Course Description

 

Modern video games employ a variety of sophisticated algorithms to produce groundbreaking 3D rendering pushing the visual boundaries and interactive experience of rich environments. This course brings state-of-the-art and production-proven rendering techniques for fast, interactive rendering of complex and engaging virtual worlds of video games.

 

This year the course includes speakers from the makers of several innovative game companies, such as Ubisoft, Sledgehammer Games, Activision, NVIDIA, Unity Technologies, Lucasfilm,  and Epic Games. Topics range from variations on the latest techniques for modern rendering pipelines, high-quality depth-of-field post processing, advances in material representation for deferred rendering pipelines, including improvements in subsurface scattering and PBR material representations, improvements to diffuse multi-scattering BRDFs, volumetric rendering techniques, using ray tracing techniques for accurate soft shadows in real-time, and many more.

 

 

This is the course to attend if you are in the game development industry or want to learn the latest and greatest techniques in real-time rendering domain!

 

Previous years’ Advances course slides: go here

Syllabus

Advances in Real-Time Rendering in Games: Part I

Monday, 13 August 2018 9am - 12:15pm |  East Building, Ballroom BC, Vancouver Convention Centre

Advances in Real-Time Rendering in Games: Part II

 

Monday, 13 August 2018 2pm - 5:15pm |  East Building, Ballroom BC, Vancouver Convention Centre

Prerequisites

 

Working knowledge of modern real-time graphics APIs like DirectX or Vulkan or Metal and a solid basis in commonly used graphics algorithms. Familiarity with the concepts of programmable shading and shading languages. Familiarity with shipping gaming consoles hardware and software capabilities is a plus but not required.

Intended Audience

 

Technical practitioners and developers of graphics engines for visualization, games, or effects rendering who are interested in interactive rendering.

Advances in Real-Time Rendering in Games: Part I

 

9:00 am

Natalya Tatarchuk

Welcome and introduction

 

9:10 am

Steve McAuley (Ubisoft)

The challenges of rendering an open world in Far Cry 5

 

10:10 am

Danny Chan (Sledgehammer)

Material advances in Call of Duty: WWII

 

11:10 am

Guillaume Abadie (Epic Games)

A life of a bokeh

 

12:10 pm

Closing Q&A

Advances in Real-Time Rendering in Games: Part II

 

2:00 pm

Natalya Tatarchuk

Welcome (and welcome back!)

 

2:05 pm

Sebastien Lagarde (Unity Technologies)

Evgenii Golubev (Unity Technologies)

The road toward unified rendering with Unity’s high definition rendering pipeline

 

3:05 pm

Evgenii Golubev (Unity Technologies)

Efficient screen-space subsurface scattering using Burley’s normalized diffusion in real-time

 

3:35 pm

Matt Pharr (NVIDIA)

Real-time rendering’s next frontier: adopting lessons from offline ray tracing to practical real-time ray tracing pipelines

 

4:35pm

Stephen Hill (Lucasfilm) (presenter)

Morgan McGuire (NVIDIA) (presenter)

Eric Heitz (Unity Technologies) (non-presenting contributor)

Real-time ray tracing of correct* soft shadows (* without a shadow of a doubt)

 

5:15 pm

Natalya Tatarchuk

Closing Remarks

 

 

Course Organizer

Natalya Tatarchuk (@mirror2mask) is a graphics engineer and a rendering enthusiast at heart. As the VP of Graphics at Unity Technologies, she is focusing on driving the state-of-the-art rendering technology and graphics performance for the Unity engine. Previously she was the Graphics Lead and an Engineering Architect at Bungie, working on innovative cross-platform rendering engine and game graphics for Bungie’s Destiny franchise, including leading graphics on the upcoming Destiny 2 title. Natalya also contributed graphics engineering to the Halo series, such as Halo: ODST and Halo:Reach. Before moving into game development full-time, Natalya was a graphics software architect and a lead in the Game Computing Application Group at AMD Graphics Products Group (Office of the CTO) where she pushed parallel computing boundaries investigating advanced real-time graphics techniques. Natalya has been encouraging sharing in the games graphics community for several decades, largely by organizing a popular series of courses such as Advances in Real-time Rendering and the Open Problems in Real-Time Rendering at SIGGRAPH. She has also published papers and articles at various computer graphics conferences and technical book series, and has presented her work at graphics and game developer conferences worldwide. Natalya is a member of multiple industry and hardware advisory boards. She holds an M.S. in Computer Science from Harvard University with a focus in Computer Graphics and B.A. degrees in Mathematics and Computer Science from Boston University.

 

 

The Challenges of Rendering an Open World in Far Cry 5

 

Abstract: Open worlds with dynamic time of day cycles pose a significant challenge to graphics development. There are fewer hiding places for the weaknesses and edge cases in our rendering features, as care has to be taken to ensure features work in all scenarios. This talk discusses some of the challenges faced on Far Cry 5 developing a water system, a physically-based time of day cycle and closes by looking at some small techniques to improve the lives of artists.

 

Bio: Stephen McAuley is a 3D Technical Lead at Ubisoft Montreal on the Far Cry Central Tech team, where he holds the vision for the future of the 3D engine. During his time at Ubisoft, he has worked on many Far Cry games, spearheading the switch to physically-based lighting and shading, working towards a more data-driven rendering architecture, and focusing on visual quality. Previously, he was a graphics programmer at Bizarre Creations, shipping games such as Blur and Blood Stone. 

 

Materials (Updated: September 4h 2018): PPTX (382 MB), PDF (17 MB)

 

 

Material Advances in Call of Duty: WWII

 

 

Abstract: This session will describe a number of improvements that were made to the Call of Duty:WWII shaders for lit opaque surfaces. The authors extended the Diffuse BRDF to model Lambertian microfacets with multi-scattering. They derived a simple method to mipmap normal and gloss maps, by directly correlating gloss to average normal length of a distribution, bypassing an intermediate representation, like variance. Then cavity maps are auto-generated for every surface that has a normal map, and the presentation will also show how the authors handled occlusion and indirect lighting in cavities represented by these textures. The presenters will demonstrate how the environment split integral precalculation can be easily reused for energy conserving diffuse. Finally, they will show how Rational Functions can be a useful tool to approximate many functions that can't be represented analytically or efficiently.

 

Bio: Danny Chan is Principal Software Engineer at Sledgehammer Games, leading the rendering effort on Call of Duty: Modern Warfare 3, Call of Duty: Advanced Warfare and Call of Duty: WWII. Previously, he’s worked at Crystal Dynamics, Naughty Dog (as Lead Programmer on Crash Team Racing), EA and Namco.

 

Materials (Updated: February 5, 2019): PPTX (140MB), PDF (4MB), Course Notes PDF (4MB)

 

A Life of a Bokeh

 

 

Abstract: Lens of a physical camera have a depth of field phenomena that has an importance in cinematography to bring focus on desired subject of a frame. The challenge of real-time depth of field is to output the highest bokeh quality while remaining fast. This talk will journey through the step by step implementation of a depth of field algorithm starting from existing state-of-the-art, and fixing artifacts one after the other to converge to the final implementation released in Unreal Engine 4.20. In order to achieve its opposing goals including scalability across a large variety of hardware, high quality and performance, the self-contained algorithm blends between:

-          The fast performance of scatter-as-gather approach while efficiently solving a physically plausible geometric occlusion with large variety of blurring radii for background, and hole filling for the foreground;

-          The quality of scattered sprites for highlights and deals with all the complexity implication of combining them plausibly;

-          The details of sub-pixel accuracy of slight out of focus convolution with the challenges of running with a Temporal Anti-Aliasing from older but sadly never published algorithm used in Unreal Engine 4.

 

 

Bio: Guillaume Abadie is Graphic Programmer at Epic Games, working directly on Unreal Engine 4's renderer, more specifically on post processing. Notably, he implemented temporal up-sampling & dynamic resolution duo shipping on Fortnite Battle Royal running at 60Hz on consoles. He also built the compositing of the SIGGRAPH real-time live 2017 awarded The Human Race demo. (twitter: @GuillaumeAbadie).

 

Materials (Updated: August 24th 2018): PPTX (150MB)

 

 

The Road toward Unified Rendering with Unity’s High Definition Render Pipeline

 

Abstract: When designing a rendering engine architecture, one frequently must choose whether to implement a forward or a deferred renderer, as each choice presents an important number of design decisions for material and lighting pipelines. Each of the approaches (forward-, forward-plus, or deferred) has a number of strengths and deficiencies, widely covered in previous conference presentations from shipping games’ engines. The features offered by each rendering architecture varies widely, and often can be content-centric.

 

When designing the high-definition rendering pipeline (HDRP) for the Unity engine, the authors desired to leverage the strengths of each rendering approach, as necessary for various application contexts (a console game, VR application, etc.). Thus, an important design constraint for the architecture of HDRP became a unified feature set between the deferred and forward rendering paths.

 

This presentation will explain how the team tackled designing the lighting, material and rendering architecture for the HDRP with the feature parity design constraint as one of the main pillars. It will cover the details of the flexible G-buffer layout architecture, explain the logic behind the taken design choices, and necessary optimizations for efficient execution on modern console hardware. In addition, the authors will present advanced developments made in the field of physically-based rendering, material advances, focusing on the novel BRDF model used for the rendering pipeline. The talk will also provide a framework for correctly mixing normals with complex material for evaluation at runtime. Lastly, the architecture and technical details of the physically-based volumetric lighting approach will be described, along with the necessary optimizations for fast performance.

Bios: Sébastien Lagarde is a lead graphics programmer at Unity Technologies where he's driving the architecture of Unity's high definition rendering pipeline, amongst other projects. Prior to Unity, he has worked on many game consoles for variety of titles, from small casual games to AAA (Remember Me, Mirror's Edge 2, StarWars Battlefront, among some).  Sébastien has previously worked for Neko Entertainment, Darkworks, Trioviz, Dontnod and Electronic Arts / Frostbite.

Evgenii Golubev, Graphics Programmer in the Unity’s Paris Graphics team, where he is contributing to Unity’s high definition rendering pipeline architecture, focusing on physically-based materials and advanced light transport, making them run efficiently on modern hardware. Previously worked on the rendering technology at Havok and Microsoft.

 

Materials (Updated: August 25th 2018): PDF (14 MB), Book of the Dead HD RP Video, Automotive HD RP Video, Volumetrics in HD RP Video (Slide 134), HD RP Participating Media Authoring Video (Slide 137), Spotlight with highly forward-scattering fog GIF (Slide 140)

 

Efficient Screen-Space Subsurface Scattering Using Burley’s Normalized Diffusion in Real-Time

 

 

Abstract: Screen-space subsurface scattering by Jimenez et al and the multitude of parameterization options developed for it, such as Gaussian lobe, two Gaussian lobe, SVD, etc.,have produced impressive results in rendering organic materials and characters. However, these methods focused primarily on the multi-scattering contribution rather than the single scattering contribution (aside from simple approximations via an extra weight for the original diffuse texture.) These previously introduced techniques rely on a separable Gaussian filter to achieve fast performance on modern GPUs.

 

In this talk, the authors build on the recently introduced Burley normalized diffusion method, bringing this technique developed for offline rendering pipelines into real-time with a novel screen-space subsurface scattering method. This method accounts for both multi and single scattering appearance models. Its parameterization model is simple and is derived based on the values from real world measurements. The authors will compare the quality of this approach to the more commonly used screen-space subsurface scattering techniques and explain the necessary optimizations required to evaluate its non-separable kernel efficiently in real-time. This method is used in production and is highly performant on commodity console hardware, such as Playstation 4 and Xbox One at 1080p resolution.

Bio: Evgenii Golubev, Graphics Programmer in the Unity’s Paris Graphics team, where he is contributing to Unity’s high definition rendering pipeline architecture, focusing on physically-based materials and advanced light transport, making them run efficiently on modern hardware. Previously worked on the rendering technology at Havok and Microsoft.

 

Materials (Updated: August 25th 2018): PDF (5MB)

 

Real-Time rendering’s next frontier: Adopting lessons from offline ray tracing to real-time ray tracing for practical pipelines

Abstract: Ray-tracing has finally arrived to the real-time graphics pipeline, tightly integrated with rasterization, shading, and compute.  The ability to trace rays offers the promise of finally being able to accurately simulate global light scattering in real time graphics.  Just as physically-based materials have revolutionized real-time graphics in recent years, physically-based global lighting offers the chance to have a similar impact on both image quality and developer and artist productivity.

 

Getting there won't be easy, however: currently, only a handful of rays can be traced at each pixel, which requires great care and creativity from the graphics programmer.  There are lessons to be learned from offline rendering, which broadly adopted ray tracing ten years ago.  The author will discuss the experience of the offline pipelines with the technology and highlight a variety of key innovations that have been developed in that realm that are worth knowing about for developers adopting real-time ray tracing.

 

Bio: Matt Pharr is a Distinguished Research Scientist at NVIDIA where he works on ray-tracing and real-time rendering.  He is the author of the book Physically Based Rendering for which he and the co-authors were awarded a Scientific and Technical Academy Award in 2014 for the book's impact on the film industry.

 

Materials (Updated: August 24th 2018): PDF (74 MB)

 

 

Real-Time Ray Tracing of Correct* Soft Shadows

(* without a shadow of a doubt)

 

 

Abstract: With recent advances in real-time shading techniques, we can now light physically based materials with complex area light sources. Alas, one critical challenge remains: accurate area-light shadowing. The arrival of DXR opens the door to solving this problem via ray tracing, but the correct formulation isn't obvious and there are several potential pitfalls. For instance, the most popular strategy consists of tracing rays to points randomly distributed on the light source and averaging the visibility, but we will show that this is incorrect and produces visual artifacts. Instead, we propose a definition for soft shadows that allows us to compute the correct result, along with an efficient implementation that works with existing analytic area lighting solutions.

Note: this talk is an extended presentation of the paper Combining Analytic Direct Illumination and Stochastic Shadows (presented at I3D) that includes additional practical details.

 

Speakers: Stephen Hill (Lucasfilm) and Morgan McGuire (NVIDIA)

Non-speaking contributor: Eric Heitz (Unity Technologies)

 

Bios: Stephen Hill is a Senior Rendering Engineer within Lucasfilm’s Advanced Development group, where he is engaged in physically based rendering R&D for real-time productions such as the Carne y Arena VR installation experience.

 

Morgan McGuire is a scientist at NVIDIA, holds faculty appointments at the University of Waterloo, Williams College, and McGill University, and has co-authored several books, including The Graphics Codex and Computer Graphics: Principles and Practice.

 

Eric Heitz is a Research Scientist at Unity Technologies. His research focuses on physically based rendering including materials, lighting, sampling, LoDs, anti-aliasing, etc.

 

 

Materials (Updated: August 27th 2018): PDF (78MB), Video

 

 

 

 

 

 

 

 

Contact: