search

Young98CN / LoRA_Composer

LoRA-Composer: Leveraging Low-Rank Adaptation for Multi-Concept Customization in Training-Free Diffusion Models
Other
41 stars 3 forks source link

A <ckc>character holding a <wpe>weapon while driving a <mror>motorbike #6

Open Sundragon1993 opened 3 weeks ago

Sundragon1993 commented 3 weeks ago

Dear team, thank you for the great contribution to the community, I'm wondering if this algorithm could be applied to train a Lora(s) that combines several different concepts like a character holding a gun while driving a motorbike? I have several images depicting each concept separately, but I couldn't find any experiments in your paper that elaborate on how the model handles interactions between such multi-concept scenarios. Could you provide further insights on this?

Young98CN commented 4 days ago

Apologies for the delayed response—I’ve been caught up with a tight deadline. We’ve included an example in the code that demonstrates the interaction between two persons.

Additionally, we recommend using more refined image conditions (human pose or canny edge) to control the model, as this can potentially lead to improved results.

Thank you for your patience, and feel free to reach out if you have further questions.