Civitai Workflow for Flux: Fine-Tuning and Upscaling with LoRA Support

Enhance image generation through a two-pass process, incorporating fine-tuning and upscaling, with seamless LoRA integration.

Workflow Steps:

Initial Setup:

• Ensure ComfyUI is installed and updated to the latest version to support the required model components.

Model Preparation:

• Load the ae.safetensors model for VAE.

• Load the clip_l.safetensors for CLIP processing.

• Load the T5 text encoder.

• Choose between Schnell or Dev based on your needs

Image Generation:

• First Pass Generation: This is where the base output is created based on your prompts and settings.

• LoRA Support: You can add up to 3 LoRAs

Fine-Tuning (Second Pass):

• Refinement is optional. This is a second pass to refine details and enhance quality.

Upscaling:

• Apply Upscaling Models: Use a suitable upscaling technique to enhance the resolution of the fine-tuned images