ComfyUI Templates

Upload your custom ComfyUI workflows to Genvid and run them as managed serverless endpoints—no infrastructure management required.

Overview

ComfyUI Templates allow power users to bring their own ComfyUI workflows into Genvid. When you upload a workflow, Genvid automatically:

1. Analyzes the workflow to detect dependencies (custom nodes and models)
2. Builds a containerized version of your workflow
3. Deploys it as a serverless endpoint on high-performance GPUs
4. Makes it available for generation alongside other AI models

This means you can use your custom image and video generation pipelines within Genvid's production workflow without managing servers, containers, or GPU infrastructure.

Prerequisites

Before uploading a ComfyUI template, ensure you have:

• A valid ComfyUI workflow in API format (not the web UI format)
• Organization membership in Genvid
• Understanding of your workflow's requirements (GPU memory, custom nodes, models)

---

Uploading a Template

Step 1: Access ComfyUI Templates

1. Click your profile icon in the top-right corner of any Genvid page
2. Select ComfyUI Templates from the dropdown menu
3. The ComfyUI Templates page displays your organization's uploaded workflows

Step 2: Start Upload

1. Click the Upload Template button in the top-right corner
2. The Upload ComfyUI Template dialog opens

Step 3: Fill in Template Details

Step 4: Select GPU Tier

Choose the GPU tier that matches your workflow's memory requirements:

Step 5: Upload Workflow File

1. Click Choose Workflow File in the upload area
2. Select your ComfyUI workflow JSON file
3. Wait for validation to complete

Validation checks:

• File must be valid JSON
• Must contain ComfyUI API format structure (nodes with class_type fields)
• Cannot be empty

If validation succeeds, you'll see a green checkmark and the number of nodes detected.

Step 6: Create Template

1. Click Create Template
2. The template is created with "Pending" status
3. Deployment begins automatically in the background

---

Exporting in API Format

ComfyUI has two save formats—you need the API format for Genvid.

Enable Developer Mode

1. Open ComfyUI in your browser
2. Click the Settings icon (gear)
3. Find Enable Dev Mode Options and turn it on
4. Click Save Settings

Export Your Workflow

1. Design or load your workflow in ComfyUI
2. Click the Save button
3. Select Save (API Format) from the menu
4. Save the JSON file to your computer

---

Understanding Deployment Status

After uploading, your template progresses through several deployment stages:

Monitoring Progress

The templates list automatically refreshes every 10 seconds while templates are deploying. You can also:

• Click the Refresh button to manually update status
• Click on a template row to view detailed status information

---

Viewing Template Details

Click on any template in the list to see its full details, including:

Detected Dependencies

Custom Nodes - Third-party node packs detected in your workflow:

• Node name and source repository
• Resolution status (whether Genvid can automatically install it)

Models - AI models referenced in your workflow:

• Model filename and type (checkpoint, LoRA, VAE, etc.)
• Estimated file size

Configurable Parameters

The system extracts parameters from your workflow that can be modified at runtime:

• seed - Random seed for reproducibility
• steps - Number of generation steps
• cfg - Classifier-free guidance scale
• Other node-specific inputs

Configuration

GPU Tier - The selected GPU type for execution

Deployment Strategy:

• Baked Models - Models included in the container (faster startup, used for smaller total model size)
• Network Volume - Models loaded from shared storage (used when total model size exceeds 5GB)

---

Deleting a Template

1. From the templates list, click the Delete icon (trash can) on the template row

   OR

   Open the template details and click the Delete icon in the top-right corner

2. Confirm the deletion when prompted

---

Troubleshooting

Template Upload Fails

Template Stuck in Building

If a template stays in "Building" status for more than 15 minutes:

1. Check the template details for error messages
2. Large models (>10GB) may take longer to download
3. If stuck indefinitely, delete and re-upload the template

Template Deployment Failed

1. Click on the template to view the error message

2. Common causes:

   • Unresolvable custom nodes - The workflow uses custom nodes that cannot be found or installed
   • Model not available - A required model file cannot be downloaded
   • Resource limits - The workflow exceeds available resources

3. Fix the underlying issue and re-upload the workflow

Custom Node Not Working

If a workflow deploys but a custom node doesn't function:

1. Check if the node shows as "Unresolved" in template details
2. Verify the custom node is available in the ComfyUI-Manager registry
3. Try updating to a newer version of the custom node in your local ComfyUI, then re-export and re-upload

---

Best Practices

Optimize Your Workflows

1. Use standard models when possible - Common models like SDXL are pre-cached and deploy faster
2. Minimize custom nodes - Each custom node adds build time and potential compatibility issues
3. Test locally first - Verify your workflow runs correctly in local ComfyUI before uploading

Choose the Right GPU Tier

1. Start with RTX 4090 - Sufficient for most workflows
2. Monitor for OOM errors - If jobs fail with out-of-memory errors, upgrade to a higher tier
3. Consider model count - Multiple large models may require 48GB+ VRAM

Naming Conventions

Use descriptive names that help identify workflows:

• Include the main model (e.g., "SDXL Cinematic Portraits")
• Include the output type (e.g., "Video Upscaler 4K")
• Add version numbers for iterations (e.g., "Character Generator v2")

---

Related Features

• AI Models - Understanding AI models available in Genvid
• Video Editor - Using generated outputs in video workflows
• Budget Management - Tracking credit usage across templates