Jibs double Turbo ZIB to ZIT Workflow — v1.0

Use Jibs double Turbo ZIB to ZIT Workflow to generate 1024×102 images; its graph has text-entry and image-loading nodes, plus image-preview and image-saving nodes.

The graph includes ControlNet, sampling, VAE decoding, and upscaling nodes.

Textbox, Text to Conditioning, and LoadImage nodes provide text-entry and image-loading points; PreviewImage and Image Saver nodes provide image outputs.

This workflow is v1.0 by J1B and names ZImageBase as its base model.

Named ComfyUI node packs: chibi, comfyroll, ComfyUI_Comfyroll_CustomNodes, comfyui_controlnet_aux, comfyui_ultimatesdupscale, comfyui-custom-scripts; comfyui-easy-use, comfyui-image-saver, mikey_nodes, RES4LYF, rgthree-comfy, and was-node-suite-comfyui.

Named model files include Jib_Mix_ZIT_1_L_00001_.safetensors, qwen_2.5_vl_7b_fp8_scaled.safetensors, Ultra_flux_For_Z-image-vae.safetensors, and x1_ITF_SkinDiffDetail_Lite_v1.pth.

Other named model files are z_image_bf16.safetensors, Z-image-ae.safetensors, and Z-Image-Turbo-Fun-Controlnet-Tile-2.1-8steps.safetensors.

The description links to a [VAE Tech Merge Node](https://civitai.com/models/2231351?modelVersionId=2638152) and a [ZIB 8 step lora](https://civitai.com/models/2362961?modelVersionId=2657506).

The description pairs “decent,” “good,” and “really good” image labels with 12, 30, and 60 seconds; those timings have not been tested here.

Before running it, check that your ComfyUI setup has the named node packs and model files.

Suggestion · non vérifié