Pixlift AI Logo
PixliftAI
Docs

Documentation

Learn how to upscale images and enhance photos with Pixlift AI. Use local desktop processing or connect your Google Colab GPU runtime to the web app.

Quick Start

Install the Pixlift AI desktop app or use our web platform to start upscaling images in seconds.

Privacy First

Desktop processing can run locally. Web processing sends images to your connected runtime and uses managed storage for uploads and results.

Core Concepts

Image upscaling increases an image’s pixel dimensions. AI models estimate additional detail from the source image. Results depend on the model and input quality; generated details may differ from the original scene.

The NCNN Engine

Our desktop application uses the NCNN high-performance neural network inference framework, optimized for mobile and desktop GPUs via Vulkan. This allows for lightning-fast upscaling even on consumer hardware.

Architecture

The Pixlift ecosystem consists of three main components:

  • Pixlift Desktop: A Tauri-based application that runs the AI models locally.
  • Pixlift Web: A React-based dashboard for managing projects and accessing cloud upscaling features.
  • Connected processing: Your Google Colab runtime runs AI inference. The web app uses Firebase for accounts and project data, managed asset storage, and Vercel serverless endpoints.

How to upscale an image online

  1. Open the web app and sign in to your Pixlift AI account.
  2. Start your Google Colab runtime and connect its endpoint using the app’s Colab connection controls. Keep the runtime active while processing.
  3. Create or open a project, add your images, and select an available AI model and scale factor.
  4. Start processing, compare the enhanced image with the source, and download the result.

Batch image upscaling

Add multiple images to a project and process them through the queue. The connected runtime’s GPU memory and model determine how much work can run at once. Try a small batch first and check the output before processing a larger collection.

Desktop or Google Colab processing?

Choose desktop processing to use your hardware locally. Choose the web app with Google Colab to use a connected GPU runtime from your browser. Colab availability, session limits, and performance depend on your runtime and plan.

For photos, start with a model intended for photographic images. For illustrations, choose a model suited to artwork when available. Inspect textures and faces at full size: a larger output does not guarantee an accurate reconstruction.