HowTo better inpaint with Differential Diffusion in ComfyUI AI, awesome results, precise controll
HTML-код
- Опубликовано: 10 апр 2024
- In this tutorial I walk you through a simple Differential Diffusion inpainting workflow in ComfyUI. Differential Diffusion gives you uch more controll over your inpainting with maps / masks which enables you to precisely controll the amount of inpaint at every area of your image.
Be sure to install all the needed custom nodes which are linked below.
Please comment if you have questions or tell me your suggestions for future videos.
Sintific paper:
differential-diffusion.github...
Custom Nodes:
github.com/ltdrdata/ComfyUI-I...
github.com/cubiq/ComfyUI_esse...
The Model:
huggingface.co/stabilityai/st...
Connect with me:
/ _neuron_ai
/ _neuron_ai
The equipment that I use (affiliate links):
GPU: ASUS ROG STRIX RTX 4090 amzn.to/42ptUTi
#comfyui #stablecascade #stablediffusion #tutorial #animatediff #automatic1111 #aiart #ai #inpaint #interpolation - Наука
Thank you! It works very well for me.
Good stuff again and thanks Olivio :)
Olivio?
How would you add a Lora Loader to this workflow? I tried adding one between the checkpoint and the diffusion node and it seems to obliterate my Lora?
I answered my own question: set the sampler denoise to less than 1. I found it working best at 0.8 and below.
How to get Work Flow in Comfy UI, like Only Mask in Automatic 1111
I dont know auromatic1111 so well and dont know what you mean. do you want to load the mask from somewhere else?
Hey, where did you find the differentialDiffusion node ? I can't find it on the manager
If you update your comfyui to the latest version it should be there. It is no custom node.
Hi! May I ask a question?
Can "differential diffusion" be used with controlnet? I heard that there's a new, more accurate inpainting tool called "BrushNet".
I've tried it out, but it seems like it can't be used with a controlnet.
Is it possible to use "differential diffusion" with controlnet? If it's possible, I'd like to give it a try. Thanks!
Didn't try it myself yet. So can not say.
I tried for several hours, but it didn't work out. It's sad, but I'm probably giving up thanks for your reply though 👍
@@rei6477 sorry for that, I will tell you if I find a way.