A user is experiencing inconsistent results when using a custom Identity LoRA trained on 66 images within StableDiffusion via ComfyUI. The generated faces sometimes match the trained identity and other times deviate significantly. The user is seeking advice on how to improve facial consistency, considering options like specific prompting, adjusting LoRA strength, or retraining the LoRA. AI
IMPACT This issue highlights potential challenges in fine-tuning and applying custom models for specific identities within generative AI tools.
RANK_REASON User-generated content about a specific tool's functionality.
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →