And you could say that the difference is that high-level languages are deterministically transformed down, but in practice the compiler is so complex you'd have no idea what it's doing and most people don't look at the machine code anyway. You may as well take a look at the LLM's prompt and make an assumption of the high-level code that it spits out.
Crazy to have a company like this with perhaps the most revolutionary product in the last 20 years decide that they want to pivot to porn and ads. Whose idea is this? They're supposed to be the superintelligence company. There's not even that much money in it. What about automating jobs, making a model that can predict the stock market, anything that would meet the hype?
Even if they're different people, they must be a vocal minority since the games that they seem to complain about tend to be the ones that sell the most copies
It's a video game engine. It's got a ton of optimisations and tweaks to make it run in realtime, but if you're making a movie there's no reason not to spend hours rendering each frame. You don't need to optimise meshes at a distance, or use real-time raytracing with noise reduction rather than just simulating a thousand bounces or limit yourself to 4K textures, you can use as many particle effects and simulations as you'd like. You can't do this with a game engine though - Unreal does have the ability to render out video but it's not going to be the same fidelity.
I didn't think they were actually using the video straight out of the Volume though - my assumption was they'd just use it to make sure the lighting reflected on to the actors nicely and then redo the CGI elements with something else.
>> It's got a ton of optimisations and tweaks to make it run in realtime, but if you're making a movie there's no reason not to spend hours rendering each frame.
That's how it's used though? It only runs real time for preview, but the final product is very much not rendered in real time at all. Obviously it's a very flexible tool that works with what you need and what you can afford - Disney runs it on their compute farm and they can throw the resources at it to render at the fidelity required. But obviously there are plenty of production houses which don't have those kind of resources and they have to make do with less. But then you wouldn't expect Pixar's own pipeline to work in those circumstances, would you.
>> Unreal does have the ability to render out video but it's not going to be the same fidelity.
I really encourage you to look into what's possible with UE nowadays. Custom made pipelines from Pixar or Dreamworks are better still, of course, but UE can absolutely stand next to them.
The problem is the way surface lighting/scattering is calculated, which does not match what traditional offline renders do.
My issue with UE is the opposite, the engine went too far into cinema production, and making it a performant game engine requires code refactoring. At which point an open-source engine might be a better choice. Its a mix of two (three) worlds, and not the best choice for one specific use.
For what is actually hard to do, like character animation, UE is a good choice. The lighting can be replaced more easily than the animation system.
> At the end, movies are about the stories, not just pretty graphics.
The great people at Pixar and DreamWorks would be a bit offended. Over the past three or so decades they have pushed every aspect of rendering to its very limits: from water, hair, atmospheric effects, reflections, subsurface scattering, and more. Watching a modern Pixar film is a visual feast. Sure, the stories are also good, but the graphics are mind-bendingly good.
Yes definitely. The average quality of Github projects, startup code may improve but there will probably not be many more stars on them or successful startups because it's immaterial and so easy to switch, the best project to do X becomes 100x more successful than the second-best project.
There are a lot of people who still say that coding agents don't work at all, it's all a NFT-style fad or scam pushed mainly by bad-faith hucksters looking to get a quick buck, etc, so it's refreshing to read something arguing otherwise - and this is antirez who created Redis, so someone who can speak from experience.
Would it make more sense to instead train a model and tokenise the syntax of languages differently so that white space isn’t counted, keywords are all a single token each and so on?
After watching models struggle with string replacement in files I've started to wonder if they'd be better off in making those alterations in a lisp: where it's normal to manipulate code not as a string but as a syntax tree.
reply