comfyui-vto

ComfyUI Virtual Try-On Node (Gemini)

A ComfyUI custom node for photorealistic virtual try-on generation using Google's Gemini API.

Features

• Model Selection: Choose between gemini-3-pro-image-preview and gemini-3.1-flash-image-preview
• Flexible Image Inputs: Up to 12 image input slots with smart placeholder system
• Editable System Prompt: Fully customizable prompt with {image_1} to {image_12} placeholders
• Multiple Aspect Ratios: 1:1, 3:4, 4:3, 2:3, 3:2, 4:5, 5:4, 9:16, 16:9, 21:9
• Resolution Control: 1K, 2K, or 4K output
• Temperature Control: Adjust creativity vs consistency (0.0 - 2.0)
• Batch Generation: Generate 1-4 images per execution
• Seed Control: Reproducible results with seed parameter (0 = random)
• Graceful Input Handling: Unconnected image inputs are automatically skipped
• Optional Overrides: Fine-tune pose, camera angle, and add custom prompts

Installation

1. Install the Node

Clone directly into your ComfyUI custom nodes directory:

cd ComfyUI/custom_nodes/
git clone https://github.com/YOUR_USERNAME/comfyui-vto.git

Or manually copy the comfyui-vto folder into ComfyUI/custom_nodes/.

2. Install Dependencies

cd ComfyUI/custom_nodes/comfyui-vto
pip install -r requirements.txt

3. Configure API Key

cp .env.example .env

Edit .env and add your Gemini API key:

GEMINI_API_KEY=your_actual_api_key_here

Get your API key from: https://aistudio.google.com/app/apikey

4. Restart ComfyUI

Restart ComfyUI to load the new node.

Usage

Quick Start

1. Search for "Virtual Try-On (Gemini)" in the node browser
2. Select your model from the dropdown
3. Connect your images to the input slots
4. Set aspect ratio, resolution, temperature, and batch size
5. Execute the workflow

Default Image Layout

The default system prompt expects images in this order:

| Input | Role | |-------|------| | image_1, image_2 | Identity references (face, body, features) | | image_3 | Pose and scene reference | | image_4 - image_12 | Clothing and accessory references |

Only connected inputs are sent to the API. Unconnected slots are automatically skipped.

Custom Prompt

You can rewrite the system prompt to change how images are used. Reference images with {image_1} through {image_12} placeholders:

Use {image_1} for identity, {image_2} for pose, and {image_3} to {image_6} for clothing.

Parameters

| Parameter | Type | Default | Description | |-----------|------|---------|-------------| | model | Dropdown | gemini-3.1-flash-image-preview | Gemini model to use | | system_prompt | String | VTO prompt | Main instruction with image placeholders | | aspect_ratio | Dropdown | 3:4 | Output aspect ratio | | image_size | Dropdown | 4K | Output resolution (1K, 2K, 4K) | | batch_size | Int | 1 | Images to generate per run (1-4) | | seed | Int | 0 | Seed for reproducibility (0 = random) | | temperature | Float | 1.0 | Creativity control (0.0 = deterministic, 2.0 = max variation) | | image_1 - image_12 | Image | Optional | Image inputs | | human_pose_override | String | Empty | Override pose from text description | | camera_position_override | String | Empty | Override camera angle from text description | | additional_prompt | String | Empty | Extra context or instructions | | project_name | String | Empty | Project identifier (for reference) |

Tips

• Identity: Use 2+ high-quality photos of the same person for best likeness
• Pose: Use a clear, well-lit photo with the desired body position
• Clothing: Multiple angles of clothing items improve accuracy
• Temperature: Lower values (0.2-0.5) for consistent production runs, higher (1.0-1.5) for creative exploration
• Batch + Seed: Use batch_size > 1 with a fixed seed to get variations

Troubleshooting

"GEMINI_API_KEY not found"

• Ensure .env file exists in the comfyui-vto folder
• Check that the API key is correctly formatted
• Restart ComfyUI after creating/editing .env

"Image generation is not available in your country"

• Some Gemini models have regional restrictions
• Try switching to a different model in the dropdown
• Check Google's available regions

"No images connected"

• Connect at least one image input before generating

Import Errors

• Run pip install -r requirements.txt in the comfyui-vto folder
• Ensure you're using the same Python environment as ComfyUI

API Costs

This node uses Google's Gemini API. Check current pricing at: https://ai.google.dev/pricing

License

MIT License - See LICENSE file for details.