Using neural networks as the final rendering element in the graphics pipeline in Unity.

How it works

  • A Unity scene consisting of objects is rendered using a regular pipeline.
  • At the output, each object has a special mask in the form of an image, which is processed by a neural network that converts it into the final image.
  • The generated image is the result of the final render displayed on the screen.

Technology stack

  • The neural network runs on the server
  • Unity acts as a client application
  • For the VR experience, the HTC Vive and SteamVR helmet were used

We see the development potential in creating a tool for drawing applications in virtual reality and obtaining a photorealistic result.


  • Potential for art – creating visuals for music or installations 
  • GauGAN in virtual reality