Workflow - 3 different characters in one picture — v1.0

This workflow is for modifying an image with three different characters: it uses a loaded image and a preferred character reference image, applies PuLID Flux face detection with configurable options, and outputs a modified picture.

The 66-node workflow includes face detection, image concatenation, previews, comparisons, saving, image scaling, and custom upscaling.

The workflow includes a LoadImage stage and uses a preferred character reference image as inputs.

It includes PreviewImage, Image Comparer (rgthree), and comfy-image-saver stages for previewing, comparing, and saving image results.

Open the v1.0 workflow file named [Flux]_[PULID]_3-face.json.

Suggestion · non vérifié

The listed node packs are cg-use-everywhere, comfy-image-saver, comfyui_essentials_mb, comfyui_patches_ll, comfyui_pulid_flux_ll, comfyui_ultimatesdupscale, and comfyui-detail-daemon.

The remaining listed node packs are comfyui-easy-use, ComfyUI-GGUF, comfyui-kjnodes, and rgthree-comfy.

The listed model files are ae.safetensors, clip_l.safetensors, DAT_x4.pth, flux1-dev-Q5_K_S.gguf, pulid_flux_v0.9.1.safetensors, and t5-v1_1-xxl-encoder-Q5_K_S.gguf.

Use PulidFluxOptions to configure face-related options.

Suggestion · non vérifié

Detail Daemon is optional, so its group can be disabled.

Suggestion · non vérifié

Use 1024 px as the stated working width; for another size, calculate the setting separately.

Suggestion · non vérifié

Before setup, verify the exact file or package for ae.safetensors, DAT_x4.pth, pulid_flux_v0.9.1.safetensors, t5-v1_1-xxl-encoder-Q5_K_S.gguf, and ComfyUI-GGUF.

Suggestion · non vérifié

The workflow package does not include the six model files, so plan to provide them separately.

Suggestion · non vérifié