+

DOCUMENTATION

Everything you need to get started.

A complete guide to setting up, configuring, and using VisionForge for real-time image analysis.

Quick Start

1

Install dependencies

VisionForge is built with TanStack Start, React, and Tailwind CSS. Clone the repository and install all required packages.

bun install
2

Configure your API key

Create a .env file in the project root and add your NVIDIA API key. This is required for the vision model to process images.

VITE_NVIDIA_API_KEY=nvapi-...
3

Start the dev server

Run the development server and navigate to the Analysis page to begin uploading and scanning images.

bun run dev

Analyzing an Image

1. Upload

Drag and drop an image or click to browse. Supports JPG, PNG, and WebP formats up to 5MB.

2. Scan

Click the Scan button to send the image to the NVIDIA Nemotron vision model for AI-generated content analysis.

3. Result

View the AI-generated percentage, reasoning, and confidence breakdown displayed in the results panel.

API Reference

POST/api/analyze

Sends an image to the NVIDIA Nemotron vision model and returns an analysis of whether the image is AI-generated.

Request Body

{ "image": "data:image/png;base64,..." }

Response

{ "result": "80%" }

Configuration

Environment Variables

VITE_NVIDIA_API_KEY — Your NVIDIA API key for the Nemotron vision model. Required for image analysis.

Model Parameters

Temperature: 0.6, Top-p: 0.95, Max tokens: 65536. These are tuned for balanced accuracy and detail in analysis output.

Upload Limits

Accepted formats: JPG, PNG, WebP. Maximum file size: 5MB. Images are converted to base64 data URLs before being sent to the API.

Privacy

All processing happens on your local infrastructure. Images are processed in-memory and immediately discarded. No data is stored.

Tech Stack

React

UI framework

TanStack Start

Full-stack framework

Tailwind CSS

Styling

NVIDIA Nemotron

Vision model

Three.js

3D rendering

Axios

HTTP client