3D Generative Adversarial Networks

Physical Hands
Deep Learning

Using some of the latest deep learning technologies to bring voxel assets to life

Pose Detection
Tensorflow

Tensorflow was used to generate these assets

XRI
Game Ready assets

Although these assets are a little unrefined, the images are direct from the unity editor

The Why

Deep learning has a vast range of applications, from recognizing numbers to generating entire games. In this project, I utilized TensorFlow and TensorLayer to generate 3D models from a dataset, specifically focusing on cars.

Training the Generative Adversarial Network (GAN) took over 24 hours and around 3000 epochs to produce the resulting images. These generated car models, though slightly undefined, are fully usable as game assets. This project highlights the potential of deep learning in creating complex 3D models for use in various applications.