SIMPLE FLUX IMG2IMG WORKFLOW by weshouldhaveshotguns — v1.0

Use this ComfyUI workflow to transform a source image with a text prompt through Flux img2img and save the resulting image; it is designed for standard Flux1.dev models.

The ComfyUI graph names nodes for loading and scaling the source image, encoding image and text, applying Flux guidance, sampling, decoding, and saving.

Input: a source image and a text prompt for the img2img transformation.

Output: an image saved with SaveImage.

The listed base model is Flux.1 D, and the workflow is designed for standard Flux1.dev models.

The description includes links to the files needed to start.

Model files: ae.safetensors (VAE), clip_l.safetensors (text encoder), flux1-dev.safetensors (UNet), and t5xxl_fp8_e4m3fn.safetensors (text encoder).

Named ComfyUI nodes include DualCLIPLoader, KSamplerSelect, FluxGuidance, UNETLoader, RandomNoise, Power Lora Loader (rgthree), VAEEncode, ImageScaleToTotalPixels, and LoadImage.

Other named ComfyUI nodes are BasicGuider, BasicScheduler, VAELoader, VAEDecode, SamplerCustomAdvanced, CLIPTextEncode, and SaveImage.

Use the img2img settings guide to adjust essential settings and understand how each component affects the output.

Sugestão · não verificado

Use the prompting tips when changing prompts to help preserve key elements of the source image.

Sugestão · não verificado

Before you start, verify that ae.safetensors, clip_l.safetensors, flux1-dev.safetensors, and t5xxl_fp8_e4m3fn.safetensors are available in ComfyUI.

Sugestão · não verificado

Verify that ComfyUI has the named nodes, including DualCLIPLoader, FluxGuidance, Power Lora Loader (rgthree), and SamplerCustomAdvanced.

Sugestão · não verificado