Flux Q8 or Nf4v2 + Batch + Upscale 4xUltraSharp
5.0
0 reviewsDescription
My old workflow (Nf4) but enhanced with proper groups and flux Q8 (nf4 node still present but bypassed u can connect if u still use it) with nf4 change flux guidance from 4 to 3.5, it changes render times a lot for me
but i recommend Q8 and change the values for the resolution if it takes too long
custom node https://github.com/city96/ComfyUI-GGUF (install via manager, simpler)
go to ComfyUI/custom_nodes/ComfyUI-GGUF
.\python_embeded\python.exe -s -m pip install -r .\ComfyUI\custom_nodes\ComfyUI-GGUF\requirements.txt
(i think the manager does this by itself, but i'm noob so have it)
Get the Q8 gguf and put in UNET folder
https://huggingface.co/city96/FLUX.1-dev-gguf/blob/main/flux1-dev-Q8_0.gguf
As always, ignore the cursed prompts, fbi is knocking rn...
Photos done with Q8
Sorry if errors, sill a noob to this
Discussion
(No comments yet)
Loading...
Resources (2)
Reviews
No reviews yet
Versions (2)
- latest (a year ago)
- v20240818-185416
Node Details
Primitive Nodes (24)
Anything Everywhere (4)
CheckpointLoaderNF4 (1)
EmptySD3LatentImage (1)
Fast Groups Bypasser (rgthree) (2)
FluxGuidance (1)
Image Comparer (rgthree) (1)
Note (5)
Reroute (8)
UnetLoaderGGUF (1)
Custom Nodes (21)
- CM_SDXLResolution (1)
ComfyUI
- CLIPTextEncode (2)
- VAEDecode (1)
- SaveImage (2)
- UpscaleModelLoader (1)
- ImageSharpen (1)
- CheckpointLoaderSimple (1)
- KSampler (1)
- easy float (1)
- ImpactFloat (1)
- ImageGenResolutionFromImage (1)
- SDXLAspectRatioSelector (1)
- FilmGrain (1)
- Preview Chooser (1)
- MathExpression|pysssss (4)
- UltimateSDUpscale (1)
Model Details
Checkpoints (2)
flux1-dev-bnb-nf4-v2.safetensors
flux1-dev-fp8.safetensors
LoRAs (0)