ComfyUI Flux Redux + PuLiD Workflow Consistency Face in Any Style

Ever wanted to create AI-generated art with perfect face consistency, no matter what style you’re going for? Well, today’s your lucky day! In this guide, we’ll walk you through how to use the ComfyUI Flux Redux and PulID workflow to make sure the face stays just right while applying any style—without losing those key facial features.

If you see that “Import Failed” error, we need to go a bit deeper. Here’s what you do:

1. Head to your ComfyUI custom node folder. In your ComfyUI directory, look for the pulid_flux folder.
2. Open the folder, open that folder in cmd
3. pip install -r requirements.txt

No need to panic—this one’s easy to fix too! First, you’ll want to check your Python version. For this workflow, you need Python 3.12. If you’re on Python 3.11 or earlier, you’ll need to grab the insideface module that fits your version.

You can download it from here: https://huggingface.co/hanamizuki-ai/insightface-releases/tree/main

Since I’m on Python 3.12, I downloaded the insideface file for that version. Once you’ve got it, paste it into your ComfyUI folder

Here’s how to install it:

1. Right-click the file and copy its path.
2. Open up the command line and type the install command at the top of the directory.

Next up, head to your Python embedded folder. open that Folder in cmd

python.exe -m pip install F:\2\ComfyUI_windows_portable\insightface-0.7.3-cp312-cp312-win_amd64.whl

This will install the insideface module. Once that’s done, restart ComfyUI to make sure everything’s good to go.

Error(s) in loading state_dict for CLIPVisionModelProjection:
size mismatch for vision_model.embeddings.patch_embedding.weight: copying a param with shape torch.Size([1152, 3, 14, 14]) from checkpoint, the shape in current model is torch.Size([1024, 3, 14, 14]).

• Just Update the ComfyUI

• pulid_flux_v0.9.0.safetensors goes in ComfyUI/models/pulid/.
• Download flux redux safetensors file from: https://huggingface.co/black-forest-labs/FLUX.1-Redux-dev (Place the redux model inside comfyui/models/style_models)
• Download the clip vision model from:  https://huggingface.co/google/siglip-so400m-patch14-384/tree/main
• Insightface-releases: https://huggingface.co/hanamizuki-ai/insightface-releases/tree/main