← sources

Graphic producing new model Ming works fine in Comfy

source: reddit (reddit:StableDiffusion)published 2026-09-23 16:17ingested 2026-09-24 02:38

retained text

Original Post : [https://www.reddit.com/r/StableDiffusion/comments/1wnhqm3/open\_source\_mingimage01design\_family/](https://www.reddit.com/r/StableDiffusion/comments/1wnhqm3/open_source_mingimage01design_family/)  
Kijai's PR : [https://github.com/Comfy-Org/ComfyUI/pull/16482](https://github.com/Comfy-Org/ComfyUI/pull/16482)

NB: I gave practically zero detailed info to the prompt (going for proof of concept) , so it winged it . It is NOT yet in the main branch.

**(EDIT IN: see the pic in the comments - used a very detailed prompt this time)** 

Method : made a new comfy install and installed the PR , copied the workflow from Kijai's PR page and bunged in a brief into ChatGPT for a Llamafarm .

And it works, can obviously be tweaked to whatever you want. I used the Qwen38BFluxKlein clip model in my workflow but I think it might need the better model partially noted in the PR Pic (search engine here I come) . It was using \~19.7gb in my 4090 and took \~50s to make a 2048 x 2048 graphic at only 12 steps (it might need more).

document fetched 2026-09-24 00:35 · canonical https://reddit.com/r/StableDiffusion/comments/1woagyt/graphic_producing_new_model_ming_works_fine_in