FLUX EASY WORKFLOW [LOWVRAM] [GGUF]

Effective Workflow for GGUF Variants with LORA and Upscaling

Despite the release of Flux some time ago, finding a functioning workflow for the GGUF variants has proven challenging. I’m sharing this simple yet effective workflow that supports both LORA and upscaling.

For the FLUX-schnell model, ensure that the FluxGuidance Node is disabled.

Installation:

Use ComfyUI-Manager to install missing nodes: ComfyUI-Manager

Models used in workflow:

• FLUX GGUF: flux-gguf
  -> Place in: /ComfyUI/models/unet

• FLUX Text Encoders: flux_text_encoders

  -> Download  

  • t5xxl_fp8_e4m3fn.safetensors

  • clip_l.safetensors

  -> Place in: /ComfyUI/models/clip

• Upscaling Model: 4x-Ultrasharp
  -> Place in: /ComfyUI/models/upscale_models/

• Lora Models:

  • Flux Meets Midjourney v6.1 SDXL

  • Flux Image Upgrader: Detail Maximizer & Contrast Fix for Low CFG SDXL & SD 1.5

  -> Place in: /ComfyUI/models/loras/