【QWEN】IMG to IMG — v1.1

【QWEN】IMG to IMG — v1.1 is a ComfyUI workflow for text-to-image and image-to-image generation: use a text prompt and loaded model, plus an image for image-to-image work, to create images saved with generation parameters in their metadata.

The eight-node workflow includes a Detail Refiner sampler, built-in SeedVR2 tiled upscaling, and image saving with generation parameters embedded in metadata.

The SeedVR2 upscaler can be switched on or off, and a Lightning LoRA is included.

Write a text prompt and load a model; for image-to-image use, also load a source image.

Output images are saved with all generation parameters embedded in their metadata.

In ComfyUI, turn on “Use Nodes V2 (Vue)” in Settings, refresh the page, and then load the workflow; without this interface, it may have display problems.

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

ComfyUI-UmeAiRT-Toolkit is described as packaging the upscaler, face detailer, and metadata saver.

For a manual base setup, the listed files are an fp8 or bf16 model in models\diffusion_models, qwen_2.5_vl_7b_fp8_scaled.safetensors in models\clip, qwen_image_vae.safetensors in models\vae, and either Qwen-Image-Lightning-8steps-V2.0.safetensors or Qwen-Image-Lightning-4steps-V2

Other listed files and node packs include Qwen_Image_Distill-Q8_0.gguf, Qwen2.5-VL-7B-Instruct-UD-Q4_K_S.gguf, qwen_image_fp8_e4m3fn.safetensors, 4x_NMKD-Siax_200k.pth, wan_2.1_vae.safetensors, rgthree-comfy, was-node-suite-comfyui, comfyui_ultimatesdupscale, comfyui-mxtoolkit, q

Each of the three LoRA slots has an on/off toggle and strength control, and additional LoRA modules can be connected.

SeedVR2 tiled upscaling is optional, and the workflow includes a Detail Refiner sampler.

The instructions use separate base and GGUF paths; verify the model filenames and node-pack names against your ComfyUI installation before use.

Suggestion · non vérifié

No license is stated for this workflow, so check the usage terms before publishing or remixing images.

Suggestion · non vérifié