冒昧问一项关于修图功能的希望

[longfei796]

如果话说的不符合一些事情请见作者谅，我是外行小白
这个作者的开源项目https://github.com/Sanster/IOPaint/tree/iopaint-1.5.0
修图功能惊为天人
这个开源项目可以加到ballontranslator里吗
我也不会py跟pip

[bropines]

如果话说的不符合一些事情请见作者谅，我是外行小白 这个作者的开源项目Sanster/IOPaint@iopaint-1.5.0 修图功能惊为天人 这个开源项目可以加到ballontranslator里吗 我也不会py跟pip

It uses one of our models trained by DmMaze. So essentially we don't need to add anything. We already have a normal Inpaint model that works well with manga

[longfei796]

如果话说的不符合一些事情请见作者谅，我是外行小白 这个作者的开源项目Sanster/IOPaint@iopaint-1.5.0 修图功能惊为天人 这个开源项目可以加到ballontranslator里吗 我也不会py跟pip

It uses one of our models trained by DmMaze. So essentially we don't need to add anything. We already have a normal Inpaint model that works well with manga

what？i don't know that
Which one is it
opencv—tela，patchmatch，aot，lama-mpe，lama-large-512px

[bropines]

lama-large-512px

This release describes this:
https://github.com/Sanster/IOPaint/releases/tag/iopaint-1.5.0

[longfei796]

lama-large-512px

This release describes this: https://github.com/Sanster/IOPaint/releases/tag/iopaint-1.5.0

thanks and May I ask what these two functions are

[bropines]

1. Roughly speaking, this is the maximum size (like in pixels) that it uses to generate new data, but I could be wrong.
2. This is a method of calculation. If you have a 20xx-40xx Nvidia video card, then feel free to turn on bf16; if you have lower models, fp32 is better.

[longfei796]

1. Roughly speaking, this is the maximum size (like in pixels) that it uses to generate new data, but I could be wrong.
2. This is a method of calculation. If you have a 20xx-40xx Nvidia video card, then feel free to turn on bf16; if you have lower models, fp32 is better.

thank you so much
i will ask the author for more details