Rig / Setup: I built a reusable ComfyUI 'character slot' so swapping characters doesn't mean rebuilding the whole graph
Sharing an organizational setup that saved my sanity across a multi-character project. If you've ever rebuilt your whole node graph just to render a different character, this is for you.
The 'character slot' pattern:
Built one master graph with a single swappable CHARACTER group node that takes: a LoRA, an InstantID reference image, and a trigger-word string.
Each character is saved as a tiny preset (just those three things). Swapping from Character A to Character B is loading a preset, not rewiring anything.
The rest of the graph (pose control, lighting, upscale, motion prep) is character-agnostic and never changes.
Why it matters for consistency specifically: because every character goes through the IDENTICAL downstream pipeline, they all sit in the same 'world' — same lighting logic, same render treatment, same grain. Characters built in different ad-hoc graphs look like they're from different movies even when each is individually consistent. One graph = one world.
It's basic software hygiene applied to an art pipeline but nobody does it and everyone suffers. Graph + the preset structure in comments.