Update README.md
Browse files
README.md
CHANGED
|
@@ -1,3 +1,69 @@
|
|
| 1 |
-
---
|
| 2 |
-
license: cc-by-4.0
|
| 3 |
-
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: cc-by-4.0
|
| 3 |
+
---
|
| 4 |
+
|
| 5 |
+
[Link to Github Release](https://github.com/Phhofm/models/releases/tag/4xFFHQDAT)
|
| 6 |
+
|
| 7 |
+
Name: 4xFFHQDAT
|
| 8 |
+
Author: Philip Hofmann
|
| 9 |
+
Release Date: 25.08.2023
|
| 10 |
+
License: CC BY 4.0
|
| 11 |
+
Network: DAT
|
| 12 |
+
Scale: 4
|
| 13 |
+
Purpose: 4x upscaling model for faces
|
| 14 |
+
Iterations: 122000
|
| 15 |
+
epoch: 2
|
| 16 |
+
batch_size: 4
|
| 17 |
+
HR_size: 128
|
| 18 |
+
Dataset: FFHQ - full dataset till 50k, then first 10k img multiscaled (resulted in ~260k imgs, 126GB)
|
| 19 |
+
Number of train images: 259990
|
| 20 |
+
OTF Training: Yes
|
| 21 |
+
Pretrained_Model_G: DAT_x4.pth
|
| 22 |
+
|
| 23 |
+
Description: 4x photo upscaler for faces with otf jpg compression, blur and resize, trained on FFHQ dataset. This has been trained on and for faces, but i guess can also be used for other photos, might be able to retain skin detail. This is not face restoration, but simply a 4x upscaler trained on faces, therefore input images need to be of good quality if good output quality is desired.
|
| 24 |
+
|
| 25 |
+
Examples 4xFFHQDAT:
|
| 26 |
+
[Imgsli1](https://imgsli.com/MjAwNjUz)
|
| 27 |
+
[Imgsli2](https://imgsli.com/MjAwNjU0)
|
| 28 |
+
[Imgsli3](https://imgsli.com/MjAwNjU2)
|
| 29 |
+
[Imgsli4](https://imgsli.com/MjAwNjU3)
|
| 30 |
+
[Imgsli5](https://imgsli.com/MjAwNjU4)
|
| 31 |
+
[Imgsli6](https://imgsli.com/MjAwNjU5)
|
| 32 |
+
[Imgsli7](https://imgsli.com/MjAwNzk0)
|
| 33 |
+
|
| 34 |
+
|
| 35 |
+
|
| 36 |
+
|
| 37 |
+
|
| 38 |
+
|
| 39 |
+
|
| 40 |
+
|
| 41 |
+
---
|
| 42 |
+
|
| 43 |
+
|
| 44 |
+
Since the above 4xFFHQDAT model is not able to handle the noise present in low quality input images, i made a small variant/finetune of this, the 4xFFHQLDAT model. This model might come in handy if your input image is of bad quality/not suited for above model. I basically made this model in a response to an input image posted in upscaling-results channel as a request to this upscale model (since 4xFFHQDAT would not be able to handle noise), see Imgsli1 example below for result.
|
| 45 |
+
|
| 46 |
+
Name: 4xFFHQLDAT
|
| 47 |
+
Author: Philip Hofmann
|
| 48 |
+
Release Date: 25.08.2023
|
| 49 |
+
License: CC BY 4.0
|
| 50 |
+
Network: DAT
|
| 51 |
+
Scale: 4
|
| 52 |
+
Purpose: 4x upscaling model for low quality input photos of faces
|
| 53 |
+
Iterations: 44000
|
| 54 |
+
epoch: 0
|
| 55 |
+
batch_size: 4
|
| 56 |
+
HR_size: 128
|
| 57 |
+
Dataset: FFHQ - full dataset till 50k, then first 10k img multiscaled (resulted in ~260k imgs, 126GB)
|
| 58 |
+
Number of train images: 259990
|
| 59 |
+
OTF Training: Yes
|
| 60 |
+
Pretrained_Model_G: 4xFFHQDAT
|
| 61 |
+
|
| 62 |
+
Examples 4xFFHQLDAT:
|
| 63 |
+
[Imgsli1](https://imgsli.com/MjAwNjYx)
|
| 64 |
+
[Imgsli2](https://imgsli.com/MjAwNjYy)
|
| 65 |
+
[Imgsli3](https://imgsli.com/MjAwNjYz)
|
| 66 |
+
|
| 67 |
+
|
| 68 |
+
|
| 69 |
+
|