This FLUX Character Prompt Workflow is designed for generating and refining AI-driven art and character prompts in a visually structured manner. The process involves:

1. Input and Model Loading:

   • Images or text prompts are imported via nodes like LoadImage and DF_Text_Box.

   • Models, such as Florence2ModelLoader and UNETLoader, handle AI functionality for generating and refining outputs.

2. Prompt Creation and Conditioning:

   • Text prompts are created or modified using tools like Text Find and Replace or Prompt Selector.

   • Conditioning nodes (ConditioningConcat, CLIPTextEncodeFlux) refine prompts or merge details for more specific outputs.

3. Image Processing and Refinement:

   • Nodes like SamplerCustomAdvanced, RandomNoise, and VAEEncode process images with input conditions, creating latent images.

   • Outputs are previewed with tools like PreviewImage or compared with Image Comparer nodes.

4. Upscaling and Enhancements:

   • Advanced tools like UltimateSDUpscale and FaceDetailer improve image resolution and add finer details, such as realistic lighting and textures.

5. Final Output:

   • Enhanced images are saved with metadata using the Save Image w/Metadata node.

   • Outputs are previewed and compared for quality control.

6. Customizability:

   • Notes provide guidance for setting prompt types, adjusting parameters, or fixing potential issues in the workflow.

7. Put this in your comfy directory\ComfyUI_windows_portable\ComfyUI\models\LLavacheckpoints

   https://huggingface.co/bartowski/NemoMix-Unleashed-12B-GGUF/blob/main/NemoMix-Unleashed-12B-IQ4_XS.gguf?language=python

Install this in your comfy directory\ComfyUI_windows_portable\ComfyUI\models\LLM

git clone https://huggingface.co/microsoft/Florence-2-large