By：Reverent Elusarca
Workflow source: https://openart.ai/workflows/reverentelusarca/flux-consistent-character-sheet/oSEKBwDLvkt9rHMfdU1b
This workflow uses SDXL or sd1.5 models as base image generation, with ControlNet Pose and IPAdapter as styles. Since Flux does not yet support ControlNet and IPAdapte, this is the current approach.
The IPAdapter can be bypassed.
Your ControlNet pose reference image should look like this workflow.
There are two CLIP positive inputs, and they should be the same.
You can use the Upscaler node for better resolution.