【FLUX】IMG to IMG — v6.2

【FLUX】IMG to IMG — v6.2 is a ComfyUI workflow from UmeAiRT for Flux.1 D image-to-image generation: provide a prompt and input image to create images, with optional tiled upscaling.

The workflow uses eight nodes and includes a built-in SeedVR2 tiled upscaler that can be turned on or off.

Saved images include generation parameters as embedded metadata for online publishing and remixing.

Manual setup lists flux1-dev-fp8.safetensors or flux1-dev-fp16.safetensors for the base version, and Q8, Q6, Q5, Q4, or Q3 model and CLIP files for the GGUF version.

Common manual files include a CLIP file, text encoder, and VAE, with examples such as clip_l.safetensors, t5xxl_fp8_e4m3fn.safetensors, and ae.safetensors.

Enable “Use Nodes V2 (Vue)” in ComfyUI, refresh the page, and then load the workflow; otherwise it may have display problems.

Install ComfyUI-UmeAiRT-Toolkit through ComfyUI Manager by searching “UmeAiRT,” or use the UmeAiRT Auto-Installer.

Listed model files include 4x_NMKD-Siax_200k.pth, ae.safetensors, clip_l.safetensors, flux1-dev-fp8.safetensors, flux1-dev-Q5_K_S.gguf, svdq-int4_r32-flux.1-dev.safetensors, t5-v1_1-xxl-encoder-Q8_0.gguf, t5xxl_fp8_e4m3fn.safetensors, and ume_sky_v2.safetensors.

Listed node-pack names include ComfyUI-GGUF, ComfyUI-nunchaku, comfyui-image-saver, comfyui-kjnodes, comfyui-mxtoolkit, comfyui_ultimatesdupscale, rgthree-comfy, and was-node-suite-comfyui.

Use the three LoRA slots with individual on/off toggles and strength controls.

The auto version can download models automatically.

Before downloading, verify that the page shows “【FLUX】IMG to IMG — v6.2” from UmeAiRT and version v6.2.

Suggestion · not verified

Before loading, confirm that your chosen setup has the exact model files and node-pack names listed for it.

Suggestion · not verified