Oliver (@fuwidabos) header
Oliver (@fuwidabos) profile picture

Rendering stuff @jangafx Previously @Framestore @TheFoundryTeam @Arm

Edinburgh
Joined October 2024
Photo by Oliver (@fuwidabos)Photo by Oliver (@fuwidabos)Photo by Oliver (@fuwidabos)Photo by Oliver (@fuwidabos)Photo by Oliver (@fuwidabos)Photo by Oliver (@fuwidabos)Photo by Oliver (@fuwidabos)Photo by Oliver (@fuwidabos)Photo by Oliver (@fuwidabos)Photo by Oliver (@fuwidabos)
Software should look cool as fuck
1
6
923
Playing around with EG2 physically based combustion.
2
8
128
4,104
Oliver retweeted
why no, I have not heard of rust, please tell me more
43
13
5
674
41,978
No other traveler has ever journeyed so far from home. Voyager 1 is about to become the first human-made object to reach a distance of one light-day from Earth, or about 16 billion miles. ½
338
5,340
626
22,492
1,835,960
Oliver retweeted
Being involved in various communities, I've noticed that people who used to make mediocre stuff before AI are now empowered to make more mediocre stuff faster. Conversely, people who used to make great contents can make slightly greater things now - when applying the same care.
17
37
2
475
15,726
The discourse around Big Walk has made it abundantly clear to me that all existing AI discourse revolves around how poorly introduced AI was. Instead of focusing on a tools-first approach, AI companies focused on stupid content generation shit which produced some of the worst garbage ever seen on the internet. Time for a stupid rant about AI from me. I want to make a few things clear: - Yes, all reasonable people agree that AI generated artwork/media is soulless garbage in its current state. Unless there was substantial effort on the part of the human behind it (e.g. the shape store) it has no original vision or cohesion. - Programming can 100% be an artform. But it also covers a wide breadth of technical knowledge. Even with my own CSCI degree, there's still a lot of shit that I personally didn't know how to do, and there weren't a lot of great on-boarding guides out there for those things. - House House did not do anything wrong here, they are airing far more on the side of caution than they frankly need to at this point. I think their reluctance to disclose this was justified given how massively overblown the reaction has been. They frankly shouldn't have done so. -- Programming and Art both have incredibly monotonous elements to them. Art is a lot more freeform and expressive, but programming at scale (especially in gamedev) is a logistical clusterfuck where you have to juggle dozens, if not hundreds (sometimes even thousands) of things around with all sorts of rigid scripts and build chains, configurations, etc. No one is having fun writing a CMake file, no one enjoys writing python scripts to organize and curate data, even if it's something they're good at. Programmers have complained for years about the logistical bullshit they have to deal with, and now they don't have to. This is the key difference between the shitty generative usage everyone knows AI for, and the pragmatic usage of AI as a tool that programmers are using it for. Generative AI art literally only exists for cynical reasons, and it deserves the hate it gets. It actually puts people out of a job, and doesn't function as a tool to accelerate artists in any way, merely to replace them. Programmers aren't replaced by AI, they are accelerated by it. We can use generative AI to rapidly drive the creation of scripts and toolchains to solve logistical problems much faster than we'd normally be able to on our own, allowing us to focus less on monotonous bullshit, and more on the fun elements of system archetyping, technical design, and steering all of the logistics together. Instead of needing to hand-write all our scripts, we can just describe the problems we're trying to solve, provide contextual files, and it can stitch everything together. It also grants a wider scope of accessibility to the breadth of technical knowledge in the industry, some of which is gatekept behind being in said industry. If there are things that we may not actually know how to do (because it's low-level or system specific), the agent has a deep knowledge resouvoir it can tap into. Then we can review it, restylize it by hand to understand it, and start iterating by hand from there. Having performance problems? The agent is good at finding shit and optimizing. Having trouble finding a bug? Describe it and the agent can look through the code and work out what the problem likely is, and it can even test the code and build isolated test cases to get things working as intended. It's a wonderful tool that saves a lot of mental strain. For people like me with limited attention and energy, it helps a fuck ton. -- Anyway, my point isn't to shill AI, it's your choice whether you use it or not. My focus here was more illustrating why AI is different for programmers versus artists. Ultimately, this moral grandstanding over other people's choice to use AI to assist in their work has become so overblown that it's genuinely irritating. If you refuse to play a game just because someone has interacted with AI in the course of development, then you might as well get mad at them for using Google, or stack overflow, or books instead of paying someone to solve their problem instead. Clearly we're so fucking deep into late stage capitalism, money always has to be a barrier to solving problems. And if your issue is with how much electricity and water is being consumed, push for all of that generative shit to be priced out in favor of AI as a tool for accelerating human hands in the creation of content. I agree that we should not be building compute centers in people's backyards, but the water consumption is a distration from things that are causing far more harm on the environment.
89
553
54
4,520
360,474
-Indie dev encounters issues at the end of their development cycle -indie dev attempts to troubleshoot said issues -indie dev consults the MANUFACTURER of their game engine, THEY can't solve their issues -as a last-ditch effort to avoid shipping a game with a glitch they can't iron out, they spitball with an LLM -surprisingly, LLM assists in workarounds that fix said issues -despite not needing to disclose (per steam TOS), the dev is honest about their disclosure when nominated for awards to be transparent. I would say House House did everything right. An LLM was used to solve problems humans could not. It didnt generate assets. It didnt infringe on the creative process. I hate what AI, and particularly GenAI has done to the gaming industry, but this is a stupid fucking hill to die on. This isn't vibe coding. This isn't slop. This is what AI was SUPPOSED to do in industry. Pick the right battles, people.
Big Walk used AI in 'the final stretch' of development, kicking off intense awards debate: aftermath.site/big-walk-ai-i…
Readers added context they thought people might want to know
House House used LLMs sparingly near the end of Big Walk's development, mainly to fix bugs in third-party code. They estimate well under 0.1% of the game's code was affected, and they didn't use generative AI elsewhere. bigwalk.game/faq/#ai pcgamer.com/games/puzzle/b…
233
5,630
122
51,973
886,723
NoGraphicsAPI is cool, but a bit of a shame to see the misleading AI usage since it seems almost everything was AI generated :/
PETAFAB just released on Steam at $9.99: store.steampowered.com/app/5… PC, Mac and Steam Deck/Machine. 90Hz on Steam Deck OLED. Up to 540Hz + 8000Hz late-latch mouse on high end PC for ultra-low latency. Let's see if you can beat my high scores :D
8
1
59
17,486
Especially given it's not like he hides AI usage on twitter so just being upfront would probably make it more appealing
1
9
1,767
#EmberGen Forward, isotropic, and backward spot light + multiple scattering.
3
8
88
3,043
Oliver retweeted
and the difference between the mediocre and best of the best programmers? The best ones never stop dreaming of designing a ground breaking programming language. The mediocre ones resign themselves to mediocrity. "implementing a PL for learning" as you might say.
Young programmers often think they want to design a ground breaking programming language. What they really want is to get far enough to be able to do something new. That being said implementing PL for learning is very good.
3
5
90
4,270
Sounds like a “you” problem Im full of joy writing my own compiler/PL from the ground up in C, without ai I’ve effectively replaced video games with recreational programming Software isnt dead, its just “dead” for those who only wanted a paycheck
Replying to @badlogicgames
pretty joyless now. guess i switch to gardening.
20
28
5
541
16,245
Oliver retweeted
Understanding is the noblest of pursuits. People who actively strive to understand are pushing against uncertainty. People who give up on understanding are willingly falling into chaos.
48
347
25
3,065
51,451
They called us mad men for doing it this way
I always thought it was a shame that modest default stack setups made alloca() a generally-frowned-upon technique in C/C++. With 64 bit architectures, give every stack a virtual address space equal to the physical memory by default! You can do a surprisingly good job of memory management with just the stack, and I contend that it has both correctness and performance advantages. Allocations are a trivial pointer adjustment, and frees are literally free. Keeping everything linearly packed is friendlier for all layers of the cache hierarchy, and it makes keeping everything in huge / gigantic pages easy. Leaving temporary intermediate blocks on the stack can seem wasteful, but unless you are dealing with algorithms that make > O(n) temporary allocations, it is usually not a big deal. Static analysis can check if you store a stack pointer into a structure deeper in the stack, which is a use-after-free, and then memory management correctness is guaranteed. The discipline of caller-allocates-return is awkward in some dynamic cases. It would take a change in low level calling conventions, but you could plausibly have functions that return with the stack extended from where it was when they were called. I’m not holding my breath for that. It isn’t obvious if having AI coding agents makes discussions of software architecture more or less important!
1
181
Maybe now that Carmack has said it people will stop doing malloc/free whack-a-mole that so many languages seem adamant on pushing you into
56
I'll believe software engineering is solved when society stop using j*vascript
Merging Bun AOT next week, then testing on Claude Code Expecting 2-3x less memory + CPU & a slightly smaller binary This is the future of server-side JavaScript Instead of you rewriting in Rust, we rewrote JavaScript's compiler to emit native code ahead of time
69
Oliver retweeted
What I think is important is to continue caring about the things you love despite anything. The storm will pass. The chaff will be blown away. Good work will continue standing.
41
262
11
2,993
47,138
EmberGen 2 LPV Cloud Rendering R&D EG 1 could never dream of this many voxels at this rendering quality. Oh how far we have come!
14
43
4
410
21,118
Still tweaking the combustion solver in EmberGen 2. A bit closer to photo-realism as well, particularly at the end.
8
8
172
6,119
I think we highly underestimate the slippery slope where even sensible applications can result in eventual vibe coding. As per my colleague 'its like having the devil on your shoulder whispering "do you want me to fix this?"'
The problem isn’t LLMs
6
4
330
16,008
Oliver retweeted
i remember early in my career wondering why most people missed out on building early internet products but i look around now and i get it so many people fiddling around with "my custom agent setup" and so few using this miracle of a tool to actually do anything
253
256
57
7,077
312,016