Gerold Meisinger
commited on
Commit
•
4918ff0
1
Parent(s):
0b3513f
control-edgedrawing-cv480edpf-drop0-fp16-checkpoint-90000.safetensors
Browse files
README.md
CHANGED
|
@@ -27,6 +27,7 @@ For usage see the model page on [Civitai.com](https://civitai.com/models/149740)
|
|
| 27 |
**Example**
|
| 28 |
|
| 29 |
sampler=UniPC steps=20 cfg=7.5 seed=0 batch=9 model: v1-5-pruned-emaonly.safetensors cherry-picked: 1/9
|
|
|
|
| 30 |
prompt: _a detailed high-quality professional photo of swedish woman standing in front of a mirror, dark brown hair, white hat with purple feather_
|
| 31 |
|
| 32 |
|
|
@@ -58,7 +59,7 @@ accelerate launch train_controlnet.py ^
|
|
| 58 |
|
| 59 |
# Versions
|
| 60 |
|
| 61 |
-
**
|
| 62 |
|
| 63 |
Conditioning images generated with [edpf.py](https://gitlab.com/-/snippets/3601881) using [opencv-contrib-python::ximgproc::EdgeDrawing](https://docs.opencv.org/4.8.0/d1/d1c/classcv_1_1ximgproc_1_1EdgeDrawing.html).
|
| 64 |
|
|
@@ -71,9 +72,13 @@ edges = ed.detectEdges(image)
|
|
| 71 |
edge_map = ed.getEdgeImage(edges)
|
| 72 |
```
|
| 73 |
|
| 74 |
-
45000 steps
|
|
|
|
|
|
|
|
|
|
|
|
|
| 75 |
|
| 76 |
-
**
|
| 77 |
|
| 78 |
Images converted with https://github.com/shaojunluo/EDLinePython
|
| 79 |
|
|
@@ -94,9 +99,9 @@ Default settings are:
|
|
| 94 |
Trained for 40000 steps with default settings
|
| 95 |
=> conditioning images are too noisy
|
| 96 |
|
| 97 |
-
**
|
| 98 |
|
| 99 |
-
Same as
|
| 100 |
|
| 101 |
Update: bug in algorithm produces too sparse images on default, see https://github.com/shaojunluo/EDLinePython/issues/4
|
| 102 |
|
|
|
|
| 27 |
**Example**
|
| 28 |
|
| 29 |
sampler=UniPC steps=20 cfg=7.5 seed=0 batch=9 model: v1-5-pruned-emaonly.safetensors cherry-picked: 1/9
|
| 30 |
+
|
| 31 |
prompt: _a detailed high-quality professional photo of swedish woman standing in front of a mirror, dark brown hair, white hat with purple feather_
|
| 32 |
|
| 33 |
|
|
|
|
| 59 |
|
| 60 |
# Versions
|
| 61 |
|
| 62 |
+
**Experiment 4 - control-edgedrawing-cv480edpf-drop0-fp16-checkpoint-90000**
|
| 63 |
|
| 64 |
Conditioning images generated with [edpf.py](https://gitlab.com/-/snippets/3601881) using [opencv-contrib-python::ximgproc::EdgeDrawing](https://docs.opencv.org/4.8.0/d1/d1c/classcv_1_1ximgproc_1_1EdgeDrawing.html).
|
| 65 |
|
|
|
|
| 72 |
edge_map = ed.getEdgeImage(edges)
|
| 73 |
```
|
| 74 |
|
| 75 |
+
90000 steps (45000 steps on original, 45000 steps with left-right flipped images)
|
| 76 |
+
|
| 77 |
+
**Experiment 3 - control-edgedrawing-cv480edpf-drop0-fp16-checkpoint-45000**
|
| 78 |
+
|
| 79 |
+
see experiment 4. 45000 steps. This is version 0.1 on civitai.
|
| 80 |
|
| 81 |
+
**Experiment 2 - control-edgedrawing-default-noisy-drop0-fp16-checkpoint-40000**
|
| 82 |
|
| 83 |
Images converted with https://github.com/shaojunluo/EDLinePython
|
| 84 |
|
|
|
|
| 99 |
Trained for 40000 steps with default settings
|
| 100 |
=> conditioning images are too noisy
|
| 101 |
|
| 102 |
+
**Experiment 1 - control-edgedrawing-default-drop50-fp16-checkpoint-40000**
|
| 103 |
|
| 104 |
+
Same as experiment 2.
|
| 105 |
|
| 106 |
Update: bug in algorithm produces too sparse images on default, see https://github.com/shaojunluo/EDLinePython/issues/4
|
| 107 |
|
control-edgedrawing-cv480edpf-drop0-fp16-checkpoint-90000.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:db33348a906ea79b17ba343cdb9fd02f990a175e34f2eb31c5daac7ec9cb840c
|
| 3 |
+
size 722598616
|