This is operated by a business or businesses owned by Informa PLC and all copyright resides with them. Informa PLC's registered office is 5 Howick Place, London SW1P 1WG. Registered in England and Wales. Number 8860726.
|Session Name:||Machine Learning Summit: Advanced Heightmap Compression Using Deep Learning in 'Dune: Awakening'|
|Speaker(s):||Clemens Rgner, Calle Lejdfors|
|Company Name(s):||Tencent Games, Tencent Games|
|Track / Format:||Machine Learning Summit|
Did you know free users get access to 30% of content from the last 2 years?
|Overview:||The vast desert landscapes of Arrakis are the key feature of Dune: Awakening. The players are confronted with dangerous environments and surviving in the harshness of the world. The large scale and its importance for the gameplay poses various challenges for the developers, not least of which is the landscape's enormous scale and the substantial on-disk size of the game.To tackle the challenge of such large install sizes of terrain data, the team developed a novel approach using Deep Learning models to compress and decompress heightmap textureswhich will be discussed in this session. With their lossy compression, the team was able to achieve compression rates ranging from 8x to 18x, satisfying critical games requirements for low storage space. They also kept the quality loss low and free of obvious artifacts, so the in-game experience did not suffer.|