The first Flux model, is a refined version of the official fp8 of Black Forest, using a subset of my showgirl photo collection (a small part) trained Lora. Previously released standalone Lora tests occasionally had watermarks, materials have been reorganized, all watermarks removed and retrained.

      The large model already integrates clip and VAE, you need to use my dedicated workflow to generate (use online lib or local, no need to download clip and VAE models separately), input image resolution, image batch, and prompt to generate.

      Local usePlease place it under the ComfyUI\models\Stable-diffusion directory.

      Since using Flux, can't go back, believe that with the power of Flux, in the future, it should dominate the open-source large models and compete with Midjourney.

      The flourishing series of photos (SD1.5) will no longer be updated. The prompts for generating images previously in the flourishing series can be copied and pasted into the workflow here as well. It's best to use prompts described in natural language.