Diffusion Single File
English
image-generation
flux

change the name to FLUX.1-Redux-dev-lora?

#1
by martintomov - opened

Amazing work. But as this is an adapter, maybe it would be less confusing to include that in the title as well. Initially, I thought it was a base model like Fill-dev, Canny-dev, and Depth-dev.

Amazing work. But as this is an adapter, maybe it would be less confusing to include that in the title as well. Initially, I thought it was a base model like Fill-dev, Canny-dev, and Depth-dev.

But it is not applied as a LoRA, but as a "style model." Is that different than a LoRA?

what do you mean... this is a lora, it sits on top of the standard dev to make the model act as the redux model would so we deont have to have multiple 23gb models

If it was a lora, then it would be applied as a lora. This is not applied as a lora, nor downloaded to the "lora" folder, but as a style model downloaded to the "style_model" folder. Are style models a new thing?

how do you use this in any of the current webuis?

@Hesajon is right, it's a style model.

1.Place this CLIP model in ComfyUI/models/clip_vision:

https://huggingface.co/Comfy-Org/sigclip_vision_384/resolve/main/sigclip_vision_patch14_384.safetensors

2.Place the redux model in ComfyUI/models/style_models:

https://huggingface.co/black-forest-labs/FLUX.1-Redux-dev/resolve/main/flux1-redux-dev.safetensors

martintomov changed discussion status to closed

Is there any example workflow for this?

Is there any example workflow for this?

Yes, here: https://comfyanonymous.github.io/ComfyUI_examples/flux/#redux

Sign up or log in to comment