IN NO EVENT SHALL BLACK FOREST LABS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH USE OF THIS MODEL.

A realistic base model block merged from schnellMODE and flux.1 dev.

Parameters

FOR ALL MODELS - Use Euler / Beta, Euler / Normal, or Euler / Simple and 8 steps.
Recommended resolutions: 1Mpix or 1.5Mpix, try 2Mpix, but your results will vary.

Models with a 'g' in the version have text/prompt guidance layers. Use prompt guidance between 1 and 2.5 for the best images.

Notes

Note: Somtimes the flux base model will push it's way through giving that flux sheen to skin and the "flux chin", most of the time you can try again and it'll go away.

This is not based on schnell so is subject to the full flex dev license terms.

NOTICE: NF4 in ComfyUI - I highly recommend this node: DenkingOfficial/ComfyUI_UNet_bitsandbytes_NF4
You must install bitsandbytes version >0.43.3 using pip pip install bitsandbytes==0.45.2

Versioning

v0.3g - Added guidance to allow guidance config to work.

v0.3 - Much more stable, a lot more good seeds, better text

v0.2 - Less unhinged, more good seeds, experimentation needed!

v0.1 - is fairly unhinged, some good seeds, some bad.

Quality

Quality (best first): FP16 > Q8 > FP8 > Q6_K > Q4_1 > Q4_K > NF4 > FP4 > Q3_K_S
Speed (fastest first): FP8 - GGUF - FP16

1MPix

1:1 - 1024 x 1024
4:3 - 1152 x 896
16:9 - 1344 x 768

1.5MPix

1:1 - 1280 x 1280
4:3 - 1440 x 1120
16:9 - 1680 x 960
21:9 - 1920 x 800

2Mpix
1:1 - 1408 x 1408
4:3 - 1664 x 1216
16:9 - 1920 x 1088