Introduction
This workflow combines PuLID Flux with ComfyUI to achieve precise facial learning and integration, tailored for creating Portraits inspired by paper-cut art. By blending photorealistic details with the bold and minimalistic aesthetic of paper-cutting, it ensures professional-grade, visually unique results.
PuLID Flux
PuLID Flux is a powerful module for refining textures, lighting, and details. It excels at blending intricate features, making it ideal for achieving the layered, bold aesthetics of paper-cut art while preserving natural facial details.
Read more and download: https://github.com/balazik/ComfyUI-PuLID-Flux.git
Workflow Overview
How to use this workflow?
Step 1: Upload Face Reference Image
Upload a clear reference image into the Load Image module.
Ensure it is a frontal view for accurate facial learning. If the reference is a side profile, the resulting output will align with this perspective.
Step 2: Input Text Prompts
Enter the LoRA trigger word AJFJZ in the Positive Prompt field in CLIP Text Encode.This word is mandatory for activating the pre-trained LoRA model.
Step 3: Perform Secondary Face Learning
Activate the Face Analysis module to conduct a second stage of facial learning. All parameters are preset for optimal performance and do not require modification.
Step 4: Initial Sampling with KSampler
Preconfigured settings, including denoise, ensure sharp and high-quality results. Avoid modifying denoise, as it directly affects image clarity; any change might lead to blurry results.
Step 5: Export the Final Image
Finalize the process by exporting the edited image through the Save Image node.
Additional Notes
- Version Restriction: This workflow is only compatible with ComfyUI version 0.3.6 or earlier.
- Parameter Optimization: All settings in the workflow, including LoRA weight and sampling parameters, are already tuned for best performance.
- Use Case: Ideal for face matching, style customization, paper-cut inspired designs, and photorealistic image generation.