tvt commited on
Commit
57f52d5
1 Parent(s): cf8460e
Files changed (1) hide show
  1. app.py +5 -3
app.py CHANGED
@@ -56,7 +56,9 @@ if __name__ == '__main__':
56
  examples=[['test-img.jpg', "up2x-latest-denoise2x.pth", 2]],
57
  article='[https://github.com/bilibili/ailab/tree/main/Real-CUGAN](https://github.com/bilibili/ailab/tree/main/Real-CUGAN)<br>'
58
  '感谢b站开源的项目,图片过大会导致内存不足,所有我将图片裁剪小,想体验大图片的效果请自行前往上面的链接。<br>'
59
- '修改bbb'
60
- 'The large image will lead to memory limit exceeded. So I crop and resize image. '
61
- 'If you want to experience the large image, please go to the link above.')
 
 
62
  iface.launch()
 
56
  examples=[['test-img.jpg', "up2x-latest-denoise2x.pth", 2]],
57
  article='[https://github.com/bilibili/ailab/tree/main/Real-CUGAN](https://github.com/bilibili/ailab/tree/main/Real-CUGAN)<br>'
58
  '感谢b站开源的项目,图片过大会导致内存不足,所有我将图片裁剪小,想体验大图片的效果请自行前往上面的链接。<br>'
59
+ '参数路径,可更换,最前面是超分倍率,过大会内存不足<br>'
60
+ '降噪版(denoise):如果原片噪声多,压得烂,推荐使用;目前2倍模型支持了3个降噪等级<br>'
61
+ '无降噪版(no-denoise):如果原片噪声不多,压得还行,但是想提高分辨率/清晰度/做通用性的增强、修复处理,推荐使用<br>'
62
+ '保守版(conservative):如果你担心丢失纹理,担心画风被改变,担心颜色被增强,总之就是各种担心AI会留下浓重的处理痕迹,推荐使用该版本。<br>'
63
+ 'tile越大,越省显存,速度越慢')
64
  iface.launch()