Grég Angerand

Did a quick and dirty implementation of a spatial hash structure to speedup RTAO, ray results are stored in cells indexed by pos/normal/cell size and after storing a few rays occlusion can be queried from the cell instead of raytracing it. 3x faster RTAO with no denoising.

@_x1m4 You don't need pumps, pressurizing the atmosphere inside the base is enough. With a proper breathing mixture and acclimation humans can go well bellow 300m. It's called saturation diving and I believe it's mostly used in the oil/gas industry.

RenderDoc v1.41 is now available! It adds threaded support for shader debugging, enabling debugging of shaders that use subgroup/wave operations and simulating all threads in a compute group. Full release notes: github.com/baldurk/render… Binary builds: renderdoc.org/builds

It might be the worst possible outcome. intel Arc GPU dead Nvidia ARM CPU dead

@metehanbolatt Using a heightmap and a moving threshold value seems much simpler.

Can we find the paths of a vector one by one and animate its color? All we need is a VectorGroup and its children to traverse it one by one... And a little knowledge of vector drawing and path. And the magic happens inside convertVectorPathToAndroidPath() #Kotlin #Android #AndroidDevs #JetpackCompose #Animation #PathAnimation #Vector

@matiasgoldberg The inherited isGroundUnit bool implying that Ship already contains part of the ground behavior makes this actual art.

Nothing screams Object-Oriented Programming in Gamedev more than "A ground unit is a ship that walks on land"

@SebAaltonen I am regularly disappointed by MSVC's inability to optimize fairly trivial stuff, but not inlining memcmp when the size is a small constant is wild.

FU MSVC! :D

Compiler Explorer is super important for high performance workflow. Often you have to confirm your assumptions: A) A struct of 8x uint8 will be passed as a single 64 bit register B) memcmp provides optimal codegen -> just a single 64 bit integer compare.

On branching in shaders: If-statements are generally better than math operations. The concerns about branching are mostly misplaced based on very old hardware. The main focus should be on avoiding divergence: pixels near each other should do similar things. Branching can become a problem when adjacent pixels are doing wildly different operations. One such example is stochastic path-tracing, where each pixel follows a random path with little coherence. In simple cases, if-statements are better than using mix/step

@SebAaltonen I thought that renderpasses (and subpasses) saved bandwidth on tiled mobile GPUs. Is it the case or can we forget about them and use dynamic rendering everywhere now?

If you are interested in low level rendering architecture design, I recommend my 2 year old SIGGRAPH talk: advances.realtimerendering.com/s2023/Aaltonen…

I decided to refactor our Vulkan 1.0 style render pass API and move to Metal/WebGPU/Vulkan 1.3 style API with no persistent render pass objects. We had persistent render pass layout objects too. Those are gone as well. Thread...

My lab will be recruiting at all levels. PhD students, postdocs, and a research engineering position (worldwide for PhD/postdoc, EU candidates only for the engineering position). If you're at SIGGRAPH, I'd love to talk to you if you are interested in any of these.

This article on virtual shadow maps is insanely detailed (and good). I'm not sure how similar the implementations are, but this technique is also used in Unreal Engine. ktstephano.github.io/rendering/stra…

@FreyaHolmer Option 5: snap each vertex to the closest grid point and watch the chaos.

if a translation operation of selected vertices wore grid-snapped pants, would it wear it like this or like this or like this or like this

@Phantom_TheGame I am pretty sure it's because we don't usually fall from so high. Does it feel more realistic for a more normal jump height (like 1-2m) ?

I always wondered why games almost never use realistic gravity. Turns out realistic gravity just seems way too floaty. Does anyone know why? Maybe we're just not used to actually falling such distances?

@newincpp This seems like a great idea! Having per call overrides like this would significantly reduce ImGui code for the more complex widgets. It doesn't seem too hard to do with a C++ wrapper over ImGui. I will definitely try.

There is probably also a way to do something like: ui::text("my text", { Padding(15), BGColor(0.3, 0.3, 0.5) }); but that's not incompatible with the text version.

Using wpf is making me want to make a C++ immediate mode UI where you can pass style files (that can include shaders) to calls. Something like: ui::text("mytext", "maintext.style"); So it can be as easy as imgui but as stylish as html/css.

you can explain most weird AAA game business decisions with the "WoW problem". "The average MMO makes 500 million dollars" when that is 99% WoW, so investors go "we want an MMO" not realizing the market was for WoW and not "any mmo". first noticed back when WoW was dominant

shader programming is for when off-by-1 errors just aren't doing it for you anymore, and you're looking to get into new things like • off by 1/2 • off by multiple of -1 • off by power of 2 • off by a reflection around 1/2 • off by 1 graphics api

@tha_rami Awesome ! DR400 ?

Looks like it's this time of the year when I spend an entire week trying to make something trivial that can be explained in a single sentence in CMake

@PirateSoftware If someone ever makes a factorio version of this concept I am done for.

Really enjoying having Rusty's Retirement running on the bottom of my screen. Reminds me of the old days with desktop pets and fun mini-widgets. Would love to see more of this in the industry. store.steampowered.com/app/2666510/Ru…

@BenjiGameDev D or C. I feel like which one is the "correct" one depends on the player habits. As a mostly shooter player I tend to actively aim at what I want to engage, and hate when the camera moves on its own, so D. But I can see a fighting game player preferring C.

There are four enemies nearby. You press lock-on. Which enemy gets targeted and why? A) Closest distance, off-screen B) Second closest, behind player C) Third closest, in player's line of sight D) Farthest distance, but nearest center of camera