Z_Image_turbo - Face detailer - Upscaler - Workflow

This ZImageTurbo workflow is for face detailing and image upscaling. It takes text and an empty latent, then decodes images for preview and saving.

The 22-node graph includes sampling, face detailing, sharpening, model-based upscaling, film grain, resolution control, preview, and image saving.

The listed starting nodes are EmptyLatentImage and CLIPTextEncode.

The graph lists KSampler, VAEDecode, ImageUpscaleWithModel, ImageSharpen, PreviewImage, and SaveImage for sampling, decoding, upscaling, sharpening, previewing, and saving.

The listing identifies ZImageTurbo as the base model and Hikarias as the publisher.

The listing identifies v2.0 as the version with fixed nodes.

Listed node packs are comfyui_ultimatesdupscale, comfyui-workflow-encrypt, controlaltai-nodes, gguf, RES4LYF, and rgthree-comfy.

Listed model files are 4x_NMKD-Siax_200k.pth, qwen_3_4b.safetensors, Qwen3-4B-abliterated.Q8_0.gguf, z_image_turbo_bf16.safetensors, and zImageTurbo_vae.safetensors.

The listed pins are comfyui_ultimatesdupscale 1.6.0, comfyui-workflow-encrypt 1.0.0, gguf 2.7.3, and rgthree-comfy 1.0.2511270846.

The listed pins for controlaltai-nodes and RES4LYF are 721492b66c9cede8ae23ae10615462ad80cfd061 and 46de917234f9fef3f2ab411c41e07aa3c633f4f7.

Before using it, verify that all six listed node packs and five listed model files are available in your ComfyUI setup.

Sugestão · não verificado

If you choose the V1.0 Reactor version, back up your VENV folder first; the listing warns that some installations may have dependency issues that could break ComfyUI.

Sugestão · não verificado