This is a Flux LoRA extracted from Krea Dev model using https://github.com/kijai/ComfyUI-FluxTrainer

The purpose of this model is to be able to plug it into Flux Kontext (tested) or Flux Schnell

Image details might not be matching the original 100%, but overall it's very close

Model rank is 256. When loading it, use model weight of 1.0, and clip weight of 0.0.

Rank 64 and 512 are added, too. 64 seems not enough to replicate the original outputs. 512 is almost identical to 256 with Dev, but with Schnell rank 512 does a much better job on solid colors, patterns, gradients, and details - here's an example of 256 vs 512. Prompt: solid gray, 256 outputs noise image/png

Downloads last month

-

Downloads are not tracked for this model. How to track
Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support

Model tree for vafipas663/flux-krea-extracted-lora

Adapter
(4)
this model