Deep Learning for Games
Deep learning, a branch of artificial intelligence is revolutionizing modern computing. It is being used to develop solutions that range from improved cancer screening to self-driving cars. It has been used to create art, play games and deliver customer insights. NVIDIA brought presentations, demos and training materials to GDC17.
Applications for game development include, customer insights, content creation and AI for characters. Deep learning has dramatically improved technologies like text to speech, speech to text and speech synthesis, all of which can be used to increase the sense of immersion delivered by games.
We had 2 talks on deep learning at GDC. Bryan Catanzaro, our Vice President of Applied Deep Learning Research will kick things off with an Introduction to deep learning and then Andrew Edelsten, Senior Developer Technology Engineer will lead a talk on generating and enhancing materials using deep learning techniques.
Introduction to Deep Learning
Deep learning, a branch of machine learning and artificial intelligence (AI), is changing the the entire computing industry. Learn how to leverage deep learning to create, develop, market, run and tune higher quality and more appealing games for mobile, console and PC.PDF | Video
Zoom, Enhance, Synthesize! Magic Image Upscaling and Material Synthesis using Deep Learning
Recently deep learning has revolutionized computer vision and other recognition problems. Everyday applications using such techniques are now commonplace with more advanced tasks being automated at a growing rate. During 2016, “image synthesis” techniques started to appear that used deep neural networks to apply style transfer algorithms for image restoration. The speakers review some of these techniques and demonstrate their application in image magnification to enable “super resolution” tools.PDF | Video
Deep Learning Institute
We ran two training sessions for developers at GDC. These covered much of the material presented above, but also included hands on labs where developers got to train a neural network to solve real world problems. If you didn't make it to a session you can try the lab for yourself using a DLI Quiklab.
We suggest you try the Getting Started with Deep Learning lab.
GameWorks: Materials & Textures
GameWorks: Materials & Textures is a set of tools targeted at 3D and graphics artists that leverages the power of Deep Learning and NVIDIA CUDA.
- Super-Resolution Allows the user to get a 2x or 4x increase in resolution of textures and photos.
- Photo To Material: 2shot Using two photos of a real-world surface, generates diffuse, normals, specular and gloss maps.
- Texture Multiplier Takes a texture or photo as input and provides the user with an organic variation.
Learn more and sign up for the beta here.
More Deep Learning from NVIDIA
News and resources for developers interested in deep learning and artifical intelligence
GPU Technology Conference
NVIDIA hosts GTC, the world's biggest conference focused on GPU acceleration and AI. If you want to learn more about deep learning, this is an event you shouldn't miss.Learn more