lol I randomly searched for a page from collabora about how they reverse engineer GPU drivers (since I’ve seen that they’ve done that for other GPUs) and the very first page I went to was written by this person. Apparently this was a side project while she did fairly similar work on Mali GPUs at Collabora.

Anyways, here’s what she said about the process:

Typically, for Linux or Android driver reverse-engineering, a small wrap library will be written to inject into a test application via LD_PRELOAD that hooks key system calls like ioctl and mmap in order to analyze user-kernel interactions. Once the “submit command buffer” call is issued, the library can dump all (mapped) shared memory for offline analysis…

it took some elbow grease to identify the three critical calls: memory allocation, command buffer creation, and command buffer submission…

With those obstacles cleared, we can finally get to the shader binaries, black boxes in themselves. However, the process from here on out is standard: start with the simplest fragment or compute shader possible, make a small change in the input source code, and compare the output binaries. Iterating on this process is tedious but will quickly reveal key structures, including opcode numbers.

From https://alyssarosenzweig.ca/blog/asahi-gpu-part-1.html , there’s a lot more details about the subsequent parts of the process in the next articles but they’re harder to summarize

lol why would you try to compete with something that no one’s buying

It’s more than good enough for CPU imo but the GPU is going to really struggle. I think though most people just want onboard processing for the foveated streaming support and plan to use it with an actual PC, where I’ve heard the headset is extremely comfortable and lightweight despite what is then an overspecced processor

For standalone it should be faster than a Quest 3, so I think it’s a stretch to call it outdated, it’s still not great though

https://www.collabora.com/news-and-blog/news-and-events/building-an-arch-linux-aarch64-port-for-holo-core.html

That isn’t the first impression that I got, although of course I’m just one data-point. The composition, lighting, and post-processing of the stills are really good which is not common for games that are actually lazily thrown together assets. Looking a bit closer though, I do see what you mean and I think the main problem is that the level of quality doesn’t seem that consistent. IDK what’s going on on that 13th still with the monitor. A lot of the scenery in the trailer looks worse that the scenery in the stills, and some of the UI stuff seems a bit meh. I get that as an indie dev it takes forever to do this sort of stuff, and definitely there’s very little that looks outright bad, the problem is mostly the inconsistency in quality. People only have so much time to play games, and it’s not that easy to judge how fun a game is from it’s trailers and description, so they look for any clues they can of less than complete effort, as any inattention to detail could indicate a lack of attention to the actual gameplay parts as well. Unfortunately for both gamers and developers, how fun or engaging your game is doesn’t necessarily correspond that well with how appealing it is to people who see the trailer.

IDK, this is all kind of unfounded speculation from someone who really doesn’t have much personal experience in the area, so take it with a grain of salt

I think the money accumulates over time, so you’ll eventually get a payment once you have $100 at the start of a month after fees. But there’s also the $100 up front cost to list a game, so even after the first payment you haven’t really made money

You’d probably have to get like $200 in sales from a game before you have actually started to make money (from taxes and steam’s 30% cut)

Idk this is just what I’ve heard on the internet, I haven’t experienced it myself

Yes, Google Translate has used neural networks for quite a while. Recently they’ve also switched it to using the same Gemini LLMs that they give you chat interfaces with, so it’s pretty equivalent to using ChatGPT for translations. It’s just another company’s version of the same thing, wrapped in a translation interface.

It’s good to encourage reuse, which is eBay’s main thing. I wouldn’t have a reason to buy anything new from them however.

I got my home server (Lenovo thinkcentre, i7 6700) for $30 minus ram or storage at my local university surplus store a few years ago, and I have no regrets. Added a 256gb sata SSD, 16 gb RAM, 8tb HDD all refurbished for like +$150 when that was still cheap.

If you are playing games where trackpads are useful, there’s not really another option, so it’s automatically a good value. I know my steam deck experience would have been a whole lot worse without those, and I would probably never consider a gaming handheld without them. But for the gaming I do with an xbox controller, I currently just use it for some video games at my computer (where I have access to a mouse anyways), and maybe split screen with family. I think my Gamesir Cyclone 2 controller (at half the price) is an unambiguously better deal for that, in the premium controller space. If I were using it mostly for couch gaming, that might put the Steam controller in a better position, or if I were mostly playing games that support whatever haptic trigger things Sony has, that controller might be in a better position.

One of the other more unique features is the tracking in the Steam Frame. It would be cool if they could standardize that sort of thing so it would work with other headsets. I wonder if they’ve considered that.

As an amateur computer graphics person, the best way to draw accurate stars is to just pre render it onto a cubemap. But if you really need that subpixel worth of parallax to be completely accurate for every star, there are a couple ways I can think of off of the top of my head. With any you’d want to make sure you only store position, size, and color, since stars are all spheres anyways. With effort, you can be very flexible with how these are stored. (4 bits color temperature, 4 bits size, 3*32 bits coordinates maybe)

• splat each star into the screen texture with atomics
• some sort of tiled software rasterization thing, like in Gaussian Splatting

Worse ideas:

• instanced hardware rasterization
• ray tracing

This is not that well suited to most usual rendering techniques, because most stars are probably going to be much smaller than a pixel. Ray tracing would mean you need to just hit every star by chance (or artificially increase star size and then deal with having tons of transparency), hardware rasterization is basically the same and additionally is inefficient with small triangles. I guess you could just live with only hitting stars by chance and throw TAA at it, there’s enough stars that it doesn’t matter if you miss some. That would react badly to parallax though and defeats the purpose of rendering every star in the first place.

It’s much more efficient to do a manual splatting thing, where for each star you look at what pixel(s) it will be in. You can also group stars together to cull out of view stars more efficiently. Subpixel occlusion will be wrong, but it probably doesn’t matter.

This is all just for the viewport, though. Presumably there are other objects in the game besides stars, which need to have reflections on them of the stars. Then that becomes an entirely different problem.

The real answer though is that you wouldn’t try to render all of the stars, even if you want parallax. Maybe some of the closer and larger ones as actual geometry, simplify a ton of stuff in the background, render things as volumes or 2d billboards, have a cubemap for the far distance, etc

Edit: also ofc this presumes you know the position, scale, temperature of every star

I also like the idea of baking all of the stars into a volume in spherical coordinates, centered around the origin

Oh, yeah, I didn’t see those. I think my point still stands though, really those specular highlights shouldn’t be that bright, but the AI can figure out that it’s plausible for them to be brighter and that it would fit the target style better.

Yeah, probably the main reason it’s getting the little bit of praise that it does is that they’re showing it off on games with fairly flat-looking skin shaders. Unfortunately a problem with this sort of thing is that getting that “2023” image is the result of giving a whole team a huge amount of time to model one man’s face. If you’re Bethesda and you just want to get NPCs into Starfield, it would be a similar amount of work. A bit less, since the first people already gave a talk on it, but still much more work then just getting a diffuse BRDF with some subsurface scattering and calling it good. But you also need a process that can be applied to every single NPC…

And looking at Striking Distance Studios, the company where that 2023 image is from:

In February 2025, it was reported that most of the studio’s developers had been laid off.

Yeah, I think it’s safe to say that the work those people put in will never be directly reused.

Another reason the DLSS version looks a bit more realistic there is because of the specular highlights on the eyes, for example. They probably aren’t reflecting anything real, or else they would be there in the original. But the AI knows that specular highlights add realism and are plausible in this scene, so it puts them there. That’s something that an artist could do if given a specific shot and camera angle, but in the general case they can’t really do that without causing problems.

Fun fact that you may or may not have heard before: the light flicker animation in Half Life Alyx is actually the exact same one used in the original Quake. Half Life 1 was built on the Quake engine, and the same animation was carried over into Source and then Source 2.

https://www.alanzucconi.com/2021/06/15/valve-flickering-lights/

sounds like that’s planned but maybe not in yet

yep, I got through about 50 hours of Subnautica at ~15 fps back in the day.

Even portal with RTX is kinda playable on the steam deck though if you use that config that one person made to enable fsr3 and do some default settings changes, and put the gpu clock up to full

Sure, I could definitely see situations where it would be useful, but I’m fairly confident that no current games are doing that. First of all, it is a whole lot easier said than done to get real-world data for that type of thing. Even if you manage to find a dataset with positions of various features across various biomes and train an AI model on that, in 99% of cases it will still take a whole lot more development time and probably be a whole lot less flexible than manually setting up rulesets, blending different noise maps, having artists scatter objects in an area, etc. It will probably also have problems generating unusual terrain types, which is a problem if the game is set in a fantasy world with terrain that is unlike what you would find in the real world. So then, you’d need artists to come up with a whole lot of datat to train the model with, when they could just be making the terrain directly. I’m sure Google DeepMind or Meta AI whatever or some team of university researchers could come up with a way to do ai terrain generation very well, but game studios are not typically connected to those sorts of people, even if they technically are under the same company of Microsoft or Meta.

You can get very far with conventional procedural generation techniques, hydraulic erosion, climate simulation, maybe even a model of an ecosystem. And all of those things together would probably still be much more approvable for a game studio than some sort of machine learning landscape prediction.

I don’t know of any games that use machine learning for procedural generation and would be slightly surprised if there are any. But there is a little bit of a distinction there because that is required at runtime, so it’s not something an artist could possibly be involved in.

Are there any alternatives that are decently fast for large files? My computer and my phone both get at least 300 mbps from the router, and I have yet to find a local file transfer application that will be anywhere near that fast for large files (destiny, local send, kde connect, might have tried others, I don’t remember)

The index was a good headset when it came out 6 years ago. It isn’t really anymore. (In terms of visual quality at least)