I’ve been dicking with stablediffusion for a while, now using comfyui… I have never been able to generate anything this good. What’s the deal? Am I using the wrong tools or just using them wrong? Anyone have any advise?
Play around with sampling methods, clip skipping, guidance etc. It takes a bit of getting used to, and there isn’t always a one-size fits all to each model/checkpoint
I’ve been dicking with stablediffusion for a while, now using comfyui… I have never been able to generate anything this good. What’s the deal? Am I using the wrong tools or just using them wrong? Anyone have any advise?
You need a lot of VRAM and a large visual model for higher complexity.
Lower VRAM means you run models that only do one thing consistently/well.
See: FLUX
I have 2x 24G 3090 but IIRC comfyui doesn’t support multiple GPU. That seem too low?
Play around with sampling methods, clip skipping, guidance etc. It takes a bit of getting used to, and there isn’t always a one-size fits all to each model/checkpoint