ComfyUI users are debating the best way to keep characters consistent in Illustrious scenes
sparereddit1234 · reddit · 2026-07-25
A user asks how to keep character styling consistent when generating adult VN CGs with Illustrious in ComfyUI.
They describe a workflow where ControlNet handles pose and framing well, but IPAdapter struggles to preserve the exact character look. They’re considering regional masking, separate prompts and reference images per character, a shared ControlNet pass for the whole scene, and possibly training a LoRA per character as the long-term solution.
More from coding & agent
- Why many agents fail because they cannot tell whether they are making progress — orvi2014 · 2026-07-25
- A developer is building one workspace to debug multi-step AI workflows — Impressive-Iron5216 · 2026-07-25
- Laguna S 2.1 says users want open source, simpler agentic coding stacks — max_paperclips · 2026-07-25
- AI coding agents won’t replace engineers who still know how to design software — bendee983 · 2026-07-25
- ComfyUI users want a simpler XY grid workflow for testing LoRA strength and versions — TheWebbster · 2026-07-25
- Startup building an agent execution layer is hiring core runtime, cloud infra and DX engineers — dotey · 2026-07-25