Hugging Face's logo

drbaph
/
FLUX.1-schnell-dev-merged-fp8

Text-to-Image
English
Image-generation
flux
Safetensors
Model card Files Community
FLUX.1-schnell-dev-merged-fp8 / README.md
drbaph's picture
drbaph
Update README.md
0a92880 verified
preview code
|
raw
history blame
2.46 kB
metadata
license: other
license_name: non-commercial
license_link: LICENSE
language:
  - en
pipeline_tag: text-to-image
tags:
  - Image-generation
  - flux
  - Safetensors

FLUX.1-schnell-dev-merged-fp8

This repository provides the merged .safetensors model file with the params as provided by sayakpaul for black-forest-labs/FLUX.1-dev and black-forest-labs/FLUX.1-schnell. Please be aware of the licenses of both the models before using the weights commercially.

For ComfyUI place the model in "models\unet"

Inference Steps: 6-8

Guidance Scale: 3.5

Sampler: Euler

I recommend you try:

6 steps (First Pass) / 2 Steps (Second Pass)
6 steps (First Pass) / 3 Steps (Second Pass)
8 steps (First Pass) / 2 Steps (Second Pass)

The second pass helps refine noisier generations

Merged (6 steps first pass / 2 steps second pass) Merged (6 steps first pass / 3 steps second pass) Merged (8 steps first pass / 2 steps second pass)
Merged 6 steps (1st pass) / 2 Steps (2nd pass) Merged 6 steps (1st pass) / 3 Steps (2nd pass) Merged 8 steps (1st pass) / 2 Steps (2nd pass)
@drbaph

License

This model falls under the FLUX.1 [dev] Non-Commercial License.