=OFF -DJPEGXL_ENABLE_OPENEXR:BOOL=ON -DJPEGXL_ENABLE_PLUGINS:BOOL=OFF -DJPEGXL_ENABLE_SJPEG:BOOL=OFF -DJPEGXL_ENABLE_SKCMS:BOOL=OFF -DJPEGXL_ENABLE_TCMALLOC:BOOL=ON -DJPEGXL_FORCE_SYSTEM_BROTLI:BOOL=ON -DJPEGXL_FORCE_SYSTEM_GTEST:BOOL=ON -DJPEGXL_FORCE_SYSTEM_HWY:BOOL=ON -DJPEGXL_FORCE_SYSTEM_LCMS2:BOOL=ON -DJPEGXL_INSTALL_JARDIR:PATH=share/java -DJPEGXL_INSTALL_JNIDIR:PATH=lib/x86_64-linux-gnu/jni -DJPEGXL_TEST_DATA_PATH:PATH=/usr/share/libjxl-testdata
```
**Expected behavior**
libjxl should compile without errors
**Screenshots**
* https://buildd.debian.org/status/fetch.php?pkg=jpeg-xl&arch=amd64&ver=0.8.0%7Egit20230118.bcacd51-1&stamp=1674124116&raw=0
**Environment**
- OS: Linux
- Compiler version: gcc 12
- CPU type: x86_64
- cjxl/djxl version string: git bcacd51
| /usr/bin/ld: cannot find -ljpegli-static: No such file or directory | https://api.github.com/repos/libjxl/libjxl/issues/2087/comments | 2 | 2023-01-19T13:07:20Z | 2023-01-26T08:17:23Z | https://github.com/libjxl/libjxl/issues/2087 | 1,549,120,898 | 2,087 |
[
"libjxl",
"libjxl"
] | Currently the meaning for cjxl:
```
--num_threads=N
Number of worker threads (-1 == use machine default, 0 == do not use multithreading).
```
and for djxl it is:
```
--num_threads=N
Sets the number of threads to use. The default 0 value means the machine default.
```
This should be changed so that both cjxl and djxl function in the same way
| harmonize --num_threads flag between cjxl and djxl | https://api.github.com/repos/libjxl/libjxl/issues/2073/comments | 0 | 2023-01-16T00:51:27Z | 2023-01-16T10:19:55Z | https://github.com/libjxl/libjxl/issues/2073 | 1,534,083,557 | 2,073 |
[
"libjxl",
"libjxl"
] | Re-enable the github ci tests for Windows MSYS2 clang32 and clang64.
Currently webp fails to build
| Windows MSYS2 clang32 and clang64 | https://api.github.com/repos/libjxl/libjxl/issues/2072/comments | 9 | 2023-01-15T23:01:08Z | 2024-03-04T10:24:19Z | https://github.com/libjxl/libjxl/issues/2072 | 1,534,041,148 | 2,072 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
When converting JPEG images to JXL with lossy compression, some pictures will convert too dark. I've included the files and examples. I've only experienced this with some older exports from Lightroom 6.14, although I haven't tested on that many images from different sources.
When converting back to JPEG with _djxl_, pictures again gain their original gamma/brightness.
**To Reproduce**
I don't know, I've only experienced this with older exports I still have saved as JPEGs. I've tried opening the JXL files with XnViewMP ver. 1.4.0 and ImageGlass 8.7.11.6.
**Expected behavior**
The pictures should keep the same gamma/brightness.
**Screenshots**
**Environment**
- OS: Windows 10
- cjxl/djxl version string: cjxl v0.7.0 f95da13 [AVX2,SSE4,SSSE3,Unknown] _**AND**_ cjxl v0.8.0 fadac12 [AVX2,SSE4,SSSE3,Unknown]
Attachments:
[Files (JPG + JXL).zip](https://github.com/libjxl/libjxl/files/10420513/Files.JPG.%2B.JXL.zip)
[Comparison Picture.zip](https://github.com/libjxl/libjxl/files/10420514/Comparison.Picture.zip)
| Some images convert too dark when using lossy compression | https://api.github.com/repos/libjxl/libjxl/issues/2070/comments | 10 | 2023-01-15T15:24:59Z | 2023-08-28T13:33:55Z | https://github.com/libjxl/libjxl/issues/2070 | 1,533,867,183 | 2,070 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
When using as input a simple b/w checkerboard pattern the lossless compression is fine. But using VarDCT (default settings) the result shows a good visible 64 by 64 px pattern if viewing the resulting jxl (also when back converting to png using djxl).
**To Reproduce**
Using "JPEG XL encoder v0.9.0 cf939d7 [AVX2,SSE4,SSSE3,Unknown]" from a checkerboard (white and black pixel by pixel) picture (PNG): -d 1 -e 7 (default)
**Expected behavior**
No or not so obvious artifacts.
**Additional context**
I uploaded my used checkerboard file as lossless jxl (with jpg-ending to be able to upload it -- please rename after download)
Please use djxl to create the larger (in byte size) PNG for test.
| 64x64 Pattern artifact at checkerboard input | https://api.github.com/repos/libjxl/libjxl/issues/2067/comments | 4 | 2023-01-13T18:09:34Z | 2025-01-23T20:49:41Z | https://github.com/libjxl/libjxl/issues/2067 | 1,532,718,753 | 2,067 |
[
"libjxl",
"libjxl"
] | cjxl can read sBIT chunk in PNG and use the information to encode JXL using the appropriate sample bit depth. The bit depth is correctly reported by jxlinfo.
djxl does not recreate the sBIT chunk so the information about the original bit depth is not preserved. If bit depth used by the JXL file is other than 8 bits or 16 bits, djxl should add sBIT chunk to the PNG file. | djxl does not generate sBIT chunk in PNG | https://api.github.com/repos/libjxl/libjxl/issues/2066/comments | 0 | 2023-01-13T17:53:00Z | 2023-05-09T09:51:58Z | https://github.com/libjxl/libjxl/issues/2066 | 1,532,698,374 | 2,066 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
When encoding lossy 32 bit float image with extra channel (eg. RGB+Alpha / GRAY+Alpha), the filesize results are a lot bigger than without extra channel.
In my test results here: [32bit-float-images](https://github.com/kampidh/Colour-Gamut-Plotter/tree/jpegxl-test-images/jpegxl-reports/32bit-float-bug) / [32bit-float-images-gdrive](https://drive.google.com/file/d/1RMKywukbnNFs3URTkexJYeMpE0UJLEXa/view?usp=share_link), the difference can be very significant:
Or even result a bigger file than lossless on grayscale + alpha:
Tried both with internal XYB and original profile yield the same results. And furthermore, when Modular mode is used instead, the output results are artifacted (looks posterized, see screenshot below).
**To Reproduce**
Steps to reproduce the behavior:
Eg. in Krita, we give these parameters to the encoder for RGBA float32 images before image frame is added:
```
pixel format:
data_type = JXL_TYPE_FLOAT;
num_channels = 4;
basic info:
bits_per_sample = 32;
exponent_bits_per_sample = 8;
alpha_bits = 32;
alpha_exponent_bits = 8;
num_color_channels = 3;
num_extra_channels = 1;
```
The rest are left default (distance 1, effort 7).
And even if `JxlEncoderSetExtraChannelInfo` also called for the alpha channel, the results are the same.
**Expected behavior**
Better compression results, or at least not too far from the one without extra channels
**Screenshots**
The output result when compressed using Modular mode:
**Environment**
- OS: Windows 10 21H2
- Compiler version: LLVM 14.0.0
- CPU type: x86_64
- cjxl/djxl version string: v0.7.0 [AVX2,SSE4,SSSE3,Unknown]
**Additional context**
Also reproducible when built with v0.8.0rc1
<!--
Currently github does not allow uploading files that end in `.jxl`, but when you
rename them for example as `image.jxl.jpg`, it will be possible to upload them
and also view them in browsers that are configured to support it.
See https://github.com/orgs/github-community/discussions/18139
-->
| Encoding lossy 32-bit float with extra channel(s) yield poor compression results | https://api.github.com/repos/libjxl/libjxl/issues/2064/comments | 2 | 2023-01-13T14:25:32Z | 2025-01-18T00:59:53Z | https://github.com/libjxl/libjxl/issues/2064 | 1,532,382,267 | 2,064 |
[
"libjxl",
"libjxl"
] | Hello,
I have book scans stored in JXL format, and I use the "djxl" command to convert them to PNG format whenever I have to manipulate them. However, I'm noticing that "djxl" is awfully unreliable. It would produce all-black PNGs from select JXLs. I am simply running:
`djxl book001.jxl book001.png`
and expecting it to work! My only relief is that the JXLs themselves aren't corrupted (I almost had a heart attack when I first noticed the problem). ImageMagick is able to convert them flawlessly:
`convert book001.jxl book001.png`
You will find attached to this post a ZIP containing:
- book001.jxl and book004.jxl - original JXLs
- book001_djxl.png and book004_djxl.png - JXLs converted to PNG format using "djxl" (they are all-black, unexpected result)
- book001_imagemagick.png and book004_imagemagick.png - JXLs converted to PNG format using ImageMagick (expected result)
[Archive.zip](https://github.com/libjxl/libjxl/files/10412287/Archive.zip)
I am using jpeg-xl: stable 0.7.0 (bottled), which is the latest version provided by Homebrew. | djxl is producing all-black PNGs from select JXLs | https://api.github.com/repos/libjxl/libjxl/issues/2063/comments | 4 | 2023-01-13T14:25:25Z | 2023-01-25T09:51:14Z | https://github.com/libjxl/libjxl/issues/2063 | 1,532,382,116 | 2,063 |
[
"libjxl",
"libjxl"
] | Consider developing an image synthesizer language that allows a fuzzer to manipulate a somewhat dense bitstream related to creating images that are giving good coverage for the encode-decode to explore 1:1 correctness between the original and the decompressed image.
Then run this language with a fuzzer. | Run synthesize-encode-decode fuzzer to verify lossless | https://api.github.com/repos/libjxl/libjxl/issues/2050/comments | 0 | 2023-01-11T13:47:06Z | 2023-01-25T09:32:50Z | https://github.com/libjxl/libjxl/issues/2050 | 1,529,077,084 | 2,050 |
[
"libjxl",
"libjxl"
] | Hello,
It seems the number of bits per sample from `basic_info` is ignored in lossy mode.
For instance, when using 3 bits per samples, I expect strong banding. This is the case when using lossless compression, but does not happen with lossy mode. It seems the image does not go below 8 bits in lossy mode.
**To Reproduce**
Here is a sample program showcasing this effect: https://github.com/afichet/jxl_compression_quantization
```
git clone https://github.com/afichet/jxl_compression_quantization.git
cd xl_compression_quantization
mkdir build
cd build
cmake ..
make
./test_jxl
```
**Expected behavior**
I am expecting banding and lower image size in lossy mode when reducing the number of bits per sample.
**Screenshots**
Lossless with 3 bits per pixel:
Lossy with 3 bits per pixel
**Environment**
- OS: Arch Linux
- Compiler version: clang 14.0.
- CPU type: x86_64
- cjxl/djxl version string: v0.7.0
| Bits per sample ignored in lossy mode | https://api.github.com/repos/libjxl/libjxl/issues/2049/comments | 1 | 2023-01-11T11:05:35Z | 2025-01-18T00:57:07Z | https://github.com/libjxl/libjxl/issues/2049 | 1,528,841,823 | 2,049 |
[
"libjxl",
"libjxl"
] | Hello,
I cannot use lossless compression with a specific image. Lossy compression works on the same image. The library returns an error when calling `JxlEncoderProcessOutput`.
To reproduce the issue, you can clone this repository: https://github.com/afichet/jxl_lossless_lossy
```
git clone https://github.com/afichet/jxl_lossless_lossy.git
cd jxl_lossless_lossy
mkdir build
cd build
cmake ..
make
./test_jxl ../data/dc.pfm
```
The problematic file is provided in `data/dc.pfm`.
The file is a 32 bits gray float image.
**Environment**
- OS: Arch Linux
- Compiler version: clang 14.0.6
- CPU type: x86_64
- cjxl/djxl version string: v0.7.0 | Cannot compress a gray float image with lossless compression | https://api.github.com/repos/libjxl/libjxl/issues/2046/comments | 7 | 2023-01-10T14:27:40Z | 2025-01-18T00:54:06Z | https://github.com/libjxl/libjxl/issues/2046 | 1,527,482,598 | 2,046 |
[
"libjxl",
"libjxl"
] | Converting the attached PNG using cjxl with `-d 0 -e 1` produces a JXL that can't be decoded. No other distance/effort is affected.
The bug seems to have been introduced with a5417e0.
**To Reproduce**
```
$ cjxl hazardous_pixels.png hazardous_pixels.jxl -d 0 -e 1
JPEG XL encoder v0.8.0 25262231 [AVX2,SSE4,SSSE3,Unknown]
./lib/extras/dec/color_hints.cc:54: No color_space/icc_pathname given, assuming sRGB
Read 451x384 image, 1137 bytes, 216.8 MP/s
Encoding [Modular, lossless, effort: 1],
Compressed to 695 bytes (0.032 bpp).
451 x 384, 92.57 MP/s [92.57, 92.57], 1 reps, 8 threads.
$ djxl hazardous_pixels.jxl --disable_output
JPEG XL decoder v0.8.0 25262231 [AVX2,SSE4,SSSE3,Unknown]
Read 695 compressed bytes.
./lib/jxl/modular/encoding/encoding.cc:566: JXL_FAILURE: Truncated input
./lib/jxl/modular/encoding/encoding.cc:597: JXL_RETURN_IF_ERROR code=1: dec_status
./lib/jxl/dec_modular.cc:357: JXL_RETURN_IF_ERROR code=1: status
./lib/jxl/dec_frame.cc:490: JXL_RETURN_IF_ERROR code=1: modular_frame_decoder_.DecodeGroup( mrect, br[i - pass0], minShift, maxShift, ModularStreamId::ModularAC(ac_group_id, i), false, dec_state_, &render_pipeline_input, false, &modular_pass_ready)
./lib/jxl/modular/encoding/encoding.cc:566: JXL_FAILURE: Truncated input
./lib/jxl/modular/encoding/encoding.cc:597: JXL_RETURN_IF_ERROR code=1: dec_status
./lib/jxl/dec_modular.cc:357: JXL_RETURN_IF_ERROR code=1: status
./lib/jxl/dec_frame.cc:490: JXL_RETURN_IF_ERROR code=1: modular_frame_decoder_.DecodeGroup( mrect, br[i - pass0], minShift, maxShift, ModularStreamId::ModularAC(ac_group_id, i), false, dec_state_, &render_pipeline_input, false, &modular_pass_ready)
./lib/jxl/dec_frame.cc:717: JXL_FAILURE: Error in AC group
./lib/jxl/decode.cc:1139: frame out of bounds
Failed to decode image
DecompressJxlToPackedPixelFile failed
```
**Expected behavior**
Successful lossless encoding/decoding.
**Environment**
- OS: Arch Linux
- Compiler version: clang 14.0.6
- CPU type: x86_64
- cjxl/djxl version string: JPEG XL encoder v0.8.0 25262231 [AVX2,SSE4,SSSE3,Unknown]
| Fast lossless can produce invalid JXLs | https://api.github.com/repos/libjxl/libjxl/issues/2039/comments | 0 | 2023-01-08T16:14:35Z | 2023-01-09T15:42:48Z | https://github.com/libjxl/libjxl/issues/2039 | 1,524,573,823 | 2,039 |
[
"libjxl",
"libjxl"
] | Hello,
I see JXL generates significant color distortion
Original:
Converted to JXL:
The top of the head is grey in original, but after converting to JXL it becomes red and blue.
Is there any encoder option to avoid this?
These are my encoding options:
Flt quality=0.97f;
Flt compression_level=0.9f;
```
Bool lossless=(quality>=1);
basic_info.uses_original_profile=lossless; // headers say this should be false for lossy, this needs to be true for lossless because 'JxlEncoderSetFrameLossless' will fail
if(JXL_ENC_SUCCESS!=JxlEncoderSetBasicInfo(enc.get(), &basic_info))goto error;
JxlColorEncoding color_encoding;
JxlColorEncodingSetToSRGB(&color_encoding, !color);
if(JXL_ENC_SUCCESS!=JxlEncoderSetColorEncoding(enc.get(), &color_encoding))goto error;
JxlEncoderFrameSettings *frame_settings=JxlEncoderFrameSettingsCreate(enc.get(), null);
JxlEncoderFrameSettingsSetOption(frame_settings, JXL_ENC_FRAME_SETTING_KEEP_INVISIBLE, 1);
if(compression_level>=0)JxlEncoderFrameSettingsSetOption(frame_settings, JXL_ENC_FRAME_SETTING_EFFORT, Mid(RoundPos(Lerp(1, 9, compression_level)), 1, 9));
if(quality >=0)
{
JxlEncoderSetFrameDistance(frame_settings, Lerp(15, 0, Sat(quality)));
JxlEncoderSetFrameLossless(frame_settings, lossless);
}
``` | Color Distortion | https://api.github.com/repos/libjxl/libjxl/issues/2035/comments | 2 | 2023-01-06T04:22:07Z | 2025-01-18T00:47:45Z | https://github.com/libjxl/libjxl/issues/2035 | 1,521,906,242 | 2,035 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
failed encoding from jpg with -e10
**To Reproduce**
```shell
cjxl flower.png.im_q85_rgb.jpg out.jxl --effort=10 --verbose
```
(found [flower.png.im_q85_rgb.jpg](https://github.com/libjxl/testdata/blob/81c7e5247c57051bf077b879ee38efd4c309c9e2/jxl/flower/flower.png.im_q85_rgb.jpg) in https://github.com/libjxl/testdata)
**Expected behavior**
exit code 0 with `out.jxl` as output
**Screenshots**
exit code 1
```
JPEG XL encoder v0.8.0 ee70a3ff [AVX2,SSE4,SSSE3,Unknown]
Note: Implicit-default for JPEG is lossless-transcoding. To silence this message, set --lossless_jpeg=(1|0).
Read JPEG image with 1382645 bytes.
Encoding [Container | JPEG, lossless transcode, effort: 10 | JPEG reconstruction data],
JxlEncoderProcessOutput failed.
EncodeImageJXL() failed.
```
**Environment**
- OS: Linux
- Compiler version: clang 15.0.6
- CPU type: x86_64
- cjxl/djxl version string: cjxl v0.8.0 ee70a3ff [AVX2,SSE4,SSSE3,Unknown]
**Additional context**
encoding with -e9 succeed.
<!--
Currently github does not allow uploading files that end in `.jxl`, but when you
rename them for example as `image.jxl.jpg`, it will be possible to upload them
and also view them in browsers that are configured to support it.
See https://github.com/orgs/github-community/discussions/18139
-->
| failed encoding from jpg with -e10 | https://api.github.com/repos/libjxl/libjxl/issues/2034/comments | 3 | 2023-01-06T02:30:45Z | 2023-01-24T21:01:26Z | https://github.com/libjxl/libjxl/issues/2034 | 1,521,778,962 | 2,034 |
[
"libjxl",
"libjxl"
] | It actually returns the entire raw box size as the name suggests, including the 4 bytes of the box length, and the 4 bytes of the box type (at least using libjxl 0.7.0), vs. just the box content length as claimed.
So either the API documentation needs to be updated, or the implementation needs to change to match (breaking)... | JxlDecoderGetBoxSizeRaw documentation is incorrect | https://api.github.com/repos/libjxl/libjxl/issues/2022/comments | 0 | 2023-01-03T14:44:25Z | 2024-03-04T10:12:56Z | https://github.com/libjxl/libjxl/issues/2022 | 1,517,488,452 | 2,022 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
Failed to build with vcpkg.
**To Reproduce**
Steps to reproduce the behavior:
Followed the vcpkg build instructions.
**Expected behavior**
libjxl successfully build.
**Screenshots**
If applicable, add screenshots or example input/output images to help explain your problem.
**Environment**
- OS: Windows 11 22H2 build 22621.963
- Compiler version: clang 11.0.1
- CPU type: x86_64
- cjxl/djxl version string: [e.g. cjxl [v0.3.7 | SIMD supported: SSE4,Scalar]]
- Microsoft Visual Studio Community 2022 (64-bit) - Version 17.4.3
**Additional context**
I get this error
```
Error CMake Error at C:/Users/Alifian/scoop/apps/vcpkg/current/installed/x64-windows/share/gtest/GTestTargets.cmake:42 (message):
Some (but not all) targets in this export set were already defined.
Targets Defined: GTest::gtest, GTest::gtest_main
Targets not yet defined: GTest::gmock, GTest::gmock_main C:/Users/Alifian/scoop/apps/vcpkg/current/installed/x64-windows/share/gtest/GTestTargets.cmake 42
```
with vcpkg version 2022-11-10-5fdee72bc1fceca198fb1ab7589837206a8b81ba
[CMake log files](https://drive.google.com/file/d/1unMowlCKAgG8tFAVUVZ_LHJIv1vh_U-9/view?usp=sharing) | Failed to build with vcpkg | https://api.github.com/repos/libjxl/libjxl/issues/2017/comments | 2 | 2023-01-03T05:12:47Z | 2023-03-27T14:07:27Z | https://github.com/libjxl/libjxl/issues/2017 | 1,516,938,845 | 2,017 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
In `djxl` v0.6.1 and earlier, decoding JXL animations would dump all image frames. Since v0.7.0, it only outputs 1 file and I cannot find a relevant option for it in the manpage.
**To Reproduce**
Steps to reproduce the behavior:
1. Unzip and use this as the input file -
[animation.zip](https://github.com/libjxl/libjxl/files/10332992/animation.zip)
2. Run `djxl animated.jxl output.png`
3. Observe that only 1 output file was generated
**Expected behavior**
Because the input is an animation, it should generate multiple image frames: `output-0.png`, `output-1.png`...
**Environment**
- OS: Windows 10, Ubuntu 20.04
- CPU type: x86_64
- cjxl/djxl version string: djxl v0.7.0 [SSE4,SSSE3,Unknown]
**Additional context**
I tried using wildcard (*) in the output filename, which didn't help either.
<!--
Currently github does not allow uploading files that end in `.jxl`, but when you
rename them for example as `image.jxl.jpg`, it will be possible to upload them
and also view them in browsers that are configured to support it.
See https://github.com/orgs/github-community/discussions/18139
-->
| djxl no longer outputs all frames of animations | https://api.github.com/repos/libjxl/libjxl/issues/2016/comments | 3 | 2023-01-02T18:07:06Z | 2023-11-26T14:52:35Z | https://github.com/libjxl/libjxl/issues/2016 | 1,516,616,922 | 2,016 |
[
"libjxl",
"libjxl"
] | Hi,
I use the lib not to convert pictures of any format, but to save pictures from an image processing software.
This way I have EXIF data or XMP data and want to save it inside the JPEG XL file.
There is a sample code, how to load the EXIF data, but I could find sample code how to write.
Can you help me? | Save EXIF box Sample Code | https://api.github.com/repos/libjxl/libjxl/issues/2001/comments | 3 | 2022-12-28T09:40:40Z | 2022-12-29T08:14:36Z | https://github.com/libjxl/libjxl/issues/2001 | 1,512,607,557 | 2,001 |
[
"libjxl",
"libjxl"
] | Thanks for making an excellent image codec!
I've just integrated it into my "Titan" game engine
https://esenthel.com/
https://github.com/Esenthel/EsenthelEngine/
Perhaps you want to include it on your list - https://github.com/libjxl/libjxl/blob/main/doc/software_support.md
Where can I show my support to contact Google so they integrate it into Chrome?
Is there any email I can contact or a change.org campaign? | JPEG XL software support - Titan Engine | https://api.github.com/repos/libjxl/libjxl/issues/2000/comments | 2 | 2022-12-28T05:44:33Z | 2023-01-06T04:17:30Z | https://github.com/libjxl/libjxl/issues/2000 | 1,512,425,032 | 2,000 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
cjxl fails to create readable jxl file.
**To Reproduce**
Steps to reproduce the behavior:
```
$ cjxl -q 70 -j 0 IMG_20160422_102122.jpg i2.jxl
JPEG XL encoder v0.7.0 f95da131 [SSE4,SSSE3,Unknown]
Read 2560x1536 image, 1574521 bytes, 184.2 MP/s
Encoding [Container | VarDCT, d2.800, effort: 7 | 602-byte Exif],
Compressed to 477184 bytes including container (0.971 bpp).
2560 x 1536, 2.38 MP/s [2.38, 2.38], 1 reps, 4 threads.
$ jxlinfo i2.jxl
JPEG XL file format container (ISO/IEC 18181-2)
JPEG XL image, 2560x1536, lossy, 8-bit RGB
Color space: RGB, D65, sRGB primaries, sRGB transfer function, rendering intent: Perceptual
Unexpected EOF
$ cjxl -V
cjxl v0.7.0 f95da131 [SSE4,SSSE3,Unknown]
Copyright (c) the JPEG XL Project
```
**Environment**
- OS: Arch linux with community/libjxl 0.7.0-3
- CPU type: x86_64
- cjxl/djxl version string: cjxl v0.7.0 f95da131 [SSE4,SSSE3,Unknown]
[link to the test file with the broken result](https://mega.nz/folder/VFMglCAa#-VUU_5JHlOj6McdUA8k7Xg) | Cannot re-encode jpg | https://api.github.com/repos/libjxl/libjxl/issues/1989/comments | 4 | 2022-12-20T12:58:44Z | 2025-01-18T00:42:23Z | https://github.com/libjxl/libjxl/issues/1989 | 1,504,521,169 | 1,989 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
```cjxl ~/testdata/jxl/flower/flower_cropped.jpg flower.jxl -p``` fails with
```
JPEG XL encoder v0.8.0 b28a0114 [AVX3,AVX2,SSE4,SSSE3,Unknown]
Read JPEG image with 197465 bytes.
Encoding [Container | JPEG, lossless transcode, effort: 7 | JPEG reconstruction data],
Note: Implicit-default for JPEG is lossless-transcoding. To silence this message, set --lossless_jpeg=(1|0).
./lib/jxl/enc_frame.cc:900: JXL_CHECK: enc_state_->passes.size() == 1
```
This bug was introduced in https://github.com/libjxl/libjxl/pull/1689. | cjxl with -p for jpg | https://api.github.com/repos/libjxl/libjxl/issues/1988/comments | 1 | 2022-12-20T10:27:24Z | 2025-01-18T10:53:51Z | https://github.com/libjxl/libjxl/issues/1988 | 1,504,329,695 | 1,988 |
[
"libjxl",
"libjxl"
] | **To Reproduce**
1. Download [this PNG](https://github.com/libjxl/conformance/raw/master/testcases/bench_oriented_brg/ref.png) which is taken from the conformance repo
2. Create XYB JPEG from it: `benchmark_xl --input=ref.png --codec=jpeg:libjxl:d1.0 --save_compressed --output_dir=./out`
3. Encode it to JXL, using the default lossless JPEG transcode: `cjxl ref.png.d1.0.libjxl.jpeg ref.png.d1.0.libjxl.jxl`
4. Try to decode it back to original JPEG: `jxl ref.png.d1.0.libjxl.jxl orig.jpg`
**Expected behavior**
djxl outputs an JPEG file which is identical to the original XYB JPEG.
**Actual behavior**
djxl fails with the following message:
```
JPEG XL decoder v0.7.0 f95da131 [AVX2,SSE4,SSSE3,Unknown]
Read 119401 compressed bytes.
Failed to decode image
Warning: could not decode losslessly to JPEG. Retrying with --pixels_to_jpeg...
Failed to decode image
DecompressJxlToPackedPixelFile failed
```
**Environment**
Arch x86_64 `cjxl`/`djxl` 0.7.0 (Arch Linux package) and `benchmark_xl` 0.8.0 `0a6bafee` (nightly CI artifact) | cjxl can encode XYB JPEG, but djxl fails to decode it | https://api.github.com/repos/libjxl/libjxl/issues/1986/comments | 1 | 2022-12-19T11:17:54Z | 2024-03-04T10:12:56Z | https://github.com/libjxl/libjxl/issues/1986 | 1,502,763,420 | 1,986 |
[
"libjxl",
"libjxl"
] | see here https://github.com/golang/go/issues/57351#issuecomment-1354558729 and discussion | better support golang | https://api.github.com/repos/libjxl/libjxl/issues/1981/comments | 2 | 2022-12-16T11:08:14Z | 2023-10-08T18:01:16Z | https://github.com/libjxl/libjxl/issues/1981 | 1,500,019,648 | 1,981 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
I'm transcoding a large (~1.3M) image dataset of JPEGs to JXL using the lossless transcoding. For 864 images I get the following error:
```
JPEG XL encoder v0.8.0 091b60e5 [AVX3,AVX2,SSE4,SSSE3,Unknown]
Read JPEG image with 43279 bytes.
Encoding [Container | JPEG, lossless transcode, effort: 7 | JPEG reconstruction data],
JxlEncoderAddJPEGFrame() failed.
EncodeImageJXL() failed.
```
running in verbose mode does not give any extra information.
I've attached an example inside a zip file (in case github tries to mess with the image)
[n03476991_52119.zip](https://github.com/libjxl/libjxl/files/10230108/n03476991_52119.zip)
**To Reproduce**
```
cjxl --lossless_jpeg=1 n03476991_52119.JPEG ./test.jxl
```
should reproduce the failure
**Expected behavior**
A JXL trancoding is produced
**Environment**
- OS: Linux
- Compiler version: n/a
- CPU type: x86_64
- cjxl/djxl version string: JPEG XL encoder v0.8.0 091b60e5 [AVX3,AVX2,SSE4,SSSE3,Unknown]
| JxlEncoderAddJPEGFrame() failed | https://api.github.com/repos/libjxl/libjxl/issues/1973/comments | 4 | 2022-12-14T17:13:45Z | 2024-01-31T13:35:03Z | https://github.com/libjxl/libjxl/issues/1973 | 1,497,064,383 | 1,973 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
A clear and concise description of what the bug is.
benchmark_xl --save_decompressed --write_html_report works in 0.6 but not in 0.7 or head
**To Reproduce**
Steps to reproduce the behavior:
cmake --build . -- -j$(nproc) && tools/benchmark_xl --input "$HOME/newcorpus/split/*.png" --codec jxl:d0.5:epf0,jxl:d1:epf0,jxl:d3:epf0,jxl:d5:epf0,jxl:d7:epf0,jxl:d15:epf0 --output_dir=/tmp/jxl_opt/ --save_decompressed --write_html_report --more_columns
a msg saying:
./lib/jxl/dec_external_image.cc:260: JXL_CHECK: float_out ? bits_per_sample == 16 || bits_per_sample == 32 : bits_per_sample > 0 && bits_per_sample <= 16
Illegal instruction
appears
**Expected behavior**
A clear and concise description of what you expected to happen.
files like '/tmp/opt/index.jxl_d1_epf0.html' will be written with their respective png files
**Screenshots**
If applicable, add screenshots or example input/output images to help explain your problem.
**Environment**
usual linux
**Additional context**
because of this I'm developing in 0.6 but it would be nicer to make quality improvements directly in head | benchmark_xl --save_decompressed --write_html_report works in 0.6 but not in 0.7 or head | https://api.github.com/repos/libjxl/libjxl/issues/1970/comments | 1 | 2022-12-14T11:10:27Z | 2022-12-20T15:01:44Z | https://github.com/libjxl/libjxl/issues/1970 | 1,496,373,343 | 1,970 |
[
"libjxl",
"libjxl"
] | Using `cjxl` trying to encode a JPEG file and using low quality values returns an error:
When using `-q 5` it works as expected:
```sh
$ cjxl input.jpeg output.jxl -q 5 --lossless_jpeg=1
JPEG XL encoder v0.7.0 0.7.0 [AVX2,SSE4,SSSE3]
Read JPEG image with 42537 bytes.
Encoding [Container | JPEG, lossless transcode, effort: 7 | JPEG reconstruction data],
Compressed to 9172 bytes including container
```
But when using values lower than `5` a `Setting frame distance failed` error is showed without generating output:
```sh
$ cjxl input.jpeg output.jxl -q 4 --lossless_jpeg=1
JPEG XL encoder v0.7.0 0.7.0 [AVX2,SSE4,SSSE3]
Setting frame distance failed.
```
| Using cjxl with a quality value lower than 5 returns an error | https://api.github.com/repos/libjxl/libjxl/issues/1968/comments | 1 | 2022-12-13T20:22:45Z | 2022-12-21T07:47:46Z | https://github.com/libjxl/libjxl/issues/1968 | 1,495,091,238 | 1,968 |
[
"libjxl",
"libjxl"
] | /libjxl-main/lib/extras/dec/pgx.cc line 154
Code:
Parser parser(bytes);
HeaderPGX header = {};
const uint8_t* pos; // <- maybe better init
if (!parser.ParseHeader(&header, &pos)) return false;
| init variable | https://api.github.com/repos/libjxl/libjxl/issues/1964/comments | 2 | 2022-12-13T10:42:49Z | 2022-12-20T08:46:59Z | https://github.com/libjxl/libjxl/issues/1964 | 1,493,952,067 | 1,964 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
When decoding a truncated progressive .jxl with `djxl --allow_partial_files`, the following warning is printed:
```
JPEG XL decoder v0.7.0 f95da131 [AVX3,AVX2,SSE4,SSSE3,Unknown]
Read 400000 compressed bytes.
Input file is truncated and allow_partial_input was disabled.Warning: could not decode losslessly to JPEG. Retrying with --pixels_to_jpeg...
Decoded to pixels.
3000 x 4000, 57.92 MP/s [57.92, 57.92], 1 reps, 8 threads.
```
**To Reproduce**
[1935.zip](https://github.com/libjxl/libjxl/files/10123669/1935.zip)
The .zip above has a .jpg file named 1935.jpg
```bash
$bash: cjxl 1935.jpg 1935.jxl -p
# truncate .jxl to first 400kb
$bash: dd if=1935.jxl of=1935t.jxl bs=1 count=400000
# decode to jpg
$bash: djxl 1935t.jxl 1935t.jpg --allow_partial_files
JPEG XL decoder v0.7.0 f95da131 [AVX3,AVX2,SSE4,SSSE3,Unknown]
Read 400000 compressed bytes.
Input file is truncated and allow_partial_input was disabled.Warning: could not decode losslessly to JPEG. Retrying with --pixels_to_jpeg...
Decoded to pixels.
3000 x 4000, 68.78 MP/s [68.78, 68.78], 1 reps, 8 threads.
```
**Expected behavior**
`allow_partial_input` shouldn't be disabled because ` --allow_partial_files` is set. `allow_partial_input` is also not a cli option for djxl as of 0.7.
**Environment**
- OS: `Linux 6.0.9-arch1-1 #1 SMP PREEMPT_DYNAMIC Wed, 16 Nov 2022 17:01:17 +0000 x86_64 GNU/Linux`
- Compiler version: N/A. Using binaries from `jxl-linux-x86_64-static-v0.7.0.tar.gz` of [release 0.7.0](https://github.com/libjxl/libjxl/releases/tag/v0.7.0)
- CPU type: x86_64
- cjxl/djxl version string: Encoded with `JPEG XL encoder v0.7.0 f95da131 [AVX3,AVX2,SSE4,SSSE3,Unknown]`, decoded with `JPEG XL decoder v0.7.0 f95da131 [AVX3,AVX2,SSE4,SSSE3,Unknown]` | `allow_partial_input` disabled despite `--allow_partial_files` | https://api.github.com/repos/libjxl/libjxl/issues/1935/comments | 1 | 2022-11-30T13:49:52Z | 2022-12-21T07:44:17Z | https://github.com/libjxl/libjxl/issues/1935 | 1,469,668,314 | 1,935 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
When convert/exporting a grayscale image with custom white point (in my case D50) to lossy jxl, profile gets discarded and returned as D65_linear when converted back with `djxl`. Even though that `jxlinfo` returns a correct profile info:
```
jxlinfo CC-gray-D50-srg-cjxl.jxl
JPEG XL image, 748x1000, lossy, 8-bit Grayscale+Alpha
Color space: Grayscale, Custom, white_point(x=0.345703,y=0.358538), sRGB transfer function, rendering intent: Perceptual
```
This results in a darker image.
**To Reproduce**
Convert a grayscale image with custom white point with `cjxl` lossy settings.
-or-
Call `JxlEncoderSetColorEncoding` with `color_space = JXL_COLOR_SPACE_GRAY` and `white_point = JXL_WHITE_POINT_CUSTOM` then set the custom white point and TRC accordingly.
And then convert it back to using `djxl`.
**Expected behavior**
The profile should be preserved.
**Screenshots**
Left is the original, right is when converted back.
Original file:
[CC-gray-D50-srg.png](https://github.com/kampidh/Colour-Gamut-Plotter/blob/f3bc92108e6ea7d755053eb21cfb09d30a6d6f23/jpegxl-reports/CC-gray-D50-srg.png)
Lossy JXL file:
[CC-gray-D50-srg.jxl](https://github.com/kampidh/Colour-Gamut-Plotter/blob/f3bc92108e6ea7d755053eb21cfb09d30a6d6f23/jpegxl-reports/CC-gray-D50-srg.jxl)
Converted back:
[CC-gray-D50-srg-fromjxl.png](https://github.com/kampidh/Colour-Gamut-Plotter/blob/f3bc92108e6ea7d755053eb21cfb09d30a6d6f23/jpegxl-reports/CC-gray-D50-srg-fromjxl.png)
**Environment**
- OS: Windows 10 21H2
- Compiler version: LLVM 14.0.0
- CPU type: x86_64
- cjxl/djxl version string: v0.7.0 [AVX2,SSE4,SSSE3,Unknown]
**Additional context**
N/A
<!--
Currently github does not allow uploading files that end in `.jxl`, but when you
rename them for example as `image.jxl.jpg`, it will be possible to upload them
and also view them in browsers that are configured to support it.
See https://github.com/orgs/github-community/discussions/18139
-->
| Converting lossy grayscale with custom white point resulting loss in profile | https://api.github.com/repos/libjxl/libjxl/issues/1933/comments | 5 | 2022-11-29T23:13:48Z | 2024-04-22T12:31:36Z | https://github.com/libjxl/libjxl/issues/1933 | 1,468,823,720 | 1,933 |
[
"libjxl",
"libjxl"
] | Here's a fun rabbit hole. I wanted to [request](https://gitlab.gnome.org/GNOME/gnome-backgrounds/-/issues/32) the GNOME Backgrounds (wallpapers) to switch to JPEG XL as their default universal format, to save a lot of space everywhere, because I wanted that to trickle down to Fedora and such, however... the ticket was closed as not being possible at the moment.
So when you dig around from there, you find [this](https://gitlab.gnome.org/GNOME/gnome-backgrounds/-/issues/29) (which doesn't tell us why webp was picked), [this](https://gitlab.gnome.org/GNOME/gnome-build-meta/-/issues/502) which points to [this previous attempt at adding libjxl to GNOME OS' core-deps](https://gitlab.gnome.org/GNOME/gnome-build-meta/-/merge_requests/1622)... which points right back here at issue #865.
However, reading all those, it was still a bit unclear to me what the situation was, so I asked for clarifications in `#gnome-hackers` and here are some answers I got:
> A: "It was a pain to add jpeg-xl support in GNOME OS, [as] it hard depends on skia. Someone would need to figure out how to build skia on our end"
> M: "Hard -1 from me, we should not accept a dependency on skia. Not unless skia decides to become a shared library. It's just too inconsistent with the rest of the platform."
>
> M: "It looks like libjxl is OK because it can be built with lcms instead of skcms, but the jxl gdk-pixbuf loader that you need for wallpapers really does hard depend on skcms"
>
> M: "lcms2 is already a dep of GNOME and it is a shared lib, switching to that looks like a good idea" | GDK-pixbuf loader plugin's hard dependency on SKIA / SCMS may hurt adoption in core components of Linux desktop environments and distros | https://api.github.com/repos/libjxl/libjxl/issues/1932/comments | 4 | 2022-11-29T20:35:09Z | 2023-06-13T12:31:11Z | https://github.com/libjxl/libjxl/issues/1932 | 1,468,664,890 | 1,932 |
[
"libjxl",
"libjxl"
] | Initially reported [in the GIMP bug tracker](https://gitlab.gnome.org/GNOME/gimp/-/issues/8905):
---
Open any image (ex: `~/foo.jpg`) and try to export it to "foo.jxl" while the "Lossless Mode" option is checked in the JXL export settings dialog.
It will start, get maybe 1/10th of the way through the progressbar, and then give you this generic error dialog:
> Saving '/home/user/foo.jxl' failed:
>
> JPEG XL image plug-in could not save image
Exporting without "Lossless Mode" works.
Tested with GIMP 2.10.32 from the Fedora 37 repositories. Turns out the "gimp-jxl-plugin" package was installed on my system, version 0.7.0-3.fc37. | GIMP plugin cannot export to lossless, fails with "could not save image" error | https://api.github.com/repos/libjxl/libjxl/issues/1931/comments | 3 | 2022-11-29T16:35:33Z | 2023-02-16T10:52:26Z | https://github.com/libjxl/libjxl/issues/1931 | 1,468,360,778 | 1,931 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
Consider the following GIF: [e4ec7c5d7af5342f57347c9ada429fba.gif](https://user-images.githubusercontent.com/1342577/204566989-edf60687-cce2-4f5b-a2d8-4d7e5d8579f1.gif)
**To Reproduce**
```
cjxl -d 1 -e 8 e4ec7c5d7af5342f57347c9ada429fba.gif output.jxl
```
**Expected behavior**
cjxl should produce a JXL file as it does with `-e 7` and lower.
**Actual behavior**
Using `-e 8` or `-e 9` crashes. Using `-e 7` and lower does not crash.
```
$ cjxl -e 8 -d 1 e4ec7c5d7af5342f57347c9ada429fba.gif output.jxl
JPEG XL encoder v0.8.0 759a29f9 [AVX2]
./lib/extras/dec/color_hints.cc:54: No color_space/icc_pathname given, assuming sRGB
Read 600x563 image, 3032897 bytes, 3.8 MP/s
Encoding [VarDCT, d1.000, effort: 8],
./lib/jxl/image.h:57: JXL_CHECK: xsize <= orig_xsize_
Illegal instruction (core dumped)
```
Valgrind output: [valgrind.log](https://github.com/libjxl/libjxl/files/10114328/valgrind.log)
**Environment**
- OS: Arch Linux x86_64
- Compiler version: gcc 12.2.0
- cjxl/djxl version string: JPEG XL encoder v0.8.0 759a29f9 [AVX2]
**Additional context**
Here, `-d 1` is just to force VarDCT. If you don't include that it implicitly defaults to `-d 0` and uses modular lossless. | cjxl GIF to JXL crashes using VarDCT with effort-8 or effort-9 | https://api.github.com/repos/libjxl/libjxl/issues/1929/comments | 3 | 2022-11-29T15:16:20Z | 2023-06-16T16:31:11Z | https://github.com/libjxl/libjxl/issues/1929 | 1,468,241,616 | 1,929 |
[
"libjxl",
"libjxl"
] | Weird Windows 10 message
containerfile: C:\Users\...\Downloads\libjxl-main (1).zip
file: C:\Users\...\Downloads\libjxl-main (1).zip->libjxl-main/tools/wasm_demo/index.html
webfile: C:\Users\...\Downloads\libjxl-main (1).zip|https://codeload.github.com/libjxl/libjxl/zip/refs/heads/main|pid:1920,ProcessStart:133141873410518872
| Trojan:Script/Wacatac.B!ml | https://api.github.com/repos/libjxl/libjxl/issues/1926/comments | 3 | 2022-11-29T09:26:15Z | 2022-12-02T15:37:32Z | https://github.com/libjxl/libjxl/issues/1926 | 1,467,727,550 | 1,926 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
Description and other details mostly copied from #1644.
Test suite fails to pass when compiling with `-Wp,-D_GLIBCXX_ASSERTIONS` on CXXFLAGS, showing the error `Assertion '__n < this->size()' failed.`
This flag is enabled by default on some Linux distributions build tools, [like on Arch Linux](https://github.com/archlinux/devtools/blob/20220621/makepkg-x86_64.conf#L46).
<details>
<summary>Attempting `ctest --rerun-failed --output-on-failure` gives the following (click arrow to show code):</summary>
```
Test project /home/louie/dev/libjxl/build
Start 634: DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV444CompleteInputOutputLines16BitDepth8 # GetParam() = Q85YUV444CompleteInputOutputLines16BitDepth8
1/48 Test #634: DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV444CompleteInputOutputLines16BitDepth8 # GetParam() = Q85YUV444CompleteInputOutputLines16BitDepth8 .......................Subprocess aborted***Exception: 1.38 sec
Running main() from /startdir/src/googletest-release-1.12.1/googletest/src/gtest_main.cc
Note: Google Test filter = DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV444CompleteInputOutputLines16BitDepth8
[==========] Running 1 test from 1 test suite.
[----------] Global test environment set-up.
[----------] 1 test from DecodeAPITest/DecodeAPITestParam
[ RUN ] DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV444CompleteInputOutputLines16BitDepth8
/usr/sbin/../lib64/gcc/x86_64-pc-linux-gnu/12.2.0/../../../../include/c++/12.2.0/bits/stl_vector.h:1123: std::vector::reference std::vector<unsigned char>::operator[](std::vector::size_type) [_Tp = unsigned char, _Alloc = std::allocator<unsigned char>]: Assertion '__n < this->size()' failed.
Start 635: DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV444CompleteInputOutputLines16BitDepth16 # GetParam() = Q85YUV444CompleteInputOutputLines16BitDepth16
2/48 Test #635: DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV444CompleteInputOutputLines16BitDepth16 # GetParam() = Q85YUV444CompleteInputOutputLines16BitDepth16 .....................Subprocess aborted***Exception: 1.37 sec
Running main() from /startdir/src/googletest-release-1.12.1/googletest/src/gtest_main.cc
Note: Google Test filter = DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV444CompleteInputOutputLines16BitDepth16
[==========] Running 1 test from 1 test suite.
[----------] Global test environment set-up.
[----------] 1 test from DecodeAPITest/DecodeAPITestParam
[ RUN ] DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV444CompleteInputOutputLines16BitDepth16
/usr/sbin/../lib64/gcc/x86_64-pc-linux-gnu/12.2.0/../../../../include/c++/12.2.0/bits/stl_vector.h:1123: std::vector::reference std::vector<unsigned char>::operator[](std::vector::size_type) [_Tp = unsigned char, _Alloc = std::allocator<unsigned char>]: Assertion '__n < this->size()' failed.
Start 642: DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV444InputChunks1OutputLines16BitDepth8 # GetParam() = Q85YUV444InputChunks1OutputLines16BitDepth8
3/48 Test #642: DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV444InputChunks1OutputLines16BitDepth8 # GetParam() = Q85YUV444InputChunks1OutputLines16BitDepth8 .........................Subprocess aborted***Exception: 2.61 sec
Running main() from /startdir/src/googletest-release-1.12.1/googletest/src/gtest_main.cc
Note: Google Test filter = DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV444InputChunks1OutputLines16BitDepth8
[==========] Running 1 test from 1 test suite.
[----------] Global test environment set-up.
[----------] 1 test from DecodeAPITest/DecodeAPITestParam
[ RUN ] DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV444InputChunks1OutputLines16BitDepth8
/usr/sbin/../lib64/gcc/x86_64-pc-linux-gnu/12.2.0/../../../../include/c++/12.2.0/bits/stl_vector.h:1123: std::vector::reference std::vector<unsigned char>::operator[](std::vector::size_type) [_Tp = unsigned char, _Alloc = std::allocator<unsigned char>]: Assertion '__n < this->size()' failed.
Start 643: DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV444InputChunks1OutputLines16BitDepth16 # GetParam() = Q85YUV444InputChunks1OutputLines16BitDepth16
4/48 Test #643: DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV444InputChunks1OutputLines16BitDepth16 # GetParam() = Q85YUV444InputChunks1OutputLines16BitDepth16 .......................Subprocess aborted***Exception: 2.63 sec
Running main() from /startdir/src/googletest-release-1.12.1/googletest/src/gtest_main.cc
Note: Google Test filter = DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV444InputChunks1OutputLines16BitDepth16
[==========] Running 1 test from 1 test suite.
[----------] Global test environment set-up.
[----------] 1 test from DecodeAPITest/DecodeAPITestParam
[ RUN ] DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV444InputChunks1OutputLines16BitDepth16
/usr/sbin/../lib64/gcc/x86_64-pc-linux-gnu/12.2.0/../../../../include/c++/12.2.0/bits/stl_vector.h:1123: std::vector::reference std::vector<unsigned char>::operator[](std::vector::size_type) [_Tp = unsigned char, _Alloc = std::allocator<unsigned char>]: Assertion '__n < this->size()' failed.
Start 650: DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV444InputChunks64OutputLines16BitDepth8 # GetParam() = Q85YUV444InputChunks64OutputLines16BitDepth8
5/48 Test #650: DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV444InputChunks64OutputLines16BitDepth8 # GetParam() = Q85YUV444InputChunks64OutputLines16BitDepth8 .......................Subprocess aborted***Exception: 1.37 sec
Running main() from /startdir/src/googletest-release-1.12.1/googletest/src/gtest_main.cc
Note: Google Test filter = DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV444InputChunks64OutputLines16BitDepth8
[==========] Running 1 test from 1 test suite.
[----------] Global test environment set-up.
[----------] 1 test from DecodeAPITest/DecodeAPITestParam
[ RUN ] DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV444InputChunks64OutputLines16BitDepth8
/usr/sbin/../lib64/gcc/x86_64-pc-linux-gnu/12.2.0/../../../../include/c++/12.2.0/bits/stl_vector.h:1123: std::vector::reference std::vector<unsigned char>::operator[](std::vector::size_type) [_Tp = unsigned char, _Alloc = std::allocator<unsigned char>]: Assertion '__n < this->size()' failed.
Start 651: DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV444InputChunks64OutputLines16BitDepth16 # GetParam() = Q85YUV444InputChunks64OutputLines16BitDepth16
6/48 Test #651: DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV444InputChunks64OutputLines16BitDepth16 # GetParam() = Q85YUV444InputChunks64OutputLines16BitDepth16 .....................Subprocess aborted***Exception: 1.38 sec
Running main() from /startdir/src/googletest-release-1.12.1/googletest/src/gtest_main.cc
Note: Google Test filter = DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV444InputChunks64OutputLines16BitDepth16
[==========] Running 1 test from 1 test suite.
[----------] Global test environment set-up.
[----------] 1 test from DecodeAPITest/DecodeAPITestParam
[ RUN ] DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV444InputChunks64OutputLines16BitDepth16
/usr/sbin/../lib64/gcc/x86_64-pc-linux-gnu/12.2.0/../../../../include/c++/12.2.0/bits/stl_vector.h:1123: std::vector::reference std::vector<unsigned char>::operator[](std::vector::size_type) [_Tp = unsigned char, _Alloc = std::allocator<unsigned char>]: Assertion '__n < this->size()' failed.
Start 658: DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV444InputChunks65536OutputLines16BitDepth8 # GetParam() = Q85YUV444InputChunks65536OutputLines16BitDepth8
7/48 Test #658: DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV444InputChunks65536OutputLines16BitDepth8 # GetParam() = Q85YUV444InputChunks65536OutputLines16BitDepth8 .................Subprocess aborted***Exception: 1.32 sec
Running main() from /startdir/src/googletest-release-1.12.1/googletest/src/gtest_main.cc
Note: Google Test filter = DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV444InputChunks65536OutputLines16BitDepth8
[==========] Running 1 test from 1 test suite.
[----------] Global test environment set-up.
[----------] 1 test from DecodeAPITest/DecodeAPITestParam
[ RUN ] DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV444InputChunks65536OutputLines16BitDepth8
/usr/sbin/../lib64/gcc/x86_64-pc-linux-gnu/12.2.0/../../../../include/c++/12.2.0/bits/stl_vector.h:1123: std::vector::reference std::vector<unsigned char>::operator[](std::vector::size_type) [_Tp = unsigned char, _Alloc = std::allocator<unsigned char>]: Assertion '__n < this->size()' failed.
Start 659: DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV444InputChunks65536OutputLines16BitDepth16 # GetParam() = Q85YUV444InputChunks65536OutputLines16BitDepth16
8/48 Test #659: DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV444InputChunks65536OutputLines16BitDepth16 # GetParam() = Q85YUV444InputChunks65536OutputLines16BitDepth16 ...............Subprocess aborted***Exception: 1.38 sec
Running main() from /startdir/src/googletest-release-1.12.1/googletest/src/gtest_main.cc
Note: Google Test filter = DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV444InputChunks65536OutputLines16BitDepth16
[==========] Running 1 test from 1 test suite.
[----------] Global test environment set-up.
[----------] 1 test from DecodeAPITest/DecodeAPITestParam
[ RUN ] DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV444InputChunks65536OutputLines16BitDepth16
/usr/sbin/../lib64/gcc/x86_64-pc-linux-gnu/12.2.0/../../../../include/c++/12.2.0/bits/stl_vector.h:1123: std::vector::reference std::vector<unsigned char>::operator[](std::vector::size_type) [_Tp = unsigned char, _Alloc = std::allocator<unsigned char>]: Assertion '__n < this->size()' failed.
Start 666: DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV420CompleteInputOutputLines16BitDepth8 # GetParam() = Q85YUV420CompleteInputOutputLines16BitDepth8
9/48 Test #666: DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV420CompleteInputOutputLines16BitDepth8 # GetParam() = Q85YUV420CompleteInputOutputLines16BitDepth8 .......................Subprocess aborted***Exception: 0.88 sec
Running main() from /startdir/src/googletest-release-1.12.1/googletest/src/gtest_main.cc
Note: Google Test filter = DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV420CompleteInputOutputLines16BitDepth8
[==========] Running 1 test from 1 test suite.
[----------] Global test environment set-up.
[----------] 1 test from DecodeAPITest/DecodeAPITestParam
[ RUN ] DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV420CompleteInputOutputLines16BitDepth8
/usr/sbin/../lib64/gcc/x86_64-pc-linux-gnu/12.2.0/../../../../include/c++/12.2.0/bits/stl_vector.h:1123: std::vector::reference std::vector<unsigned char>::operator[](std::vector::size_type) [_Tp = unsigned char, _Alloc = std::allocator<unsigned char>]: Assertion '__n < this->size()' failed.
Start 667: DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV420CompleteInputOutputLines16BitDepth16 # GetParam() = Q85YUV420CompleteInputOutputLines16BitDepth16
10/48 Test #667: DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV420CompleteInputOutputLines16BitDepth16 # GetParam() = Q85YUV420CompleteInputOutputLines16BitDepth16 .....................Subprocess aborted***Exception: 0.89 sec
Running main() from /startdir/src/googletest-release-1.12.1/googletest/src/gtest_main.cc
Note: Google Test filter = DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV420CompleteInputOutputLines16BitDepth16
[==========] Running 1 test from 1 test suite.
[----------] Global test environment set-up.
[----------] 1 test from DecodeAPITest/DecodeAPITestParam
[ RUN ] DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV420CompleteInputOutputLines16BitDepth16
/usr/sbin/../lib64/gcc/x86_64-pc-linux-gnu/12.2.0/../../../../include/c++/12.2.0/bits/stl_vector.h:1123: std::vector::reference std::vector<unsigned char>::operator[](std::vector::size_type) [_Tp = unsigned char, _Alloc = std::allocator<unsigned char>]: Assertion '__n < this->size()' failed.
Start 674: DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV420InputChunks1OutputLines16BitDepth8 # GetParam() = Q85YUV420InputChunks1OutputLines16BitDepth8
11/48 Test #674: DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV420InputChunks1OutputLines16BitDepth8 # GetParam() = Q85YUV420InputChunks1OutputLines16BitDepth8 .........................Subprocess aborted***Exception: 3.25 sec
Running main() from /startdir/src/googletest-release-1.12.1/googletest/src/gtest_main.cc
Note: Google Test filter = DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV420InputChunks1OutputLines16BitDepth8
[==========] Running 1 test from 1 test suite.
[----------] Global test environment set-up.
[----------] 1 test from DecodeAPITest/DecodeAPITestParam
[ RUN ] DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV420InputChunks1OutputLines16BitDepth8
/usr/sbin/../lib64/gcc/x86_64-pc-linux-gnu/12.2.0/../../../../include/c++/12.2.0/bits/stl_vector.h:1123: std::vector::reference std::vector<unsigned char>::operator[](std::vector::size_type) [_Tp = unsigned char, _Alloc = std::allocator<unsigned char>]: Assertion '__n < this->size()' failed.
Start 675: DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV420InputChunks1OutputLines16BitDepth16 # GetParam() = Q85YUV420InputChunks1OutputLines16BitDepth16
12/48 Test #675: DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV420InputChunks1OutputLines16BitDepth16 # GetParam() = Q85YUV420InputChunks1OutputLines16BitDepth16 .......................Subprocess aborted***Exception: 3.29 sec
Running main() from /startdir/src/googletest-release-1.12.1/googletest/src/gtest_main.cc
Note: Google Test filter = DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV420InputChunks1OutputLines16BitDepth16
[==========] Running 1 test from 1 test suite.
[----------] Global test environment set-up.
[----------] 1 test from DecodeAPITest/DecodeAPITestParam
[ RUN ] DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV420InputChunks1OutputLines16BitDepth16
/usr/sbin/../lib64/gcc/x86_64-pc-linux-gnu/12.2.0/../../../../include/c++/12.2.0/bits/stl_vector.h:1123: std::vector::reference std::vector<unsigned char>::operator[](std::vector::size_type) [_Tp = unsigned char, _Alloc = std::allocator<unsigned char>]: Assertion '__n < this->size()' failed.
Start 682: DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV420InputChunks64OutputLines16BitDepth8 # GetParam() = Q85YUV420InputChunks64OutputLines16BitDepth8
13/48 Test #682: DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV420InputChunks64OutputLines16BitDepth8 # GetParam() = Q85YUV420InputChunks64OutputLines16BitDepth8 .......................Subprocess aborted***Exception: 0.92 sec
Running main() from /startdir/src/googletest-release-1.12.1/googletest/src/gtest_main.cc
Note: Google Test filter = DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV420InputChunks64OutputLines16BitDepth8
[==========] Running 1 test from 1 test suite.
[----------] Global test environment set-up.
[----------] 1 test from DecodeAPITest/DecodeAPITestParam
[ RUN ] DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV420InputChunks64OutputLines16BitDepth8
/usr/sbin/../lib64/gcc/x86_64-pc-linux-gnu/12.2.0/../../../../include/c++/12.2.0/bits/stl_vector.h:1123: std::vector::reference std::vector<unsigned char>::operator[](std::vector::size_type) [_Tp = unsigned char, _Alloc = std::allocator<unsigned char>]: Assertion '__n < this->size()' failed.
Start 683: DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV420InputChunks64OutputLines16BitDepth16 # GetParam() = Q85YUV420InputChunks64OutputLines16BitDepth16
14/48 Test #683: DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV420InputChunks64OutputLines16BitDepth16 # GetParam() = Q85YUV420InputChunks64OutputLines16BitDepth16 .....................Subprocess aborted***Exception: 0.96 sec
Running main() from /startdir/src/googletest-release-1.12.1/googletest/src/gtest_main.cc
Note: Google Test filter = DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV420InputChunks64OutputLines16BitDepth16
[==========] Running 1 test from 1 test suite.
[----------] Global test environment set-up.
[----------] 1 test from DecodeAPITest/DecodeAPITestParam
[ RUN ] DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV420InputChunks64OutputLines16BitDepth16
/usr/sbin/../lib64/gcc/x86_64-pc-linux-gnu/12.2.0/../../../../include/c++/12.2.0/bits/stl_vector.h:1123: std::vector::reference std::vector<unsigned char>::operator[](std::vector::size_type) [_Tp = unsigned char, _Alloc = std::allocator<unsigned char>]: Assertion '__n < this->size()' failed.
Start 690: DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV420InputChunks65536OutputLines16BitDepth8 # GetParam() = Q85YUV420InputChunks65536OutputLines16BitDepth8
15/48 Test #690: DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV420InputChunks65536OutputLines16BitDepth8 # GetParam() = Q85YUV420InputChunks65536OutputLines16BitDepth8 .................Subprocess aborted***Exception: 0.87 sec
Running main() from /startdir/src/googletest-release-1.12.1/googletest/src/gtest_main.cc
Note: Google Test filter = DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV420InputChunks65536OutputLines16BitDepth8
[==========] Running 1 test from 1 test suite.
[----------] Global test environment set-up.
[----------] 1 test from DecodeAPITest/DecodeAPITestParam
[ RUN ] DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV420InputChunks65536OutputLines16BitDepth8
/usr/sbin/../lib64/gcc/x86_64-pc-linux-gnu/12.2.0/../../../../include/c++/12.2.0/bits/stl_vector.h:1123: std::vector::reference std::vector<unsigned char>::operator[](std::vector::size_type) [_Tp = unsigned char, _Alloc = std::allocator<unsigned char>]: Assertion '__n < this->size()' failed.
.
.
.
.
.
.
... [LINES TRUNCATED TO FIT ISSUE SIZE]
.
.
.
.
.
Start 779: DecodeAPITest/DecodeAPITestParam.TestAPI/Q85LumaSubsampleInputChunks64OutputLines16BitDepth8 # GetParam() = Q85LumaSubsampleInputChunks64OutputLines16BitDepth8
44/48 Test #779: DecodeAPITest/DecodeAPITestParam.TestAPI/Q85LumaSubsampleInputChunks64OutputLines16BitDepth8 # GetParam() = Q85LumaSubsampleInputChunks64OutputLines16BitDepth8 .........Subprocess aborted***Exception: 1.06 sec
Running main() from /startdir/src/googletest-release-1.12.1/googletest/src/gtest_main.cc
Note: Google Test filter = DecodeAPITest/DecodeAPITestParam.TestAPI/Q85LumaSubsampleInputChunks64OutputLines16BitDepth8
[==========] Running 1 test from 1 test suite.
[----------] Global test environment set-up.
[----------] 1 test from DecodeAPITest/DecodeAPITestParam
[ RUN ] DecodeAPITest/DecodeAPITestParam.TestAPI/Q85LumaSubsampleInputChunks64OutputLines16BitDepth8
/usr/sbin/../lib64/gcc/x86_64-pc-linux-gnu/12.2.0/../../../../include/c++/12.2.0/bits/stl_vector.h:1123: std::vector::reference std::vector<unsigned char>::operator[](std::vector::size_type) [_Tp = unsigned char, _Alloc = std::allocator<unsigned char>]: Assertion '__n < this->size()' failed.
Start 781: DecodeAPITest/DecodeAPITestParam.TestAPI/Q85RGBCompleteInputOutputLines16BitDepth8 # GetParam() = Q85RGBCompleteInputOutputLines16BitDepth8
45/48 Test #781: DecodeAPITest/DecodeAPITestParam.TestAPI/Q85RGBCompleteInputOutputLines16BitDepth8 # GetParam() = Q85RGBCompleteInputOutputLines16BitDepth8 .............................Subprocess aborted***Exception: 1.36 sec
Running main() from /startdir/src/googletest-release-1.12.1/googletest/src/gtest_main.cc
Note: Google Test filter = DecodeAPITest/DecodeAPITestParam.TestAPI/Q85RGBCompleteInputOutputLines16BitDepth8
[==========] Running 1 test from 1 test suite.
[----------] Global test environment set-up.
[----------] 1 test from DecodeAPITest/DecodeAPITestParam
[ RUN ] DecodeAPITest/DecodeAPITestParam.TestAPI/Q85RGBCompleteInputOutputLines16BitDepth8
/usr/sbin/../lib64/gcc/x86_64-pc-linux-gnu/12.2.0/../../../../include/c++/12.2.0/bits/stl_vector.h:1123: std::vector::reference std::vector<unsigned char>::operator[](std::vector::size_type) [_Tp = unsigned char, _Alloc = std::allocator<unsigned char>]: Assertion '__n < this->size()' failed.
Start 783: DecodeAPITest/DecodeAPITestParam.TestAPI/Q85RGBInputChunks64OutputLines16BitDepth8 # GetParam() = Q85RGBInputChunks64OutputLines16BitDepth8
46/48 Test #783: DecodeAPITest/DecodeAPITestParam.TestAPI/Q85RGBInputChunks64OutputLines16BitDepth8 # GetParam() = Q85RGBInputChunks64OutputLines16BitDepth8 .............................Subprocess aborted***Exception: 1.45 sec
Running main() from /startdir/src/googletest-release-1.12.1/googletest/src/gtest_main.cc
Note: Google Test filter = DecodeAPITest/DecodeAPITestParam.TestAPI/Q85RGBInputChunks64OutputLines16BitDepth8
[==========] Running 1 test from 1 test suite.
[----------] Global test environment set-up.
[----------] 1 test from DecodeAPITest/DecodeAPITestParam
[ RUN ] DecodeAPITest/DecodeAPITestParam.TestAPI/Q85RGBInputChunks64OutputLines16BitDepth8
/usr/sbin/../lib64/gcc/x86_64-pc-linux-gnu/12.2.0/../../../../include/c++/12.2.0/bits/stl_vector.h:1123: std::vector::reference std::vector<unsigned char>::operator[](std::vector::size_type) [_Tp = unsigned char, _Alloc = std::allocator<unsigned char>]: Assertion '__n < this->size()' failed.
Start 785: DecodeAPITest/DecodeAPITestParam.TestAPI/Q85RGBSubsampleBlueCompleteInputOutputLines16BitDepth8 # GetParam() = Q85RGBSubsampleBlueCompleteInputOutputLines16BitDepth8
47/48 Test #785: DecodeAPITest/DecodeAPITestParam.TestAPI/Q85RGBSubsampleBlueCompleteInputOutputLines16BitDepth8 # GetParam() = Q85RGBSubsampleBlueCompleteInputOutputLines16BitDepth8 ...Subprocess aborted***Exception: 1.08 sec
Running main() from /startdir/src/googletest-release-1.12.1/googletest/src/gtest_main.cc
Note: Google Test filter = DecodeAPITest/DecodeAPITestParam.TestAPI/Q85RGBSubsampleBlueCompleteInputOutputLines16BitDepth8
[==========] Running 1 test from 1 test suite.
[----------] Global test environment set-up.
[----------] 1 test from DecodeAPITest/DecodeAPITestParam
[ RUN ] DecodeAPITest/DecodeAPITestParam.TestAPI/Q85RGBSubsampleBlueCompleteInputOutputLines16BitDepth8
/usr/sbin/../lib64/gcc/x86_64-pc-linux-gnu/12.2.0/../../../../include/c++/12.2.0/bits/stl_vector.h:1123: std::vector::reference std::vector<unsigned char>::operator[](std::vector::size_type) [_Tp = unsigned char, _Alloc = std::allocator<unsigned char>]: Assertion '__n < this->size()' failed.
Start 787: DecodeAPITest/DecodeAPITestParam.TestAPI/Q85RGBSubsampleBlueInputChunks64OutputLines16BitDepth8 # GetParam() = Q85RGBSubsampleBlueInputChunks64OutputLines16BitDepth8
48/48 Test #787: DecodeAPITest/DecodeAPITestParam.TestAPI/Q85RGBSubsampleBlueInputChunks64OutputLines16BitDepth8 # GetParam() = Q85RGBSubsampleBlueInputChunks64OutputLines16BitDepth8 ...Subprocess aborted***Exception: 1.24 sec
Running main() from /startdir/src/googletest-release-1.12.1/googletest/src/gtest_main.cc
Note: Google Test filter = DecodeAPITest/DecodeAPITestParam.TestAPI/Q85RGBSubsampleBlueInputChunks64OutputLines16BitDepth8
[==========] Running 1 test from 1 test suite.
[----------] Global test environment set-up.
[----------] 1 test from DecodeAPITest/DecodeAPITestParam
[ RUN ] DecodeAPITest/DecodeAPITestParam.TestAPI/Q85RGBSubsampleBlueInputChunks64OutputLines16BitDepth8
/usr/sbin/../lib64/gcc/x86_64-pc-linux-gnu/12.2.0/../../../../include/c++/12.2.0/bits/stl_vector.h:1123: std::vector::reference std::vector<unsigned char>::operator[](std::vector::size_type) [_Tp = unsigned char, _Alloc = std::allocator<unsigned char>]: Assertion '__n < this->size()' failed.
0% tests passed, 48 tests failed out of 48
Total Test time (real) = 64.20 sec
The following tests FAILED:
634 - DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV444CompleteInputOutputLines16BitDepth8 # GetParam() = Q85YUV444CompleteInputOutputLines16BitDepth8 (Subprocess aborted)
635 - DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV444CompleteInputOutputLines16BitDepth16 # GetParam() = Q85YUV444CompleteInputOutputLines16BitDepth16 (Subprocess aborted)
642 - DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV444InputChunks1OutputLines16BitDepth8 # GetParam() = Q85YUV444InputChunks1OutputLines16BitDepth8 (Subprocess aborted)
643 - DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV444InputChunks1OutputLines16BitDepth16 # GetParam() = Q85YUV444InputChunks1OutputLines16BitDepth16 (Subprocess aborted)
650 - DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV444InputChunks64OutputLines16BitDepth8 # GetParam() = Q85YUV444InputChunks64OutputLines16BitDepth8 (Subprocess aborted)
651 - DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV444InputChunks64OutputLines16BitDepth16 # GetParam() = Q85YUV444InputChunks64OutputLines16BitDepth16 (Subprocess aborted)
658 - DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV444InputChunks65536OutputLines16BitDepth8 # GetParam() = Q85YUV444InputChunks65536OutputLines16BitDepth8 (Subprocess aborted)
659 - DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV444InputChunks65536OutputLines16BitDepth16 # GetParam() = Q85YUV444InputChunks65536OutputLines16BitDepth16 (Subprocess aborted)
666 - DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV420CompleteInputOutputLines16BitDepth8 # GetParam() = Q85YUV420CompleteInputOutputLines16BitDepth8 (Subprocess aborted)
667 - DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV420CompleteInputOutputLines16BitDepth16 # GetParam() = Q85YUV420CompleteInputOutputLines16BitDepth16 (Subprocess aborted)
674 - DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV420InputChunks1OutputLines16BitDepth8 # GetParam() = Q85YUV420InputChunks1OutputLines16BitDepth8 (Subprocess aborted)
675 - DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV420InputChunks1OutputLines16BitDepth16 # GetParam() = Q85YUV420InputChunks1OutputLines16BitDepth16 (Subprocess aborted)
682 - DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV420InputChunks64OutputLines16BitDepth8 # GetParam() = Q85YUV420InputChunks64OutputLines16BitDepth8 (Subprocess aborted)
683 - DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV420InputChunks64OutputLines16BitDepth16 # GetParam() = Q85YUV420InputChunks64OutputLines16BitDepth16 (Subprocess aborted)
690 - DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV420InputChunks65536OutputLines16BitDepth8 # GetParam() = Q85YUV420InputChunks65536OutputLines16BitDepth8 (Subprocess aborted)
691 - DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV420InputChunks65536OutputLines16BitDepth16 # GetParam() = Q85YUV420InputChunks65536OutputLines16BitDepth16 (Subprocess aborted)
698 - DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV420PROGRCompleteInputOutputLines16BitDepth8 # GetParam() = Q85YUV420PROGRCompleteInputOutputLines16BitDepth8 (Subprocess aborted)
699 - DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV420PROGRCompleteInputOutputLines16BitDepth16 # GetParam() = Q85YUV420PROGRCompleteInputOutputLines16BitDepth16 (Subprocess aborted)
706 - DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV420PROGRInputChunks1OutputLines16BitDepth8 # GetParam() = Q85YUV420PROGRInputChunks1OutputLines16BitDepth8 (Subprocess aborted)
707 - DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV420PROGRInputChunks1OutputLines16BitDepth16 # GetParam() = Q85YUV420PROGRInputChunks1OutputLines16BitDepth16 (Subprocess aborted)
714 - DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV420PROGRInputChunks64OutputLines16BitDepth8 # GetParam() = Q85YUV420PROGRInputChunks64OutputLines16BitDepth8 (Subprocess aborted)
715 - DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV420PROGRInputChunks64OutputLines16BitDepth16 # GetParam() = Q85YUV420PROGRInputChunks64OutputLines16BitDepth16 (Subprocess aborted)
722 - DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV420PROGRInputChunks65536OutputLines16BitDepth8 # GetParam() = Q85YUV420PROGRInputChunks65536OutputLines16BitDepth8 (Subprocess aborted)
723 - DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV420PROGRInputChunks65536OutputLines16BitDepth16 # GetParam() = Q85YUV420PROGRInputChunks65536OutputLines16BitDepth16 (Subprocess aborted)
730 - DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV420R13BCompleteInputOutputLines16BitDepth8 # GetParam() = Q85YUV420R13BCompleteInputOutputLines16BitDepth8 (Subprocess aborted)
731 - DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV420R13BCompleteInputOutputLines16BitDepth16 # GetParam() = Q85YUV420R13BCompleteInputOutputLines16BitDepth16 (Subprocess aborted)
738 - DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV420R13BInputChunks1OutputLines16BitDepth8 # GetParam() = Q85YUV420R13BInputChunks1OutputLines16BitDepth8 (Subprocess aborted)
739 - DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV420R13BInputChunks1OutputLines16BitDepth16 # GetParam() = Q85YUV420R13BInputChunks1OutputLines16BitDepth16 (Subprocess aborted)
746 - DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV420R13BInputChunks64OutputLines16BitDepth8 # GetParam() = Q85YUV420R13BInputChunks64OutputLines16BitDepth8 (Subprocess aborted)
747 - DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV420R13BInputChunks64OutputLines16BitDepth16 # GetParam() = Q85YUV420R13BInputChunks64OutputLines16BitDepth16 (Subprocess aborted)
754 - DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV420R13BInputChunks65536OutputLines16BitDepth8 # GetParam() = Q85YUV420R13BInputChunks65536OutputLines16BitDepth8 (Subprocess aborted)
755 - DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV420R13BInputChunks65536OutputLines16BitDepth16 # GetParam() = Q85YUV420R13BInputChunks65536OutputLines16BitDepth16 (Subprocess aborted)
757 - DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV422CompleteInputOutputLines16BitDepth8 # GetParam() = Q85YUV422CompleteInputOutputLines16BitDepth8 (Subprocess aborted)
759 - DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV422InputChunks64OutputLines16BitDepth8 # GetParam() = Q85YUV422InputChunks64OutputLines16BitDepth8 (Subprocess aborted)
761 - DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV440CompleteInputOutputLines16BitDepth8 # GetParam() = Q85YUV440CompleteInputOutputLines16BitDepth8 (Subprocess aborted)
763 - DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV440InputChunks64OutputLines16BitDepth8 # GetParam() = Q85YUV440InputChunks64OutputLines16BitDepth8 (Subprocess aborted)
765 - DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV444_1x2CompleteInputOutputLines16BitDepth8 # GetParam() = Q85YUV444_1x2CompleteInputOutputLines16BitDepth8 (Subprocess aborted)
767 - DecodeAPITest/DecodeAPITestParam.TestAPI/Q85YUV444_1x2InputChunks64OutputLines16BitDepth8 # GetParam() = Q85YUV444_1x2InputChunks64OutputLines16BitDepth8 (Subprocess aborted)
769 - DecodeAPITest/DecodeAPITestParam.TestAPI/Q85AsymmetricCompleteInputOutputLines16BitDepth8 # GetParam() = Q85AsymmetricCompleteInputOutputLines16BitDepth8 (Subprocess aborted)
771 - DecodeAPITest/DecodeAPITestParam.TestAPI/Q85AsymmetricInputChunks64OutputLines16BitDepth8 # GetParam() = Q85AsymmetricInputChunks64OutputLines16BitDepth8 (Subprocess aborted)
773 - DecodeAPITest/DecodeAPITestParam.TestAPI/Q85GrayCompleteInputOutputLines16BitDepth8 # GetParam() = Q85GrayCompleteInputOutputLines16BitDepth8 (Subprocess aborted)
775 - DecodeAPITest/DecodeAPITestParam.TestAPI/Q85GrayInputChunks64OutputLines16BitDepth8 # GetParam() = Q85GrayInputChunks64OutputLines16BitDepth8 (Subprocess aborted)
777 - DecodeAPITest/DecodeAPITestParam.TestAPI/Q85LumaSubsampleCompleteInputOutputLines16BitDepth8 # GetParam() = Q85LumaSubsampleCompleteInputOutputLines16BitDepth8 (Subprocess aborted)
779 - DecodeAPITest/DecodeAPITestParam.TestAPI/Q85LumaSubsampleInputChunks64OutputLines16BitDepth8 # GetParam() = Q85LumaSubsampleInputChunks64OutputLines16BitDepth8 (Subprocess aborted)
781 - DecodeAPITest/DecodeAPITestParam.TestAPI/Q85RGBCompleteInputOutputLines16BitDepth8 # GetParam() = Q85RGBCompleteInputOutputLines16BitDepth8 (Subprocess aborted)
783 - DecodeAPITest/DecodeAPITestParam.TestAPI/Q85RGBInputChunks64OutputLines16BitDepth8 # GetParam() = Q85RGBInputChunks64OutputLines16BitDepth8 (Subprocess aborted)
785 - DecodeAPITest/DecodeAPITestParam.TestAPI/Q85RGBSubsampleBlueCompleteInputOutputLines16BitDepth8 # GetParam() = Q85RGBSubsampleBlueCompleteInputOutputLines16BitDepth8 (Subprocess aborted)
787 - DecodeAPITest/DecodeAPITestParam.TestAPI/Q85RGBSubsampleBlueInputChunks64OutputLines16BitDepth8 # GetParam() = Q85RGBSubsampleBlueInputChunks64OutputLines16BitDepth8 (Subprocess aborted)
```
</details>
**To Reproduce**
```
$ export CXXFLAGS+=' -Wp,-D_GLIBCXX_ASSERTIONS'
$ <your_preferred_cmake_command>
$ make -C build all
$ make -C build test
```
**Expected behavior**
The test suite should pass.
**Screenshots**
N/A (build error)
**Environment**
- OS: Arch Linux under WSL2
- Compiler version: clang 14.0.6 and gcc 12.2.0 (same error on both)
- CPU type: x86_64
- cjxl/djxl version string: v0.8.0 at commit c8f1713c05e713d6f8ae2c60e477483927db5a59
**Additional context**
The test suite passes fine when adding `-Wp,-U_GLIBCXX_ASSERTIONS` at the end of CXXFLAGS to remove the definition of the `_GLIBCXX_ASSERTIONS` macro.
I have not yet attempted to search for a known-good commit.
<!--
Currently github does not allow uploading files that end in `.jxl`, but when you
rename them for example as `image.jxl.jpg`, it will be possible to upload them
and also view them in browsers that are configured to support it.
See https://github.com/orgs/github-community/discussions/18139
--> | Test suite is failing again when compiling with `-Wp,-D_GLIBCXX_ASSERTIONS` | https://api.github.com/repos/libjxl/libjxl/issues/1921/comments | 3 | 2022-11-27T17:50:07Z | 2024-03-04T10:12:56Z | https://github.com/libjxl/libjxl/issues/1921 | 1,465,525,645 | 1,921 |
[
"libjxl",
"libjxl"
] | [Pale Moon](https://www.palemoon,org), as of 31.4.0, default-supports (and enables) JPEG-XL. It offers conneg headers to negotiate dynamic image serving by servers that have .jxl images available.
It should be added to the list of supporting software.
Google has also announced it intends to [drop support for it](https://www.cnet.com/tech/computing/chrome-banishes-jpeg-xl-photo-format-that-could-save-phone-space/), so perhaps it should be removed. | Add Pale Moon to the list of jxl-supporting software (and maybe remove Chrome?) | https://api.github.com/repos/libjxl/libjxl/issues/1914/comments | 5 | 2022-11-23T11:37:02Z | 2022-11-28T10:56:59Z | https://github.com/libjxl/libjxl/issues/1914 | 1,461,594,404 | 1,914 |
[
"libjxl",
"libjxl"
] | When using cjxl and converting to jxl (from jpg and png in my test cases) the exif metadata of the pictures does't get passed through to the created jxl image by default.
**To Reproduce**
Convert an image using cjxl in the command line.
**Expected behavior**
Exif metadata gets passed through to the newly created image by default.
**Environment**
- OS: Archlinux
- CPU type: x86_64
- cjxl/djxl version: 0.7.0
<!--
Currently github does not allow uploading files that end in `.jxl`, but when you
rename them for example as `image.jxl.jpg`, it will be possible to upload them
and also view them in browsers that are configured to support it.
See https://github.com/orgs/github-community/discussions/18139
-->
| No exif passthrough by default | https://api.github.com/repos/libjxl/libjxl/issues/1911/comments | 1 | 2022-11-22T12:37:28Z | 2022-11-22T13:34:13Z | https://github.com/libjxl/libjxl/issues/1911 | 1,459,792,437 | 1,911 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
Running main() from ./googletest/src/gtest_main.cc
Note: Google Test filter = ColorManagementTestInstantiation/ColorManagementTest.VerifyAllProfiles/0
[==========] Running 1 test from 1 test suite.
[----------] Global test environment set-up.
[----------] 1 test from ColorManagementTestInstantiation/ColorManagementTest
[ RUN ] ColorManagementTestInstantiation/ColorManagementTest.VerifyAllProfiles/0
RGB_D65_SRG_Per_709
./lib/jxl/color_management_test.cc:121: Failure
Expected equality of these values:
c.rendering_intent
Which is: 4-byte object <00-00 00-00>
c2.rendering_intent
Which is: 4-byte object <01-00 00-00>
[ FAILED ] ColorManagementTestInstantiation/ColorManagementTest.VerifyAllProfiles/0, where GetParam() = ColorEncoding/RGB_D65_SRG_Per_709 (3 ms)
[----------] 1 test from ColorManagementTestInstantiation/ColorManagementTest (3 ms total)
[----------] Global test environment tear-down
[==========] 1 test from 1 test suite ran. (3 ms total)
[ PASSED ] 0 tests.
[ FAILED ] 1 test, listed below:
[ FAILED ] ColorManagementTestInstantiation/ColorManagementTest.VerifyAllProfiles/0, where GetParam() = ColorEncoding/RGB_D65_SRG_Per_709
1 FAILED TEST
**To Reproduce**
$ ctest -R RGB_D65_SRG_Per_709 -V
**Expected behavior**
Test should run without error
**Screenshots**
* https://buildd.debian.org/status/fetch.php?pkg=jpeg-xl&arch=ppc64el&ver=0.8.0%7Egit20221120.4167056-1&stamp=1669093955&raw=0
**Environment**
- OS: Linux
- Compiler version: gcc-12
- CPU type: all arches
- cjxl/djxl version string: git20221120.4167056
| ColorManagementTestInstantiation/ColorManagementTest.VerifyAllProfiles/ColorEncoding/RGB_D65_SRG_Per_709 (Failed) | https://api.github.com/repos/libjxl/libjxl/issues/1910/comments | 4 | 2022-11-22T07:09:24Z | 2024-03-04T10:12:55Z | https://github.com/libjxl/libjxl/issues/1910 | 1,459,264,975 | 1,910 |
[
"libjxl",
"libjxl"
] | **Is your feature request related to a problem? Please describe.**
When decoding JPEG-XL files to jpeg inside script there is no option to know if image can be reconstructed to jpeg losslessly without saving the output, and then discarding if we want only decode losslessly.
**Describe the solution you'd like**
Add a option to djxl to not to retry with `--pixels_to_jpeg`
**Describe alternatives you've considered**
Some other option to check if jxl was created as losslessly compressed jpeg, without decoding it to file | Option to not retry with `--pixels_to_jpeg` for `djxl` | https://api.github.com/repos/libjxl/libjxl/issues/1907/comments | 2 | 2022-11-19T13:45:07Z | 2023-01-16T16:53:00Z | https://github.com/libjxl/libjxl/issues/1907 | 1,456,483,252 | 1,907 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
Using a 0.7 build of the wasm file, a number of images fail with `JxlDecoderSetPreferredColorProfile failed` when HDR is requested (SDR works).
**To Reproduce**
Compile wasm from version 0.7 and attempt to decode various images.
**Expected behavior**
The error message should not occur and the image decode successfully.
**Environment**
- OS: Linux
- Compiler version: clang 15
- CPU type: x86_64
- cjxl/djxl version string: cjxl v0.7 | SIMD supported: SSE4,Scalar
| 0.7 wasm fails with in JxlDecoderSetPreferredColorProfile | https://api.github.com/repos/libjxl/libjxl/issues/1902/comments | 2 | 2022-11-17T02:46:37Z | 2022-11-17T07:59:46Z | https://github.com/libjxl/libjxl/issues/1902 | 1,452,604,692 | 1,902 |
[
"libjxl",
"libjxl"
] | As someone, who has scanned a lot of photos and would like to keep all metadata associated with them, I would like to directly convert my photos from the TIFF format to JXL. Trying to convert to an intermediate format, such as PNG, is possible with another utility, such as ImageMagick, but it adds significantly more time and complexity to the entire process.
For multi-TIFF images, it would be nice to have the option to split all images into separate JXL files.
| Add Support For TIFF Image File Format | https://api.github.com/repos/libjxl/libjxl/issues/1901/comments | 4 | 2022-11-17T00:34:32Z | 2025-04-01T09:17:27Z | https://github.com/libjxl/libjxl/issues/1901 | 1,452,498,731 | 1,901 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
In [decode.cc, line 2695](https://github.com/libjxl/libjxl/blob/1d2225c252ce5fd906d3a12ee0dbed92e577076d/lib/jxl/decode.cc#L2695), the value returned from `ConvertExternalToInternalColorEncoding` is `true` (for success). However, the boolean is compared to `JXL_DEC_SUCCESS` which is a zero `0` (`JXL_DEC_FAILURE` is one `1`). So it is thinking there is an error when there is not.
**To Reproduce**
Call `JxlDecoderSetPreferredColorProfile` with valid values, such as from [tools/jxl_emcc.cc, line 188](https://github.com/libjxl/libjxl/blob/cb03f7124c7f8bf90fed392ffca229b787726d7b/tools/jxl_emcc.cc#L188).
**Expected behavior**
The decoding should have completed.
**Environment**
- OS: Any
- Compiler version: emscripten 3.1.24 & clang 15.0.2
- CPU type: x86_64
- cjxl/djxl version string: cjxl [v0.7 | SIMD supported: SSE4,Scalar]
**Additional context**
Causing WASM to fail.
Could the values of `JXL_DEC_FAILURE` and `JXL_DEC_SUCCESS` be flipped so as not to interfere when checked against outside functions returning boolean `true` for success and `false` for error?
| Throwing error when there is none, values seem flipped. | https://api.github.com/repos/libjxl/libjxl/issues/1896/comments | 1 | 2022-11-15T16:39:50Z | 2022-11-15T23:55:10Z | https://github.com/libjxl/libjxl/issues/1896 | 1,450,073,846 | 1,896 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
Build fails on openSUSE Tumbleweed armv7, with libhwy 1.0.2
```
[ 40s] cd /home/abuild/rpmbuild/BUILD/libjxl-0.7.0/build/lib && /usr/bin/c++ -DHWY_DISABLED_TARGETS="(HWY_SVE|HWY_SVE2|HWY_SVE_256|HWY_SVE2_128|HWY_RVV)" -DJPEGXL_MAJOR_VERSION=0 -DJPEGXL_MINOR_VERSION=7 -DJPEGXL_PATCH_VERSION=0 -DJXL_INTERNAL_LIBRARY_BUILD -D__DATE__=\"redacted\" -D__TIMESTAMP__=\"redacted\" -D__TIME__=\"redacted\" -I/home/abuild/rpmbuild/BUILD/libjxl-0.7.0 -I/home/abuild/rpmbuild/BUILD/libjxl-0.7.0/lib/include -I/home/abuild/rpmbuild/BUILD/libjxl-0.7.0/build/lib/include -O2 -Wall -U_FORTIFY_SOURCE -D_FORTIFY_SOURCE=3 -fstack-protector-strong -funwind-tables -fasynchronous-unwind-tables -fstack-clash-protection -Werror=return-type -flto=auto -g -fno-rtti -funwind-tables -fno-omit-frame-pointer -O2 -g -DNDEBUG -fPIC -fvisibility=hidden -fvisibility-inlines-hidden -fmacro-prefix-map=/home/abuild/rpmbuild/BUILD/libjxl-0.7.0=. -Wno-builtin-macro-redefined -Wall -fmerge-all-constants -fno-builtin-fwrite -fno-builtin-fread -Wextra -Wc++11-compat -Warray-bounds -Wformat-security -Wimplicit-fallthrough -Wno-register -Wno-unused-function -Wno-unused-parameter -Wnon-virtual-dtor -Woverloaded-virtual -Wvla -fsized-deallocation -fno-exceptions -fmath-errno -std=c++11 -MD -MT lib/CMakeFiles/jxl_dec-obj.dir/jxl/compressed_dc.cc.o -MF CMakeFiles/jxl_dec-obj.dir/jxl/compressed_dc.cc.o.d -o CMakeFiles/jxl_dec-obj.dir/jxl/compressed_dc.cc.o -c /home/abuild/rpmbuild/BUILD/libjxl-0.7.0/lib/jxl/compressed_dc.cc
[ 42s] In file included from /usr/include/hwy/ops/arm_neon-inl.h:33,
[ 42s] from /usr/include/hwy/highway.h:358,
[ 42s] from /home/abuild/rpmbuild/BUILD/libjxl-0.7.0/lib/jxl/compressed_dc.cc:22,
[ 42s] from /usr/include/hwy/foreach_target.h:81,
[ 42s] from /home/abuild/rpmbuild/BUILD/libjxl-0.7.0/lib/jxl/compressed_dc.cc:21:
[ 42s] /usr/lib/gcc/armv7hl-suse-linux-gnueabi/12/include/arm_neon.h: In function 'hwy::N_NEON::StoreU(hwy::N_NEON::Vec128<float, 4u>, hwy::N_NEON::Simd<float, 4u, 0>, float*)':
[ 42s] /usr/lib/gcc/armv7hl-suse-linux-gnueabi/12/include/arm_neon.h:11024:1: error: inlining failed in call to 'always_inline' 'vst1q_f32': target specific option mismatch
[ 42s] 11024 | vst1q_f32 (float32_t * __a, float32x4_t __b)
[ 42s] | ^~~~~~~~~
[ 42s] /usr/include/hwy/ops/arm_neon-inl.h:2777:12: note: called from here
[ 42s] 2777 | vst1q_f32(unaligned, v.raw);
[ 42s] | ~~~~~~~~~^~~~~~~~~~~~~~~~~~
[ 42s] make[2]: *** [lib/CMakeFiles/jxl_dec-obj.dir/build.make:275: lib/CMakeFiles/jxl_dec-obj.dir/jxl/compressed_dc.cc.o] Error 1
```
This looks similar to what can be seen here: https://github.com/google/highway/issues/1032
**To Reproduce**
Try to build on openSUSE Tumbleweed for armv7 with:
```
[ 14s] + /usr/bin/cmake /home/abuild/rpmbuild/BUILD/libjxl-0.7.0/. '-GUnix Makefiles' -DCMAKE_INSTALL_PREFIX:PATH=/usr -DINCLUDE_INSTALL_DIR:PATH=/usr/include -DLIB_INSTALL_DIR:PATH=/usr/lib -DSYSCONF_INSTALL_DIR:PATH=/etc -DSHARE_INSTALL_PREFIX:PATH=/usr/share -DCMAKE_INSTALL_LIBDIR:PATH=lib -DCMAKE_INSTALL_LIBEXECDIR=/usr/libexec -DCMAKE_BUILD_TYPE=RelWithDebInfo '-DCMAKE_C_FLAGS=-O2 -Wall -U_FORTIFY_SOURCE -D_FORTIFY_SOURCE=3 -fstack-protector-strong -funwind-tables -fasynchronous-unwind-tables -fstack-clash-protection -Werror=return-type -flto=auto -g' '-DCMAKE_CXX_FLAGS=-O2 -Wall -U_FORTIFY_SOURCE -D_FORTIFY_SOURCE=3 -fstack-protector-strong -funwind-tables -fasynchronous-unwind-tables -fstack-clash-protection -Werror=return-type -flto=auto -g' '-DCMAKE_Fortran_FLAGS=-O2 -Wall -U_FORTIFY_SOURCE -D_FORTIFY_SOURCE=3 -fstack-protector-strong -funwind-tables -fasynchronous-unwind-tables -fstack-clash-protection -Werror=return-type -flto=auto -g' '-DCMAKE_EXE_LINKER_FLAGS=-flto=auto -Wl,--as-needed -Wl,--no-undefined -Wl,-z,now' '-DCMAKE_MODULE_LINKER_FLAGS=-flto=auto -Wl,--as-needed' '-DCMAKE_SHARED_LINKER_FLAGS=-flto=auto -Wl,--as-needed -Wl,--no-undefined -Wl,-z,now' -DCMAKE_VERBOSE_MAKEFILE:BOOL=ON -DBUILD_SHARED_LIBS:BOOL=ON -DBUILD_STATIC_LIBS:BOOL=OFF -DCMAKE_COLOR_MAKEFILE:BOOL=OFF -DCMAKE_INSTALL_DO_STRIP:BOOL=OFF -DCMAKE_MODULES_INSTALL_DIR=/usr/lib/cmake/libjxl -DJPEGXL_FORCE_SYSTEM_HWY=ON -DJPEGXL_FORCE_SYSTEM_BROTLI=ON -DJPEGXL_FORCE_SYSTEM_LCMS2=ON -DBUILD_TESTING=OFF -DJPEGXL_ENABLE_PLUGINS=OFF -DJPEGXL_ENABLE_SKCMS=OFF -DJPEGXL_ENABLE_SJPEG=OFF
[ 14s] -- The C compiler identification is GNU 12.2.1
[ 15s] -- The CXX compiler identification is GNU 12.2.1
[ 15s] -- Detecting C compiler ABI info
[ 15s] -- Detecting C compiler ABI info - done
[ 15s] -- Check for working C compiler: /usr/bin/cc - skipped
[ 15s] -- Detecting C compile features
[ 15s] -- Detecting C compile features - done
[ 15s] -- Detecting CXX compiler ABI info
[ 15s] -- Detecting CXX compiler ABI info - done
[ 15s] -- Check for working CXX compiler: /usr/bin/c++ - skipped
[ 15s] -- Detecting CXX compile features
[ 15s] -- Detecting CXX compile features - done
[ 15s] -- Performing Test JPEGXL_EMSCRIPTEN
[ 15s] -- Performing Test JPEGXL_EMSCRIPTEN - Failed
[ 15s] -- CMAKE_SYSTEM_PROCESSOR is armv7l
[ 15s] -- Performing Test CXX_FUZZERS_SUPPORTED
[ 15s] -- Performing Test CXX_FUZZERS_SUPPORTED - Failed
[ 15s] -- Performing Test CXX_CONSTRUCTOR_ALIASES_SUPPORTED
[ 15s] -- Performing Test CXX_CONSTRUCTOR_ALIASES_SUPPORTED - Failed
[ 15s] -- Performing Test CXX_MACRO_PREFIX_MAP
[ 16s] -- Performing Test CXX_MACRO_PREFIX_MAP - Success
[ 16s] -- Performing Test CXX_NO_RTTI_SUPPORTED
[ 16s] -- Performing Test CXX_NO_RTTI_SUPPORTED - Success
[ 17s] -- Found PkgConfig: /usr/bin/pkg-config (found version "1.8.0")
[ 17s] -- Performing Test JXL_HWY_DISABLED_TARGETS_FORCED
[ 17s] -- Performing Test JXL_HWY_DISABLED_TARGETS_FORCED - Failed
[ 17s] -- Compiled IDs C:GNU, C++:GNU
[ 17s] -- Performing Test CMAKE_HAVE_LIBC_PTHREAD
[ 17s] -- Performing Test CMAKE_HAVE_LIBC_PTHREAD - Success
[ 17s] -- Found Threads: TRUE
[ 17s] -- Performing Test ATOMICS_LOCK_FREE_INSTRUCTIONS
[ 18s] -- Performing Test ATOMICS_LOCK_FREE_INSTRUCTIONS - Success
[ 18s] Warning: HWY_SVE, HWY_SVE2, HWY_SVE_256, HWY_SVE2_128 and HWY_RVV CPU targets are disabled
[ 18s] -- Found HWY: /usr/lib/libhwy.so (found suitable version "1.0.2", minimum required is "0.15.0")
[ 18s] -- Checking for module 'libbrotlicommon'
[ 18s] -- Found libbrotlicommon, version 1.0.9
[ 18s] -- Checking for module 'libbrotlienc'
[ 18s] -- Found libbrotlienc, version 1.0.9
[ 18s] -- Checking for module 'libbrotlidec'
[ 18s] -- Found libbrotlidec, version 1.0.9
[ 18s] -- Found Brotli: /usr/include
[ 18s] -- Found LCMS2: /usr/lib/liblcms2.so (found suitable version "2.14", minimum required is "2.13")
[ 18s] -- Could NOT find ZLIB (missing: ZLIB_LIBRARY ZLIB_INCLUDE_DIR)
[ 18s] -- Could NOT find PNG (missing: PNG_LIBRARY PNG_PNG_INCLUDE_DIR)
[ 18s] -- Performing Test COMPILER_HAS_HIDDEN_VISIBILITY
[ 19s] -- Performing Test COMPILER_HAS_HIDDEN_VISIBILITY - Success
[ 19s] -- Performing Test COMPILER_HAS_HIDDEN_INLINE_VISIBILITY
[ 19s] -- Performing Test COMPILER_HAS_HIDDEN_INLINE_VISIBILITY - Success
[ 19s] -- Performing Test COMPILER_HAS_DEPRECATED_ATTR
[ 19s] -- Performing Test COMPILER_HAS_DEPRECATED_ATTR - Success
[ 19s] -- Performing Test LINKER_SUPPORT_EXCLUDE_LIBS
[ 20s] -- Performing Test LINKER_SUPPORT_EXCLUDE_LIBS - Success
[ 20s] -- Could NOT find GIF (missing: GIF_LIBRARY GIF_INCLUDE_DIR) (Required is at least version "5.1")
[ 20s] -- Could NOT find JPEG (missing: JPEG_LIBRARY JPEG_INCLUDE_DIR)
[ 20s] -- Could NOT find ZLIB (missing: ZLIB_LIBRARY ZLIB_INCLUDE_DIR)
[ 20s] -- Could NOT find PNG (missing: PNG_LIBRARY PNG_PNG_INCLUDE_DIR)
[ 20s] -- Checking for module 'OpenEXR'
[ 20s] -- Package 'OpenEXR', required by 'virtual:world', not found
[ 20s] -- Could NOT find Doxygen (missing: DOXYGEN_EXECUTABLE)
[ 20s] CMake Warning at CMakeLists.txt:457 (message):
[ 20s] asciidoc was not found, the man pages will not be installed.
[ 20s]
[ 20s]
[ 20s] -- Building with JPEGXL_VERSION=(unknown)
[ 20s] -- Could NOT find JPEG (missing: JPEG_LIBRARY JPEG_INCLUDE_DIR)
[ 20s] -- Could NOT find ZLIB (missing: ZLIB_LIBRARY ZLIB_INCLUDE_DIR)
[ 20s] -- Could NOT find PNG (missing: PNG_LIBRARY PNG_PNG_INCLUDE_DIR)
[ 20s] -- Checking for module 'libwebp'
[ 20s] -- Package 'libwebp', required by 'virtual:world', not found
[ 20s] -- Checking for module 'libavif'
[ 20s] -- Package 'libavif', required by 'virtual:world', not found
[ 20s] -- Building tools: cjxl;djxl;jxlinfo;cjpeg_hdr;benchmark_xl
[ 20s] -- Configuring done
[ 20s] -- Generating done
[ 20s] CMake Warning:
[ 20s] Manually-specified variables were not used by the project:
[ 20s]
[ 20s] BUILD_STATIC_LIBS
[ 20s] CMAKE_Fortran_FLAGS
[ 20s] CMAKE_INSTALL_DO_STRIP
[ 20s] CMAKE_MODULES_INSTALL_DIR
[ 20s] INCLUDE_INSTALL_DIR
[ 20s] LIB_INSTALL_DIR
[ 20s] SHARE_INSTALL_PREFIX
[ 20s] SYSCONF_INSTALL_DIR
```
**Expected behavior**
Build should succeed.
**Screenshots**
N/A
**Environment**
- OS: openSUSE Tumbleweed
- Compiler version: gcc12
- CPU type: armv7
- cjxl/djxl version string: 0.7.0
**Additional context**
N/A | Build failure on armv7 with libhwy 1.0.2 | https://api.github.com/repos/libjxl/libjxl/issues/1892/comments | 2 | 2022-11-14T07:37:33Z | 2022-11-17T07:08:54Z | https://github.com/libjxl/libjxl/issues/1892 | 1,447,544,442 | 1,892 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
A clear and concise description of what the bug is.
**To Reproduce**
Steps to reproduce the behavior:
change CMakefile.txt and add "set(CMAKE_OSX_DEPLOYMENT_TARGET "10.11" CACHE STRING "" FORCE)"
**Expected behavior**
A clear and concise description of what you expected to happen.
**Screenshots**
If applicable, add screenshots or example input/output images to help explain your problem.
```
$ cmake --build build
Consolidate compiler generated dependencies of target jxl_threads
[ 1%] Building CXX object lib/CMakeFiles/jxl_threads.dir/threads/resizable_parallel_runner.cc.o
In file included from /Users/Jeromy/Downloads/libjxl-0.7.0/lib/threads/resizable_parallel_runner.cc:8:
In file included from /Library/Developer/CommandLineTools/SDKs/MacOSX13.0.sdk/usr/include/c++/v1/algorithm:653:
In file included from /Library/Developer/CommandLineTools/SDKs/MacOSX13.0.sdk/usr/include/c++/v1/functional:500:
/Library/Developer/CommandLineTools/SDKs/MacOSX13.0.sdk/usr/include/c++/v1/__functional/function.h:44:13: error: 'operator delete' is unavailable: introduced in macOS 10.12
virtual ~bad_function_call() _NOEXCEPT {}
^
/Library/Developer/CommandLineTools/SDKs/MacOSX13.0.sdk/usr/include/c++/v1/new:185:74: note: 'operator delete' has been explicitly marked unavailable here
_LIBCPP_OVERRIDABLE_FUNC_VIS _LIBCPP_AVAILABILITY_SIZED_NEW_DELETE void operator delete(void* __p, std::size_t __sz) _NOEXCEPT;
^
/Users/Jeromy/Downloads/libjxl-0.7.0/lib/threads/resizable_parallel_runner.cc:186:3: error: 'operator delete' is unavailable: introduced in macOS 10.12
delete static_cast<jpegxl::ResizeableParallelRunner*>(runner_opaque);
^
/Library/Developer/CommandLineTools/SDKs/MacOSX13.0.sdk/usr/include/c++/v1/new:185:74: note: 'operator delete' has been explicitly marked unavailable here
_LIBCPP_OVERRIDABLE_FUNC_VIS _LIBCPP_AVAILABILITY_SIZED_NEW_DELETE void operator delete(void* __p, std::size_t __sz) _NOEXCEPT;
^
/Library/Developer/CommandLineTools/SDKs/MacOSX13.0.sdk/usr/include/c++/v1/new:245:3: error: 'operator delete' is unavailable: introduced in macOS 10.12
__builtin_operator_delete(__args...);
^
/Library/Developer/CommandLineTools/SDKs/MacOSX13.0.sdk/usr/include/c++/v1/new:271:10: note: in instantiation of function template specialization 'std::__libcpp_operator_delete<void *, unsigned long>' requested here
return __libcpp_operator_delete(__ptr, __size, __args...);
^
/Library/Developer/CommandLineTools/SDKs/MacOSX13.0.sdk/usr/include/c++/v1/new:279:12: note: in instantiation of function template specialization 'std::__do_deallocate_handle_size<>' requested here
return __do_deallocate_handle_size(__ptr, __size);
^
/Library/Developer/CommandLineTools/SDKs/MacOSX13.0.sdk/usr/include/c++/v1/new:185:74: note: 'operator delete' has been explicitly marked unavailable here
_LIBCPP_OVERRIDABLE_FUNC_VIS _LIBCPP_AVAILABILITY_SIZED_NEW_DELETE void operator delete(void* __p, std::size_t __sz) _NOEXCEPT;
^
In file included from /Users/Jeromy/Downloads/libjxl-0.7.0/lib/threads/resizable_parallel_runner.cc:8:
In file included from /Library/Developer/CommandLineTools/SDKs/MacOSX13.0.sdk/usr/include/c++/v1/algorithm:653:
In file included from /Library/Developer/CommandLineTools/SDKs/MacOSX13.0.sdk/usr/include/c++/v1/functional:500:
In file included from /Library/Developer/CommandLineTools/SDKs/MacOSX13.0.sdk/usr/include/c++/v1/__functional/function.h:20:
/Library/Developer/CommandLineTools/SDKs/MacOSX13.0.sdk/usr/include/c++/v1/__memory/shared_ptr.h:285:8: error: 'operator delete' is unavailable: introduced in macOS 10.12
struct __shared_ptr_emplace
^
/Library/Developer/CommandLineTools/SDKs/MacOSX13.0.sdk/usr/include/c++/v1/__memory/shared_ptr.h:290:14: note: in implicit destructor for 'std::__shared_ptr_emplace<std::mutex, std::allocator<std::mutex>>' first required here
explicit __shared_ptr_emplace(_Alloc __a, _Args&& ...__args)
^
/Library/Developer/CommandLineTools/SDKs/MacOSX13.0.sdk/usr/include/c++/v1/__memory/shared_ptr.h:1106:55: note: in instantiation of function template specialization 'std::__shared_ptr_emplace<std::mutex, std::allocator<std::mutex>>::__shared_ptr_emplace<>' requested here
::new ((void*)_VSTD::addressof(*__guard.__get())) _ControlBlock(__a, _VSTD::forward<_Args>(__args)...);
^
/Library/Developer/CommandLineTools/SDKs/MacOSX13.0.sdk/usr/include/c++/v1/__memory/shared_ptr.h:1115:19: note: in instantiation of function template specialization 'std::allocate_shared<std::mutex, std::allocator<std::mutex>, void>' requested here
return _VSTD::allocate_shared<_Tp>(allocator<_Tp>(), _VSTD::forward<_Args>(__args)...);
^
/Library/Developer/CommandLineTools/SDKs/MacOSX13.0.sdk/usr/include/c++/v1/condition_variable:170:14: note: in instantiation of function template specialization 'std::make_shared<std::mutex, void>' requested here
: __mut_(make_shared<mutex>()) {}
^
/Library/Developer/CommandLineTools/SDKs/MacOSX13.0.sdk/usr/include/c++/v1/new:185:74: note: 'operator delete' has been explicitly marked unavailable here
_LIBCPP_OVERRIDABLE_FUNC_VIS _LIBCPP_AVAILABILITY_SIZED_NEW_DELETE void operator delete(void* __p, std::size_t __sz) _NOEXCEPT;
^
In file included from /Users/Jeromy/Downloads/libjxl-0.7.0/lib/threads/resizable_parallel_runner.cc:8:
In file included from /Library/Developer/CommandLineTools/SDKs/MacOSX13.0.sdk/usr/include/c++/v1/algorithm:653:
In file included from /Library/Developer/CommandLineTools/SDKs/MacOSX13.0.sdk/usr/include/c++/v1/functional:500:
In file included from /Library/Developer/CommandLineTools/SDKs/MacOSX13.0.sdk/usr/include/c++/v1/__functional/function.h:20:
In file included from /Library/Developer/CommandLineTools/SDKs/MacOSX13.0.sdk/usr/include/c++/v1/__memory/shared_ptr.h:25:
/Library/Developer/CommandLineTools/SDKs/MacOSX13.0.sdk/usr/include/c++/v1/__memory/unique_ptr.h:57:5: error: 'operator delete' is unavailable: introduced in macOS 10.12
delete __ptr;
^
/Library/Developer/CommandLineTools/SDKs/MacOSX13.0.sdk/usr/include/c++/v1/__memory/unique_ptr.h:318:7: note: in instantiation of member function 'std::default_delete<std::__thread_struct>::operator()' requested here
__ptr_.second()(__tmp);
^
/Library/Developer/CommandLineTools/SDKs/MacOSX13.0.sdk/usr/include/c++/v1/__memory/unique_ptr.h:272:19: note: in instantiation of member function 'std::unique_ptr<std::__thread_struct>::reset' requested here
~unique_ptr() { reset(); }
^
/Library/Developer/CommandLineTools/SDKs/MacOSX13.0.sdk/usr/include/c++/v1/thread:308:12: note: in instantiation of member function 'std::unique_ptr<std::__thread_struct>::~unique_ptr' requested here
_TSPtr __tsp(new __thread_struct);
^
/Library/Developer/CommandLineTools/SDKs/MacOSX13.0.sdk/usr/include/c++/v1/__memory/allocator.h:154:28: note: in instantiation of function template specialization 'std::thread::thread<(lambda at /Users/Jeromy/Downloads/libjxl-0.7.0/lib/threads/resizable_parallel_runner.cc:33:31), void>' requested here
::new ((void*)__p) _Up(_VSTD::forward<_Args>(__args)...);
^
/Library/Developer/CommandLineTools/SDKs/MacOSX13.0.sdk/usr/include/c++/v1/__memory/allocator_traits.h:290:13: note: in instantiation of function template specialization 'std::allocator<std::thread>::construct<std::thread, (lambda at /Users/Jeromy/Downloads/libjxl-0.7.0/lib/threads/resizable_parallel_runner.cc:33:31)>' requested here
__a.construct(__p, _VSTD::forward<_Args>(__args)...);
^
/Library/Developer/CommandLineTools/SDKs/MacOSX13.0.sdk/usr/include/c++/v1/vector:948:21: note: in instantiation of function template specialization 'std::allocator_traits<std::allocator<std::thread>>::construct<std::thread, (lambda at /Users/Jeromy/Downloads/libjxl-0.7.0/lib/threads/resizable_parallel_runner.cc:33:31), void>' requested here
__alloc_traits::construct(this->__alloc(), _VSTD::__to_address(__tx.__pos_),
^
/Library/Developer/CommandLineTools/SDKs/MacOSX13.0.sdk/usr/include/c++/v1/vector:1706:9: note: in instantiation of function template specialization 'std::vector<std::thread>::__construct_one_at_end<(lambda at /Users/Jeromy/Downloads/libjxl-0.7.0/lib/threads/resizable_parallel_runner.cc:33:31)>' requested here
__construct_one_at_end(_VSTD::forward<_Args>(__args)...);
^
/Users/Jeromy/Downloads/libjxl-0.7.0/lib/threads/resizable_parallel_runner.cc:33:18: note: in instantiation of function template specialization 'std::vector<std::thread>::emplace_back<(lambda at /Users/Jeromy/Downloads/libjxl-0.7.0/lib/threads/resizable_parallel_runner.cc:33:31)>' requested here
workers_.emplace_back([this, i]() { WorkerBody(i); });
^
/Library/Developer/CommandLineTools/SDKs/MacOSX13.0.sdk/usr/include/c++/v1/new:185:74: note: 'operator delete' has been explicitly marked unavailable here
_LIBCPP_OVERRIDABLE_FUNC_VIS _LIBCPP_AVAILABILITY_SIZED_NEW_DELETE void operator delete(void* __p, std::size_t __sz) _NOEXCEPT;
^
In file included from /Users/Jeromy/Downloads/libjxl-0.7.0/lib/threads/resizable_parallel_runner.cc:8:
In file included from /Library/Developer/CommandLineTools/SDKs/MacOSX13.0.sdk/usr/include/c++/v1/algorithm:653:
In file included from /Library/Developer/CommandLineTools/SDKs/MacOSX13.0.sdk/usr/include/c++/v1/functional:500:
In file included from /Library/Developer/CommandLineTools/SDKs/MacOSX13.0.sdk/usr/include/c++/v1/__functional/function.h:20:
In file included from /Library/Developer/CommandLineTools/SDKs/MacOSX13.0.sdk/usr/include/c++/v1/__memory/shared_ptr.h:25:
/Library/Developer/CommandLineTools/SDKs/MacOSX13.0.sdk/usr/include/c++/v1/__memory/unique_ptr.h:57:5: error: 'operator delete' is unavailable: introduced in macOS 10.12
delete __ptr;
^
/Library/Developer/CommandLineTools/SDKs/MacOSX13.0.sdk/usr/include/c++/v1/__memory/unique_ptr.h:318:7: note: in instantiation of member function 'std::default_delete<std::tuple<std::unique_ptr<std::__thread_struct>, (lambda at /Users/Jeromy/Downloads/libjxl-0.7.0/lib/threads/resizable_parallel_runner.cc:33:31)>>::operator()' requested here
__ptr_.second()(__tmp);
^
/Library/Developer/CommandLineTools/SDKs/MacOSX13.0.sdk/usr/include/c++/v1/__memory/unique_ptr.h:272:19: note: in instantiation of member function 'std::unique_ptr<std::tuple<std::unique_ptr<std::__thread_struct>, (lambda at /Users/Jeromy/Downloads/libjxl-0.7.0/lib/threads/resizable_parallel_runner.cc:33:31)>>::reset' requested here
~unique_ptr() { reset(); }
^
/Library/Developer/CommandLineTools/SDKs/MacOSX13.0.sdk/usr/include/c++/v1/thread:310:21: note: in instantiation of member function 'std::unique_ptr<std::tuple<std::unique_ptr<std::__thread_struct>, (lambda at /Users/Jeromy/Downloads/libjxl-0.7.0/lib/threads/resizable_parallel_runner.cc:33:31)>>::~unique_ptr' requested here
unique_ptr<_Gp> __p(
^
/Library/Developer/CommandLineTools/SDKs/MacOSX13.0.sdk/usr/include/c++/v1/__memory/allocator.h:154:28: note: in instantiation of function template specialization 'std::thread::thread<(lambda at /Users/Jeromy/Downloads/libjxl-0.7.0/lib/threads/resizable_parallel_runner.cc:33:31), void>' requested here
::new ((void*)__p) _Up(_VSTD::forward<_Args>(__args)...);
^
/Library/Developer/CommandLineTools/SDKs/MacOSX13.0.sdk/usr/include/c++/v1/__memory/allocator_traits.h:290:13: note: in instantiation of function template specialization 'std::allocator<std::thread>::construct<std::thread, (lambda at /Users/Jeromy/Downloads/libjxl-0.7.0/lib/threads/resizable_parallel_runner.cc:33:31)>' requested here
__a.construct(__p, _VSTD::forward<_Args>(__args)...);
^
/Library/Developer/CommandLineTools/SDKs/MacOSX13.0.sdk/usr/include/c++/v1/vector:948:21: note: in instantiation of function template specialization 'std::allocator_traits<std::allocator<std::thread>>::construct<std::thread, (lambda at /Users/Jeromy/Downloads/libjxl-0.7.0/lib/threads/resizable_parallel_runner.cc:33:31), void>' requested here
__alloc_traits::construct(this->__alloc(), _VSTD::__to_address(__tx.__pos_),
^
/Library/Developer/CommandLineTools/SDKs/MacOSX13.0.sdk/usr/include/c++/v1/vector:1706:9: note: in instantiation of function template specialization 'std::vector<std::thread>::__construct_one_at_end<(lambda at /Users/Jeromy/Downloads/libjxl-0.7.0/lib/threads/resizable_parallel_runner.cc:33:31)>' requested here
__construct_one_at_end(_VSTD::forward<_Args>(__args)...);
^
/Users/Jeromy/Downloads/libjxl-0.7.0/lib/threads/resizable_parallel_runner.cc:33:18: note: in instantiation of function template specialization 'std::vector<std::thread>::emplace_back<(lambda at /Users/Jeromy/Downloads/libjxl-0.7.0/lib/threads/resizable_parallel_runner.cc:33:31)>' requested here
workers_.emplace_back([this, i]() { WorkerBody(i); });
^
/Library/Developer/CommandLineTools/SDKs/MacOSX13.0.sdk/usr/include/c++/v1/new:185:74: note: 'operator delete' has been explicitly marked unavailable here
_LIBCPP_OVERRIDABLE_FUNC_VIS _LIBCPP_AVAILABILITY_SIZED_NEW_DELETE void operator delete(void* __p, std::size_t __sz) _NOEXCEPT;
^
6 errors generated.
make[2]: *** [lib/CMakeFiles/jxl_threads.dir/threads/resizable_parallel_runner.cc.o] Error 1
make[1]: *** [lib/CMakeFiles/jxl_threads.dir/all] Error 2
make: *** [all] Error 2
```
**Environment**
- OS: [e.g. Windows]
- Compiler version: [e.g. clang 11.0.1]
- CPU type: [e.g. x86_64]
- cjxl/djxl version string: [e.g. cjxl [v0.3.7 | SIMD supported: SSE4,Scalar]]
**Additional context**
Add any other context about the problem here.
<!--
Currently github does not allow uploading files that end in `.jxl`, but when you
rename them for example as `image.jxl.jpg`, it will be possible to upload them
and also view them in browsers that are configured to support it.
See https://github.com/orgs/github-community/discussions/18139
-->
| Compile error in macOS Monterey with MacOSX13.0.sdk for osx target 10.11 | https://api.github.com/repos/libjxl/libjxl/issues/1891/comments | 1 | 2022-11-11T20:22:04Z | 2023-04-17T05:30:55Z | https://github.com/libjxl/libjxl/issues/1891 | 1,445,968,201 | 1,891 |
[
"libjxl",
"libjxl"
] | Take a JXL file tagged as 9-bit, for example this one: [wb-rainbow.jxl](https://user-images.githubusercontent.com/1342577/201196071-8b0d777c-cafe-4d46-890d-9d2dbbc61155.jpg)
`libjxl`-based decoders are able to read it correctly (such as ffmpeg, GIMP, etc.)
However, `djxl` produces corrupt output with the following:
```
$ djxl samples/wb-rainbow.jxl output.png
JPEG XL decoder v0.7.0 f95da131 [AVX2,SSE4,SSSE3,Unknown]
Read 352 compressed bytes.
Decoded to pixels.
2048 x 1152, 5.84 MP/s [5.84, 5.84], 1 reps, 4 threads.
```
`output.png` is too large to put on Github, but here's a link: [https://buzo.us/S.png](https://buzo.us/S.png)
**Additional Information:**
Explicitly setting `--bits_per_sample=8` and `--bits_per_sample=16` also fails (although I believe this bug was fixed on git main):
```
$ djxl samples/wb-rainbow.jxl --bits_per_sample=8 output.png
JPEG XL decoder v0.7.0 f95da131 [AVX2,SSE4,SSSE3,Unknown]
Read 352 compressed bytes.
Decoded to pixels.
Encode failed
$ djxl samples/wb-rainbow.jxl --bits_per_sample=16 output.png
JPEG XL decoder v0.7.0 f95da131 [AVX2,SSE4,SSSE3,Unknown]
Read 352 compressed bytes.
Decoded to pixels.
Encode failed
```
**Version Stuff:**
Arch Linux x86_64
djxl version: f95da131
| djxl cannot handle bit depths >8 and <16 correctly when decoding to PNG | https://api.github.com/repos/libjxl/libjxl/issues/1889/comments | 1 | 2022-11-10T20:18:43Z | 2022-11-13T20:17:00Z | https://github.com/libjxl/libjxl/issues/1889 | 1,444,506,866 | 1,889 |
[
"libjxl",
"libjxl"
] | **Describe the bug**
During noise synthesis, `libjxl` generates one batch of 16 floating point numbers for every partial 16-pixel wide row. After filling the row, it generates another batch of 16 floats for the stragglers, in case the the group width is not a multiple of 16 (which it isn't guaranteed to be, for example if the image dimensions are odd).
However, the group width usually is a multiple of 16, and if it is, then it generates this extra batch for the stragglers. And because there are no stragglers, it does nothing with it, so for a group width, of say, 512, it generates 33 batches where only 32 are required, and the last batch is unused.
Look at the loop on line 67 of `dec_noise.cc`:
```cpp
for (size_t y = 0; y < ysize; ++y) {
float* JXL_RESTRICT row = rect.Row(noise, y);
size_t x = 0;
// Only entire batches (avoids exceeding the image padding).
for (; x + kFloatsPerBatch <= xsize; x += kFloatsPerBatch) { // <-- HERE, Line 72
rng->Fill(batch);
for (size_t i = 0; i < kFloatsPerBatch; i += Lanes(df)) {
BitsToFloat(reinterpret_cast<const uint32_t*>(batch) + i, row + x + i);
}
}
// Any remaining pixels, rounded up to vectors (safe due to padding).
rng->Fill(batch);
size_t batch_pos = 0; // < kFloatsPerBatch
for (; x < xsize; x += N) {
BitsToFloat(reinterpret_cast<const uint32_t*>(batch) + batch_pos,
row + x);
batch_pos += N;
}
}
```
On line 72 (comment added), there's a non-strict inequality. I believe this should be a strict inequality.
Notice that if `xsize` is a multiple of `kFloatsPerBatch` (i.e. 16), then we have `x + 16 <= xsize` as the limiting condition, which is the same thing as `x < xsize`, so this fills the entire row. Then `rng->Fill(batch);` is called again, unconditionally.
**Additional Information**
This behavior isn't mentioned in the spec. Section K.5.2 in the spec has the following quote:
> The next 16 (or the number of remaining samples in the row, whichever is less)
It doesn't explicitly state that this is also generated if the number of remaining samples is zero, and it would be very strange to interpret it that way. | noise synthesis generates unused Xorshiro batch when noise group size is a multiple of 16 | https://api.github.com/repos/libjxl/libjxl/issues/1888/comments | 1 | 2022-11-10T17:02:56Z | 2022-12-02T12:31:54Z | https://github.com/libjxl/libjxl/issues/1888 | 1,444,257,921 | 1,888 |
[
"libjxl",
"libjxl"
] | Not related to this, but another coalescing mode that would be good to have for animation display, is one where the returned frames are blended but only the crop box containing modified pixels gets returned (so it only returns kReplace frames like the default coalescing, but not necessarily canvas-sized ones). This has obvious advantages in terms of buffer sizes and would allow e.g. browsers to have a higher likelihood of being able to keep the entire animation in memory.
So I think we basically need one boolean setting to indicate whether zero-duration frames should be blended or not, and another boolean setting to indicate whether cropped frames should be returned or canvas-sized ones. We currently only have an option where it is (true,false) (the default coalescing==true) and another where it is (false,true) (this is what coalescing==false does).
And then we need to find a way for dealing with exotic blend source choices; probably this is something most applications will not want to deal with so it's probably best if by default anything that is not a simple "stack of layers" gets converted in something that is (we can always do the blending and turn anything into a kReplace frame that overwrites the whole canvas), and only when some new option is set, you can get the original thing which may be quite exotic.
Also it would be good to have a way to tell the decoder not to return 'exotic' blend modes (i.e. not kReplace and not kBlend), since in many cases (like Gimp) the application will already know about replace and blend but not about the other blend modes.
_Originally posted by @jonsneyers in https://github.com/libjxl/libjxl/issues/1863#issuecomment-1310085823_
| intermediate coalescing modes / handling exotic blend sources/modes | https://api.github.com/repos/libjxl/libjxl/issues/1886/comments | 5 | 2022-11-10T10:39:41Z | 2022-11-29T10:58:45Z | https://github.com/libjxl/libjxl/issues/1886 | 1,443,655,484 | 1,886 |
[
"libjxl",
"libjxl"
] | I'd like to start implementing this new format on my android devices. Would it be possible to either have an .APK of the program that can then be run on a non-rooted device through a command line app? Or, more preferable, a GUI to accompany it so that we can start converting our raw files into a more bandwidth kind file format?
If the program is currently command-line-only, then I could see issues implementing it. But having an open source encoding/decoding (viewer) that can load up these files & convert them would be extremely helpful. Even just a GUI to enable converting/viewing the new file type more easily on Android could help broader adoption (barring native viewer/encoder support).
Since phones have become a primary source of image generation, offering a quick Android compatible GUI conversion & viewer app (particularly if it then enabled other file browsers to show thumbnails as a result) would enable more people to use it & thus potentially expand the window of use.
Thanks for your consideration! | Android APK or APK/GUI if not already present | https://api.github.com/repos/libjxl/libjxl/issues/1885/comments | 3 | 2022-11-10T03:26:00Z | 2025-01-18T00:13:08Z | https://github.com/libjxl/libjxl/issues/1885 | 1,443,176,575 | 1,885 |
[
"libjxl",
"libjxl"
] |
**Describe the bug**
See, https://github.com/libjxl/libjxl/pull/1863#issuecomment-1305294301
In the Wikipedia.jxl when reading the image with coalescing turned off, we see images that appear are of duration 0 and appear to be empty.
**To Reproduce**
Compile gimp plugin at https://github.com/libjxl/libjxl/commit/ddda9b2aefcefc0f8f2a15494ec048710669062d and load Wikipedia.jxl with that. (Or modify libjxl-info to also get those frames by setting coalescing to false there as well.)
**Expected behavior**
Frames with no information (like of duration 0, with no content) shouldn't appear, even with coalescing set to false.
**Additional context**
Might not really be a bug, but simply something we are not seeing in those frames that appear to be empty.
<!--
Currently github does not allow uploading files that end in `.jxl`, but when you
rename them for example as `image.jxl.jpg`, it will be possible to upload them
and also view them in browsers that are configured to support it.
See https://github.com/orgs/github-community/discussions/18139
-->
| investigate zero duration frames when coalescing is turned off | https://api.github.com/repos/libjxl/libjxl/issues/1882/comments | 0 | 2022-11-09T12:50:38Z | 2022-11-09T12:50:38Z | https://github.com/libjxl/libjxl/issues/1882 | 1,442,041,084 | 1,882 |
[
"libjxl",
"libjxl"
] | as outlined in #1366 | Implement `JxlDecoderSetOutputColorProfile` and `JxlDecoderSetCms` | https://api.github.com/repos/libjxl/libjxl/issues/1880/comments | 1 | 2022-11-09T12:10:17Z | 2024-03-04T10:12:55Z | https://github.com/libjxl/libjxl/issues/1880 | 1,441,983,661 | 1,880 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.