Skip to content

Template Library: One-Click Loading of Official Workflows and Automatic Download of Missing Models ​

This page explains how to use the ComfyUI workflow template library to pick templates by task: load official workflows with one click, automatically download missing models, and then generate images right away.

Preparation ​

  • Done First Generation: The Default Workflow Explained Node by Node, and ComfyUI starts up normally.
  • Sufficient disk space: open-source text-to-image templates download roughly 20 GB of models, and image editing templates need even more.
  • Network access to Hugging Face or Civitai—automatic download only supports safetensors files from these two sources.

Follow Along ​

Step 1: Open the ComfyUI Workflow Template Library and Pick a Template by Task ​

Launch ComfyUI, click the Templates icon in the sidebar, or open the workflow template library from the menu Workflow → Browse Workflow Templates, and browse by category and tag. For example, in the "Image" module of the "Basics" category, there are official templates such as "Z-Image-Turbo: Text to Image" (open source, supports Chinese and English), "Qwen Image 2.1: Image Editing" (character-consistent editing, alpha-channel cutout and compositing), and "Nano Banana 2.1: Text to Image" (Google's API image model).

Expected result: You can see each template's Chinese name, description, and tags (text to image, image editing, API, etc.); selecting one loads it onto the canvas.

Step 2: Load a Template and See Which Models Are Missing ​

Click a template card to load the workflow onto the canvas; nodes for models missing on your machine will be marked in red. You can view the list of missing models in the Errors tab, or open a model loader node and check the models list in the node's properties to see all models this template requires.

Expected result: The red nodes and the Errors tab list the missing model files; for example, loading Z-Image-Turbo for the first time will prompt that its model files are missing.

Step 3: Download Missing Models with One Click and Run ​

Choose automatic download in the missing-model prompt: click N errors found banner's View details, then in the Errors tab's Missing Models list in the right panel, click Download to download them one by one, or click Download all to get them all at once. Download sources are limited to safetensors files from Hugging Face and Civitai; once done, the red marks disappear and you can click Run to generate an image.

Expected result: The nodes return to their normal colors, and the workflow runs through and outputs an image.

Verification ​

  • No nodes are marked in red anymore, and the Errors tab is empty.
  • The Z-Image-Turbo template generates normally even with Chinese prompts—this model natively supports Chinese and English.
  • The output is comparable in style to the template thumbnail, showing that both the template and the models loaded correctly.

Common Pitfalls ​

  • API templates have no local models to download."Nano Banana 2.1: Text to Image" uses Google's Gemini-based API image model (not open source); it does not trigger a local model download and requires the corresponding API configuration to run.
  • **Automatic download has source restrictions.**Only safetensors from Hugging Face / Civitai are supported; models from other sources or in other formats must be downloaded manually and placed in the corresponding directory—see Model Loading Failures: Not Found, Unrecognized, Architecture Mismatch.
  • **A version that is too old cannot open new templates.**Templates specify a minimum ComfyUI version: Z-Image-Turbo requires 0.11.0, Qwen Image 2.1 image editing requires 0.37.0, and Nano Banana 2.1 requires 0.39.1—upgrade first if your version is lower, then load.
  • **Downloads are large, so check your disk first.**Z-Image-Turbo is about 20 GB and Qwen Image 2.1 about 26 GB; confirm you have enough space and a stable network before downloading.

Next Lesson ​

Once the template runs through, the next lesson covers LoRA Usage: Style Stacking and 4-Step Acceleration, layering your own style onto official templates. If you're not sure which model to choose, review Models × Tasks Overview: Pick the Right Model Before You Start.

gaodong.aiAboutPrivacyContact

Independent tutorials, not affiliated with any product mentioned. Product names and trademarks belong to their owners. Check official pages for prices and policies.