|
|
@@ -1,52 +1,65 @@
|
|
|
# Comparisons among different anime models
|
|
|
|
|
|
## Update News
|
|
|
-- 2022/04/22: Release AnimeVideo-**v3**, we have made the following improvements:
|
|
|
+
|
|
|
+- 2022/04/24: Release **AnimeVideo-v3**. We have made the following improvements:
|
|
|
- **better naturalness**
|
|
|
- - **less artifacts**
|
|
|
+ - **Fewer artifacts**
|
|
|
- **more faithful to the original colors**
|
|
|
- **better texture restoration**
|
|
|
- **better background restoration**
|
|
|
|
|
|
-## Compare methods
|
|
|
-* [waifu2x](https://github.com/nihui/waifu2x-ncnn-vulkan): the hyperparameters: `tile=0 `, `noiselevel=2`
|
|
|
-* [Real-CUGAN](https://github.com/bilibili/ailab/tree/main/Real-CUGAN):
|
|
|
- we use the [20220227](https://github.com/bilibili/ailab/releases/tag/Real-CUGAN-add-faster-low-memory-mode) version,
|
|
|
- the hyperparameters are: `cache_mode=0`, `tile=0`, `alpha=1`.
|
|
|
-* RealESRGAN-AnimeVideo-**v3** (ours latest released animevideo model)
|
|
|
+## Comparisons
|
|
|
+
|
|
|
+We have compared our RealESRGAN-AnimeVideo-v3 with the following methods.
|
|
|
+Our RealESRGAN-AnimeVideo-v3 can achieve better results with faster inference speed.
|
|
|
+
|
|
|
+- [waifu2x](https://github.com/nihui/waifu2x-ncnn-vulkan) with the hyperparameters: `tile=0`, `noiselevel=2`
|
|
|
+
|
|
|
+- [Real-CUGAN](https://github.com/bilibili/ailab/tree/main/Real-CUGAN):
|
|
|
+ we use the [20220227](https://github.com/bilibili/ailab/releases/tag/Real-CUGAN-add-faster-low-memory-mode) version, the hyperparameters are: `cache_mode=0`, `tile=0`, `alpha=1`.
|
|
|
+- our RealESRGAN-AnimeVideo-v3
|
|
|
|
|
|
-## results
|
|
|
+## Results
|
|
|
|
|
|
You may need to **zoom in** for comparing details, or **click the image** to see in the full size.
|
|
|
|
|
|
+**More natural results, better background restoration**
|
|
|
| Input | waifu2x | Real-CUGAN | RealESRGAN<br>AnimeVideo-v3 |
|
|
|
| :---: | :---: | :---: | :---: |
|
|
|
-| |  |  |  |
|
|
|
-| |  |  |  |
|
|
|
-| |  |  |  |
|
|
|
| |  |  |  |
|
|
|
-| |  |  |  |
|
|
|
| |  |  |  |
|
|
|
+| |  |  |  |
|
|
|
+
|
|
|
+**Fewer artifacts, better detailed textures**
|
|
|
+| Input | waifu2x | Real-CUGAN | RealESRGAN<br>AnimeVideo-v3 |
|
|
|
+| :---: | :---: | :---: | :---: |
|
|
|
+| |  |  |  |
|
|
|
+| |  |  |  |
|
|
|
| |  |  |  |
|
|
|
+| |  |  |  |
|
|
|
+
|
|
|
+**Other better results**
|
|
|
+| Input | waifu2x | Real-CUGAN | RealESRGAN<br>AnimeVideo-v3 |
|
|
|
+| :---: | :---: | :---: | :---: |
|
|
|
| |  |  |  |
|
|
|
-| |  |  |  |
|
|
|
-| |  |  |  |
|
|
|
| |  |  |  |
|
|
|
|  |   |   |   |
|
|
|
+| |  |  |  |
|
|
|
+| |  |  |  |
|
|
|
|
|
|
+## Inference Speed
|
|
|
|
|
|
+### PyTorch
|
|
|
|
|
|
+Note that we only report the **model** time, and ignore the IO time.
|
|
|
|
|
|
-
|
|
|
-## Speed
|
|
|
-### pytorch
|
|
|
-Note that we only report the **model** speed, and IO time is not considered.
|
|
|
-
|
|
|
-| GPU | Input Resolution | waifu2x | Real-CUGAN | RealESRGAN-AnimeVideo-v3
|
|
|
+| GPU | Input Resolution | waifu2x | Real-CUGAN | RealESRGAN-AnimeVideo-v3
|
|
|
| :---: | :---: | :---: | :---: | :---: |
|
|
|
-| V100 | 1921x1080 | - | 3.4 fps | **10.0** fps |
|
|
|
-| V100 | 1280x720 | - | 7.2 fps | **22.6** fps |
|
|
|
-| V100 | 640x480 | - | 24.4 fps | **65.9** fps |
|
|
|
+| V100 | 1921 x 1080 | - | 3.4 fps | **10.0** fps |
|
|
|
+| V100 | 1280 x 720 | - | 7.2 fps | **22.6** fps |
|
|
|
+| V100 | 640 x 480 | - | 24.4 fps | **65.9** fps |
|
|
|
|
|
|
### ncnn
|
|
|
-- [ ] update ncnn library of [Real-ESRGAN-ncnn-vulkan](https://github.com/xinntao/Real-ESRGAN-ncnn-vulkan) for a fair comparison.
|
|
|
+
|
|
|
+- [ ] TODO
|
Xintao