Flux Fill outpaint with GGUF

Use Flux Fill outpaint with GGUF to outpaint an image with Flux.1 D; it names LoadImage for the image input and SaveImage for the output.

The workflow lists these required node packs: ComfyUI-Custom-Scripts, ComfyUI-Miaoshouai-Tagger, ComfyMath, comfyui-various, was-node-suite-comfyui, and ComfyUI_essentials.

The named image input is LoadImage, and the named output is SaveImage.

Guidance: download the GGUF model from https://huggingface.co/YarvixPA/FLUX.1-Fill-dev-gguf before use.

Suggestion · non vérifié

Model files named for the workflow are flux1-fill-dev-Q5_0.gguf, Flux/ae.safetensors, Flux/clip_l.safetensors, Flux/t5xxl_fp16.safetensors, and 4x_ESRGAN-NMKD-Siax-4x.pth.

Named node types include ImageUpscaleWithModel, ImagePadForOutpaint, ImageResizeKJ, CLIPTextEncodeFlux, MaskPreview+, UpscaleModelLoader, VAELoader, Text Multiline, JWMaskResize, LoadImage, and InpaintModelConditioning.

Other named node types are DifferentialDiffusion, Miaoshouai_Tagger, KSampler, CM_NearestSDXLExtendedResolution, UnetLoaderGGUF, SaveImage, ColorMatch, DualCLIPLoader, GetImageSize+, VAEDecode, ShowText|pysssss, and ImageCompositeMasked.

It is listed with Flux.1 D as its base model, and its title specifies GGUF.

Guidance: check that the named model files and nodes are available in your setup before trying the workflow.

Suggestion · non vérifié

Guidance: check performance on your setup; the publisher reports that Flux struggled on their Mac M1 and GGUF worked slowly.

Suggestion · non vérifié