🎭 Flux Redux Instant Consistent Characters ComfyUI Workflow (No LoRA)

Hey Black Mixture Fam!

We're excited to present our latest Flux Consistent Character workflow, combining Flux Redux with PuLID and Florence2 auto-captioning. This workflow lets you generate character images by combining any reference image with any face - no LoRA training required!

💾 Workflow Versions:

• With Auto-Downloader: Florence2 installs automatically on first run

• Standard: For users who already have Florence2

📦 Required Models:

• Flux Redux Model:

  • Download from: Black Forest Labs

  • Place in: 📂ComfyUI/models/style_models

• CLIP Vision Model:

  • Download from: Google SigLIP

  • Place in: 📂ComfyUI/models/clip_vision

• Diffusion Model:

  • Download from: FLUX.1 Dev

  • Place in: 📂ComfyUI/models/diffusion_models

• CLIP Model:

  • Download from: Flux Text Encoders

  • Place in: 📂ComfyUI/models/clip

• Upscale Model:

  • 4x Foolhardy Remacri

  • Download from: OpenModelDB

  • Place in: 📂ComfyUI/models/upscale_models

✨ Features:

• Instant character style transfer without LoRA training using Flux Redux + PuLID

• Combines any reference image with any face reference

• Integrated PuLID for detail preservation and character consistency

• Automatic prompt generation using Florence2 AI - no manual prompting needed

• Smart image understanding that captures key character details

• Professional upscaling

• Multi-stage processing using Addon Picker

🎨 Workflow Sections:

🟪 IMPORTANT NOTES: Download links and instructions
🟥 LOAD MODELS + IMAGES
🟨 PROMPT: Automatic Florence2 caption generation
🟦 PULID + GENERATION SETTINGS: Face integration and generation controls 🟩 UPSCALE: High-quality image enhancement processing

⚙️ How It Works:

1. Load a Reference Image:

• This will determine the overall style, pose, and composition

• Florence2 automatically generates appropriate prompts from this image

2. Load a Face Reference:

• This can be any face image you want to incorporate

• The workflow will integrate this face into the new generation

3. Generate:

• The workflow combines both references to create a new image

• Maintains the style and pose of the first reference

• Incorporates the facial features from your face reference

• Upscales 2x after generation

• All without needing to train or load any LoRAs!

🎯 Best Practices:

• Use clear, well-lit reference images

• Choose reference images with similar angles for best results

• Face references should be clear and forward-facing when possible

• Experiment with different reference combinations to find what works best

⚙️ Quick Start:

1. Update ComfyUI to the latest version

2. Download and install all required models in their specified directories:

3. Choose your workflow version:

• Autodownload if you want automatic Florence2 installation

• Standard if you already have Florence2 installed

4. Import your chosen workflow file into ComfyUI

5. Install any missing custom nodes when prompted by ComfyUI

6. Load your reference image - Florence2 will automatically generate the appropriate prompts

7. Load your face reference image

8. Adjust generation parameters if desired (no prompt writing needed!)

9. Generate and enjoy your consistent character creations!

💡 Important Notes:

• ComfyUI will prompt you to install any missing custom nodes on workflow import

• Face reference images work best when they are clear, front-facing portraits

📹 Tutorial & Support:

• Check out our YouTube channel: Black Mixture

• Join our Discord: Black Mixture Discord

• Access our central hub: Resource Directory

Created by Nate & Chriselle Dwarika from The Black Mixture Team

💝 HUGE Thanks to all of our amazing Patreon supporters and Happy Holidays!