[flux-fill + flux-redux] Product Background Change

This workflow is for changing a product image’s background; it takes an image through LoadImage and writes the resulting image with SaveImage.

Its listed steps include Florence-2 and SAM2 processing, masking, Flux guidance, sampling, decoding, and image saving.

Input: an image loaded with LoadImage. Output: an image written by SaveImage.

The listing says you can run this workflow without setting up a ComfyUI environment, and that it is free.

The workflow lists these model files or models: ae.safetensors, clip_l.safetensors, flux1-fill-dev.safetensors, flux1-redux-dev.safetensors, microsoft/Florence-2-base, sam2_hiera_small.safetensors, sigclip_vision_patch14_384.safetensors, and t5xxl_fp16.safetensors.

Listed node types include CLIPTextEncode, CLIPVisionEncode, CLIPVisionLoader, DifferentialDiffusion, DownloadAndLoadFlorence2Model, DownloadAndLoadSAM2Model, DualCLIPLoader, easy showAnything, Florence2ModelLoader, Florence2Run, Florence2toCoordinates, FluxGuidance, and Image ‌​‌

Other listed node types are InpaintModelConditioning, JoinImageWithAlpha, KSampler, LoadImage, Mask Invert, MaskBlur+, MaskPreview+, Sam2Segmentation, SaveImage, StyleModelApply, StyleModelLoader, UNETLoader, VAEDecode, and VAELoader.

The listing identifies Flux.1 D as the base model.

Before using it, verify that the listed node types are available in your setup.

Vorschlag · nicht geprüft

Before use, verify that sam2_hiera_small.safetensors and sigclip_vision_patch14_384.safetensors are the expected files for your setup.

Vorschlag · nicht geprüft