A powerful Image2Image workflow built in ComfyUI using the Furious Mix by NukeAI v4 checkpoint, designed to generate high-quality, anime gym and fitness imagery from a reference image.

This workflow uses dual CLIP text encoding (positive + negative), a dedicated VAE Encode → KSampler → VAE Decode pipeline, and the Book-Tools Text to Image node for prompt rendering. ClearVAE is loaded separately for clean latent decoding.

Perfect for: fitness content creators, gym branding, athletic stock imagery, and motivational visual content.

Upload a reference image of a person or gym space — the workflow will repaint it with the positive prompt style

Lower denoise to 0.6–0.8 if you want to preserve more of the original image structure\

Swap the checkpoint for an anime model and adjust the prompts to use this same pipeline for any style

The negative prompt is heavily tuned to suppress bad anatomy, extra fingers, and cartoon artifacts — keep it as-is for best results