Flux2 Klein 9B Consis Lora

Trained on the 9B dataset. This LoRA is designed to maintain image consistency across multiple input images. The LoRA effect can be controlled via the LoRA weight.

Recommended weight: 0.2–1.5

Can be used with any prompt.

Higher strength provides greater stability, but will also reduce editability.

Different images may require different optimal LoRA weights.