tf-Flux2-Klein-9B-Consis-LoRA - 保持图像一致性_v2.safetensors
V1.0
tf-Flux2-Klein-9B-Consis-LoRA - 保持图像一致性_v2.safetensors
377 Views0 Likes5 Favorites
Model Information
Active
Model Type：
LoRA
Basic Model：
Flux2-Klein-9B
Trigger Words：
Flux2 Klein-9B Consis-Lora 保持图像一致性_v2 Klein Klein9B tf
Resource Name：
models/loras/tf-Flux2-Klein-9B-Consis-Lora-保持图像一致性_v2.safetensors
🧱 Base Model Foundation: FLUX.2 [klein] 9B (Flux2 Klein 9B Backbone DiT Model)
🎯 Key Features
This is a dedicated image consistency LoRA specifically trained for Flux2 Klein 9B, requiring no trigger words. It primarily solves image drift issues in image-to-image, Inpaint, Outpaint, and multi-shot continuous editing:
- Constrains the model during image editing to preserve the original character's facial features, object structures, colors, composition, and texture details, reducing image distortion, facial warping, and object shifting caused by multiple iterations;
- Suitable for multi-frame storyboards, cross-shot editing of the same character, multi-iteration retouching, and local Inpaint/Outpaint Outpaint workflows;
- Supports combination with various style LoRA and ControlNet for simultaneous use, adapting to the full Flux2 Klein9B image editing pipeline;
- Version v2 optimizes the high-low frequency color loss algorithm. Compared to the preview version, it balances image consistency with editing capabilities, reducing dull colors and blurred details.
🧱 Base Model Foundation: FLUX.2 [klein] 9B (Flux2 Klein 9B Backbone DiT Model)
🎯 Key Features
This is a dedicated image consistency LoRA specifically trained for Flux2 Klein 9B, requiring no trigger words. It primarily solves image drift issues in image-to-image, Inpaint, Outpaint, and multi-shot continuous editing:
- Constrains the model during image editing to preserve the original character's facial features, object structures, colors, composition, and texture details, reducing image distortion, facial warping, and object shifting caused by multiple iterations;
- Suitable for multi-frame storyboards, cross-shot editing of the same character, multi-iteration retouching, and local Inpaint/Outpaint Outpaint workflows;
- Supports combination with various style LoRA and ControlNet for simultaneous use, adapting to the full Flux2 Klein9B image editing pipeline;
- Version v2 optimizes the high-low frequency color loss algorithm. Compared to the preview version, it balances image consistency with editing capabilities, reducing dull colors and blurred details.