improve flux docs

This commit is contained in:
Alex "mcmonkey" Goodwin
2024-08-15 11:09:53 -07:00
parent bc2399a876
commit 13c1e6b9d6

View File

@@ -97,15 +97,16 @@ Download the model, then click "`Edit Metadata`" and select `(Temporary) AuraFlo
# Black Forest Labs' Flux.1 Models
- Black Forest Labs' Flux.1 model is fully supported in Swarm <https://blackforestlabs.ai/announcing-black-forest-labs/>
- You can download BFL's original files:
- **Recommended:** use the [NF4 Format Files](#bits-and-bytes-nf4-format-models)
- **Alternate:** the simplified fp8 file:
- Dev <https://huggingface.co/Comfy-Org/flux1-dev/blob/main/flux1-dev-fp8.safetensors>
- Schnell <https://huggingface.co/Comfy-Org/flux1-schnell/blob/main/flux1-schnell-fp8.safetensors>
- goes in your regular `(Swarm)/Models/Stable-Diffusion` dir
- **or, not recommended:** You can download BFL's original files:
- Download "Schnell" (Turbo) from <https://huggingface.co/black-forest-labs/FLUX.1-schnell>
- Or "Dev" (non-Turbo) from <https://huggingface.co/black-forest-labs/FLUX.1-dev>
- Put dev/schnell in `(Swarm)/Models/unet`
- Put the `ae.sft` file in `(Swarm)/Models/VAE`
- Or the simplified fp8 file:
- Dev <https://huggingface.co/Comfy-Org/flux1-dev/blob/main/flux1-dev-fp8.safetensors>
- Schnell <https://huggingface.co/Comfy-Org/flux1-schnell/blob/main/flux1-schnell-fp8.safetensors>
- goes in your regular `(Swarm)/Models/Stable-Diffusion` dir
- For both models, use CFG=1 (negative prompt won't work). Sampling leave default (will use Euler + Simple)
- For the Dev model, there is also a `Flux Guidance Scale` parameter under `Sampling`, which is a distilled embedding value that the model was trained to use.
- Dev can use some slightly-higher CFG values (allowing for negative prompt), possibly higher if you reduce the Flux Guidance value and/or use Dynamic Thresholding.