A user on Reddit is seeking advice on how to maintain facial identity consistency in AI-generated videos using Stable Diffusion's Wan2.2 model. They are experiencing identity drift and are exploring the effectiveness of training a character LoRA for Wan2.2. The user is also asking for guidance on what constitutes a good set of training images for strong facial identity and what cosine similarity metric to aim for to avoid identity drift. AI
Summary written by gemini-2.5-flash-lite from 1 sources. How we write summaries →
RANK_REASON User-generated content on a forum discussing technical aspects of a specific AI model's application, not a formal release or significant industry event.