comfyui-character-gen

To maintain face preservation in ComfyUI, you can implement identity-preserving workflows using methods like PuLID, InstantID, or IP-Adapter. This Skill provides structured templates and parameter recommendations to ensure facial consistency across generated images.

The best way to handle cross-domain character conversion is by selecting an appropriate identity method like InfiniteYou or FLUX Kontext. This Skill applies selection logic to choose the optimal method for transformations like 3D to photo conversions.

Multi-reference consistency is achieved by using identity-preserving workflows that support multiple input images. This Skill helps define safe defaults and templates to reliably maintain character identity across various references and iterative edits.

Yes, you can use LoRA training with identity-preserving workflows. This Skill includes hooks for LoRA training and model switching, allowing you to build pipelines that cover both zero-shot and LoRA-based character generation scenarios.

Choosing InstantID over IP-Adapter depends on your specific input scenario and desired artifacts. This Skill provides method selection logic to evaluate your use case and recommend the most suitable identity preservation approach for your workflow.

To avoid common artifacts in iterative editing pipelines, this Skill provides safe defaults and parameter recommendations. It safeguards your workflow against artifacts while maintaining face preservation during iterative image edits.