the problem with qwen-rapid-aio models
help please. wherever I use your models (I tried 19,22,23), there are noisy images everywhere in the result. it doesn't matter which workflow I use, I've tried different ones. I've been updating nodes, I've updated the graphics card driver. there are no problems with other models. please tell me the solution to the problem
I'm not sure but in my WF there's no load checkpoint, Qwen is a diffusion model so I use this :
Those 3 nodes are feeding what your loadcheckpoint does. Is your .safetensors in diffusion_models\ ?
If you paste a pic with your WF I can try it and modify it. Even the blurry image will do 😃
I'm not sure but in my WF there's no load checkpoint, Qwen is a diffusion model so I use this :
Those 3 nodes are feeding what your loadcheckpoint does. Is your .safetensors in diffusion_models\ ?
If you paste a pic with your WF I can try it and modify it. Even the blurry image will do 😃
This is an AIO (All-In-One) model that already includes the Text-Encoder, VAE, and Diffusion in a single file. Just use a Checkpoint Loader to load it; no separate nodes are required.
help please. wherever I use your models (I tried 19,22,23), there are noisy images everywhere in the result. it doesn't matter which workflow I use, I've tried different ones. I've been updating nodes, I've updated the graphics card driver. there are no problems with other models. please tell me the solution to the problem
Delete these 4 nodes. I recommend checking out Civitai to find a workflow (WF) that suits your preference instead.
Your workflow is inadequate.
Sorry, I didn't get how you're helping here, this comment is about Nuke1229? I'm glad you're trying to help but please be constructive.
Sorry, I didn't get how you're helping here, this comment is about Nuke1229? I'm glad you're trying to help but please be constructive.
just send him your workflow, that's literally the easiest fix. stop overcomplicating it
??? You just said it's inadequate. I don't do T2I with Qwen, all I can point out is Diffusion VS Checkpoint. You can have all my WF, I don't care but they won't fit the current need.
Here but I'm fearing it'll add more confusion.
??? You just said it's inadequate. I don't do T2I with Qwen, all I can point out is Diffusion VS Checkpoint. You can have all my WF, I don't care but they won't fit the current need.
Here but I'm fearing it'll add more confusion.
I actually reversed his workflow, tweaked a few things here and there, and it works fine now. and I saw your WF too. and he'll figure it out eventually. ComfyUI is just that kind of thing — one day it works, one day it doesn't, no one really knows why.