AnalysisVisual AIJuly 7, 2026
Krea 2 Identity Edit LoRA enables instruction-based image editing

The LoRA targets Krea 2's 12.9B MMDiT model for identity-preserving edits given an image and plain-language instruction. Available on Hugging Face.

The LoRA targets Krea 2's 12.9B MMDiT model for identity-preserving edits given an image and plain-language instruction. Available on Hugging Face.