search

comfyanonymous / ComfyUI

The most powerful and modular diffusion model GUI, api and backend with a graph/nodes interface.
https://www.comfy.org/
GNU General Public License v3.0
50.57k stars 5.31k forks source link

Text -> Image then Inpaint Workflow Recommendation #2022

Open ynie opened 9 months ago

ynie commented 9 months ago

Hey all,

I love using ComfyUI and thanks for the work. I want some recommendations on how to set up this workflow. I usually start with a 10 images batch to generate a background first, then I choose the best one and inpaint some items on it. Is this possible to do in one workflow? If I do like the background, I do not want comfyui to re-generate it. Thanks!

jn-jairo commented 9 months ago

I don't do exactly it but I have a big workflow with nodes separeted in groups, so I keep the groups disabled and enable only the parts I need.

If I would add something like that in my workflow I would add a group for each task and drag the background I like into the LoadImage node of the other group, enable that part of the workflow and disable the other.

Another way I could do it is using some custom node to get only the image I like from the batch (I am not sure if it can be done using basic nodes, but you can look for nodes related to batch) and have it connected to the other part of the workflow that I would enable after selecting the background I want, but for that don't generate the background again the seed should be fixed.