Game Character Generation Using Generative Adversarial Network

Advanced Deep Learning models like Generative Adversarial Networks (GANs) can help game designers to create or adopt new game characters whilst saves time and cost for the Studios.

Designers across the gaming industry go through long and tedious ideation and development cycles for creation of new game characters as multiple attributes like face morphology, gender, skin tone, clothing accessories, and expressions need to be factored in afresh each time. Via this whitepaper, we aim to help gaming studios minimize this complexity by shedding light on machine learning frameworks like GANs that can combine the power of AI with human supervision to usher in a new paradigm of creativity with productivity. GANs can empower studios to quickly sketch new characters through image generation and style mixing techniques leading to huge time and cost savings.

Learn how Generative Adversarial Networks (GANs) and their variants work and explore their effectiveness via a specific use case of developing new characters for a game called Mortal Kombat (MK) through the technique of style transfer on new and synthesized images.

Download Whitepaper

This whitepaper will provide you with exclusive access to:

Detailed analysis of effectiveness of GANs in terms of image quality produced (subjective evaluation) and FID distances (objective evaluation)

Results of style mixing performed using the trained model for creating Mortal Kombat characters

Experimental evaluation using Mortal Kombat dataset (custom dataset having 25,000) images

Want to know more about our Gaming Analytics Solutions?

Talk to our Experts today!