Flux Kontext workflow - detailers, IPAdapter, upscale, all optional

This Flux.1 Kontext workflow uses loaded pictures and a text prompt to generate images, with optional detailers, IPAdapter, and upscaling.

Load pictures in Init, select the diffusion model, CLIP, VAE, and latent size in Loader, then enter a text prompt for generation.

It uses a first pass without upscaling, followed by an upscale stage set by default to 2x with 2x2 tiling.

Inputs include loaded pictures, a text prompt, model and sampler settings, a noise seed, scheduler, steps, and optional conditioning controls.

Outputs include a generated first-pass image and an upscaled generation; SaveImage is listed for saving images.

To use it, make the listed model files and node packs available, then select the diffusion model, CLIP, VAE, and latent size in Loader.

Vorschlag · nicht geprüft

Listed node packs are comfy-image-saver, comfyui_ultimatesdupscale, ComfyUI-Crystools, comfyui-detail-daemon, comfyui-easy-use, ComfyUI-nunchaku, and comfyui-wd14-tagger.

Listed model files are 4x-UltraSharp.pth, ae.safetensors, clip_l.safetensors, clip_vision_h.safetensors, Dark_Ghibli.safetensors, flux1-kontext-dev.safetensors, kontext-turnaround-sheet-v1.safetensors, svdq-int4_r32-flux.1-kontext-dev.safetensors, and t5xxl_fp16.safetensors.

Switches can enable or disable model sampling, LoRAs, detailers, upscaling, automatic prompt tagging, CLIP Vision/unCLIP conditioning, and IPAdapter.

Sampler settings expose the noise seed, sampler, scheduler, and steps; the Conditioning group includes FluxGuidance.

Verify that all listed model files and node packs are available in your setup before using the workflow.

Vorschlag · nicht geprüft

CLIP Vision and IPAdapter behavior has not yet been tested with Flux Kontext, so verify those options for your use before relying on them.

Vorschlag · nicht geprüft