Olivio posted a video today about Differential Diffusion that lets you use the original checkpoint for inpainting. Would you consider the quality of the output as good as using the inpainting checkpoints?
Great video, as always. Thank you very much. Question: In WebUI 1111 inpainting, under the mask content, one can choose 'Fill,' 'Original,' 'Latent Noise,' etc. Is there a way to have such options in CumfyUI inpainting?
Oh that's such a good question - closest i see might be this: blenderneko.github.io/ComfyUI-docs/Core%20Nodes/Latent/inpaint/SetLatentNoiseMask/ I would have to test it though. Not finding anything for latent zeros either.
Idk man I'm not sold yet. It feels like it's hit or miss. Also there's Differential Diffusion which massively improves contextual understanding in inpainting.
Straight to the point thanks, finally can use those!
Loving your content and once again Great vid, thanks!
Thank you!!
Olivio posted a video today about Differential Diffusion that lets you use the original checkpoint for inpainting. Would you consider the quality of the output as good as using the inpainting checkpoints?
Oh good question!
Yeah, i don't know how i missed this - definitely looks like really cool stuff! Thanks for sharing!
Great video, as always. Thank you very much. Question: In WebUI 1111 inpainting, under the mask content, one can choose 'Fill,' 'Original,' 'Latent Noise,' etc. Is there a way to have such options in CumfyUI inpainting?
Oh that's such a good question - closest i see might be this: blenderneko.github.io/ComfyUI-docs/Core%20Nodes/Latent/inpaint/SetLatentNoiseMask/ I would have to test it though. Not finding anything for latent zeros either.
Thanks alot eh!
Heyo! Thank you!
a useful Tutorial as always... thanks. But Could we use an image as desired glasses for young woman ?
Idk man I'm not sold yet. It feels like it's hit or miss. Also there's Differential Diffusion which massively improves contextual understanding in inpainting.
Differential Diffusion. It works with all SDXL Models.