Sylvain Meunier

Because we all have many social networks but we are still unique human beings, I decided linking all my account here so everybody can follow me (or not...) wherever they want : twitter.com/meuns, @meuns" target="_blank" rel="nofollow noopener">mastodon.gamedev.place/@meuns, linkedin.com/in/sylvain-meu….

@valentin_galea This is a great tool !

GPU is free with Windows SDK learn.microsoft.com/en-us/windows-…

Comparing GPUView traces of Monster Hunter World (left) vs #MHWilds (right)... it's so bottlenecked I cannot properly frame it with my 2K monitor! 1080p on Laptop Nvidia 4090 with High Settings and internal upscalers Blue vertical lines are V Sync's (1/2)

@pixelsnafu @SebAaltonen The demo is clearly work in progress from their research team. It's obvious it will improve significantly.

@SebAaltonen Their ray tracing tech demo was the most underwhelming thing I've seen in a while. But still kudos to them for getting a lot of other things right. Hopefully they put out better looking demos in the future. 😅

AMD's marketing team is learning from their mistakes. This launch event was solid. "2% slower than competitor" provides much more trust than "RTX 5070 is faster than RTX 4090 (with 4x frame gen)".

Let's interact with the light waves! Here, I created a real-time simulation that can run in your browser: chiuhans111.github.io/interactwave/ You can change the beam width and the focusing power of the input light rays. Please give it a try and let me know your thoughts! #Optics #WebGL

@AndroidBlogger Not much, graphics developers are probably only able to use c++ professionally... ^^

Language wars are still an important thing these days...

@NOTimothyLottes I might be blind, but MSAA and hardware VRS are more or less the same under the hood and I can't find how to choose or find VRS sample positions for implementing custom resolving/upscaling with D3D12 nor Vulkan.

So we (the industry) really ever went back to the MSAA based TAA combinations and applied everything we learned in the past decade ...

Judging by the Blackwell whitepaper, RTX Mega Geometry is an implementation of ideas from this @HPG_Conf 2023 paper by Carsten Benthin and me. It is cool to see this broadly deployed so soon. Hopefully, cross-vendor standardization will be just as swift. x.com/MomentsInCG/st…

@NOTimothyLottes @benjedwards Reviving the Steam box or Steam machines initiative is probably not that hard. ^^

Getting back to this by @benjedwards - vintagecomputing.com/index.php/arch… ... for gaming, to get back to a place where "personal computers" actually have a customer base, one has to solve the Microsoft OS Monopoly problem, and the only long term hope is SteamOS getting PC pre-install

This new sample demonstrates how to use the VK_KHR_dynamic_rendering_local_read extension in conjunction with the VK_KHR_dynamic_rendering extension. This combination can replace core render and subpasses. github.com/KhronosGroup/V…

It's free but you should buy it !

At Z-Labs we treat OBJ file (COFF/ELF) as plugins. It is very simple with bof-launcher lib and bofObjectGetProcAddress() API. #cybersecurity #infosec #hacking

@dankbaker @SebAaltonen And worth 2k$ !

@SebAaltonen I like that he looks sadder and aged.

HypeHype mascot "Ping." Old renderer (left). Concept art with future renderer (right) :D

Portals and Quake - Computer Graphics & Programming with Pekka Väänänen 30fps.net/pages/pvs-port…

@BattleAxeVR More memory is always better ! ^^

The Laplacian operator can be expressed as the difference of a pair of smoothing operators. Here’s a canonical demonstration

@MittringMartin Quite sure atomics are future of GPU programming, not AI nor RT... ^^

Vulkanised24: GPU Atomic Performance Modeling with Microbenchmarks vulkan.org/user/pages/09.…

@NOTimothyLottes Horizon team did something interesting for PS5. Maybe the will present their work at some point.

People keep claiming non-AI is dead with regards to visual improvements, well that is 100% bogus. No one has yet scratched the surface of what is possible.

__/ HOW TO MAKE A "GOOD" NON-ML SCALING-TAA \__ ML people specialize in understanding what training data to feed a fixed network, likely instead of understanding how the network is actually solving the problem Alternative: understand the problem and write a shader solution ...

@NOTimothyLottes I tested this during Fsr2 research period but it never landed in the product. It did mitigate various corner cases like popping objects and other out of the screen objects when the neighbor clamping fails rectifying history. Fsr2/3 uses a simpler approach.

Check if forward projected motion gets back to the pixel you are filtering Be careful to dilate nearest motion

@NOTimothyLottes Grids can live in many spaces and they all have some sort of resolution. Radiance caches are world space or texture space most of the time and they workaround/trade disocclusions for other bad things.

@meuns So while one could try to wait out the increasing TDP trend until it hits wall outlet limits, it is probably far better to move to a domain immune to disocclusions (not screen space) and one where asymmetric work/pixel can be applied

IMO this is what they should have showcased in their CES keynote: youtube.com/watch?v=xt_opW…