You've been logged out of GDC Vault since the maximum users allowed for this account has been reached. To access Members Only content on GDC Vault, please log out of GDC Vault from the computer which last accessed this account.

Click here to find out about GDC Vault Membership options for more users.

close

Session Name:

Machine Learning Summit: Differentiable Rendering for Scalable Asset Pipeline in 'Honor of Kings'

Overview:

In this session, Fei Ling and Frei Zhang will tackle the challenging problem of LOD assets generation for mobile games which run on devices with a wide variety of computational capabilitiesarguing that directly minimizing image space rendering loss is crucial to LOD asset quality. This presentation will introduce a novel and unified way of game assets fitting and generation, the whole pipeline is built on top of an in-house Hybrid Differentiable Renderer called Mythal which combines both rasterization and ray-tracing, taking the advantage of their respective strengthsspeed and realism. Armed with Mythal, they can calculate the image loss between reference rendering and LOD rendering, because the whole rendering pipeline is differentiable they can backpropagate the gradients for different rendering stages' attributes, and finally to interested game asset parameters, and then update these parameters with gradient-descent-based optimization methods. They will dive into several LOD generation applications which make their game Honor Of Kings (HOK) run efficiently on mobile devices with diverse hardware specifications, these applications include baked LOD of PBR shading, auto LOD skinning, auto mesh simplification, visibility baking, and more.

Did you know free users get access to 30% of content from the last 2 years?


Get your team full access to the most up to date GDC content

  • Game Developers Conference 2023
  • Fei Ling
  • Tencent Technology Inc.
  • Frei Zhang
  • Tencent Technology Inc.
  • free content
  • Tools Summit
  • Programming
  • Machine Learning Summit
  • Programming