Alexandre BaronWatching #SIGGRAPH2024 presentations these days... In this thread I'm going to write some feedback on the ones I've seen 😀
"An Introduction to Creating Real-time Interactive Computer Graphics Applications" by Dave Shreiner, Edward Angel: I always watch these because I wanna see if there is new stuff that I missed since last year. But nope, this is basically the same presentation as last year, if you already saw it, you can safely skip it. Great introduction though
"Mathematical Optimization in Computer Graphics" by Adam Bargteil, Marc Olano: Given the title I was expecting something technical. It's actually a monstruous 3 hours (!!) review of influential papers from 1994 until today.
Out of my depth for the most part, but I liked to hear about "An Inexpensive BRDF Model for Physically‐based Rendering" by Schlick (1994), "Synthesizing Physically Realistic Human Motion
in Low-Dimensional, Behavior-Specific Spaces" by Safanova et al. (2004), and "Approximate Models for Physically Based Rendering" by Pesce and Iwanicki (2015). As a lowly engineer I always like to have papers explained to me 😁
in Low-Dimensional, Behavior-Specific Spaces" by Safanova et al. (2004), and "Approximate Models for Physically Based Rendering" by Pesce and Iwanicki (2015). As a lowly engineer I always like to have papers explained to me 😁
As someone who was (and kinda still is) in education I'm always interested in knowing what is important to teach students nowadays and what can they do after school, so I wanted to watch "Student / Futures: Creative Careers in Animation, Computer Graphics, and Interactive Techniques", by Johannes DeYoung, Claudia Davis, Clinton Fanelli, Mike Stroud, but it wasn't recorded. Too bad 😭
"Bayesian Statistics: A Practical Introduction for Computer Graphics" by Mirinda Whitaker, Sarah Creem-Regehr, Bobby Bodenheimer: A great introduction on Bayesian Statistics applied to CG, but unfortunately I don't have the background on Bayes to understand it fully 😅 It was hard...
"An Introduction to Quantum Computing" by Andrew Glassner: I don't know anything about quantum computing. I thought it was the occasion to give it a shot. The lecturer does a great job at presenting the basic concepts in layman's terms, even though at some point I thought the metaphor was a bit too heavy. I wish there was less metaphor and more introduction on what quantum mechanics *actually* are.
"Demystifying Color in Your Digital Media and Data Visualizations" by Theresa-Marie Rhyne: Since learning about gamma correction, I always thought that despite its apparent simplicity, color is one of the hardest topics to truly understand. This lecture is packed with interesting fun facts and reminders. Did you know Magenta is not a spectral color? That there's an NVidia Green? The difference between purple and violet? What the K in CMYK stands for? It's all in there.
"The Life and Legacy of Bui Tuong Phong" by Theodore Kim, Yoehan Oh, Jacinda Tran: Not technical at all, but focused on the life and death of the legendary researcher Bui Tuong Phong and common misconceptions about him. For example, "Phong" is actually his first name, so he was credited wrong all this time, and the #1 picture on google under his name is actually not him. For the first time ever, his daughter and wife were also here to answer questions. One of the best talks in my opinion!
"A Whirlwind Introduction to Computer Graphics for Complete Beginners" by Mike Bailey: Another introduction lecture, here again nothing much changed, if you've already seen it, you can safely skip it - although Mr Bailey is so passionate talking about CG that it makes you want to watch it again 😁
That's all for now. But I have more presentations to watch. I will update this thread when I'm done watching more videos...
I don't have a lot of time to watch #SIGGRAPH2024 talks 😭But I've watched some more!
"Introduction to Real-Time Ray Tracing" (Chris Cascioli, Laura Reznikov) : For someone like me who didn't have the time yet to investigate RTRT, this is a great introduction. I didn't know about all the new types of shaders used by these techniques. Exciting!
"Introduction to Real-Time Ray Tracing" (Chris Cascioli, Laura Reznikov) : For someone like me who didn't have the time yet to investigate RTRT, this is a great introduction. I didn't know about all the new types of shaders used by these techniques. Exciting!
"Advances in Real-Time Rendering in Games, Part I" (Natalya Tatarchuk, Michał Iwanicki, Simon Taylor, Shun Cao): "Advances" are always my most awaited talks because they're usually focused on tech and specifically on new things!
Part 1 in a Nutshell:
- Activision tried to use machine learning to improve their light baking process. Bottom line: it worked, but definitely not a good solution at the moment. Needs more R&D
- Tencent discussed their pipeline to achieve real-time dynamic GI on mobile: Cluster-based shading, visibility buffer, voxelization. The result kinda looks like Nanite. Neat!
- EA talked about Serac, their in-house shading language for Frostbite, and the pros and cons. Interesting!
- Activision tried to use machine learning to improve their light baking process. Bottom line: it worked, but definitely not a good solution at the moment. Needs more R&D
- Tencent discussed their pipeline to achieve real-time dynamic GI on mobile: Cluster-based shading, visibility buffer, voxelization. The result kinda looks like Nanite. Neat!
- EA talked about Serac, their in-house shading language for Frostbite, and the pros and cons. Interesting!
"The Future of Teaching Computer Graphics", Edward Angel
Dave Shreiner: Very interesting to know what other teachers think about how to teach CG to future students. Bottom line: What works for them is using WebGL (maybe WebGPU later), focus on shaders, and keeping CPU/GPU data transfer as low as possible (hardcode things!). New APIs have too high barrier to entry.
It says something that even today, GL is still relevant...
Dave Shreiner: Very interesting to know what other teachers think about how to teach CG to future students. Bottom line: What works for them is using WebGL (maybe WebGPU later), focus on shaders, and keeping CPU/GPU data transfer as low as possible (hardcode things!). New APIs have too high barrier to entry.
It says something that even today, GL is still relevant...
"There Can Be (at least) Two Introductory Graphics Courses: Teaching Introduction to Non-interactive Computer Graphics", Michael Shah: A talk mostly about how he structured a course about raytracing graphics based on Peter Shirley's "Raytracing in..." book trilogy and how it differs from a classic "GL-based" approach to teaching CG. Very cool, probably the type of course I'd have liked to have at university. Giving me ideas :)