Comment by pjmlp
3 days ago
Being an old dog that still cares about gaming, I would assert many games are also not taking advantage of current gen hardware, coded in Unreal and Unity, a kind of Electron for games, in what concerns taking advantage of existing hardware.
There is a reason there are so many complaints in social media about being obvious to gamers in what game engine a game was written on.
It used to be that game development quality was taken more seriously, when they were sold via storage media, and there was a deadline to burn those discs/cartridges.
Now they just ship whatever is done by the deadline, and updates will come later via a DLC, if at all.
They're both great engines. They're popular and gamers will lash out at any popular target.
If it was so simple to bootstrap an engine no one would pay the percentage points to Unity and Epic.
The reality is the quality bar is insanely high.
It is pretty simple to bootstrap an engine. What isn’t simple is supporting asset production pipelines on which dozen/hundreds of people can work on simultaneously, and on which new hires/contractors can start contributing right away, which is what modern game businesses require and what unity/unreal provide.
Unreal and Unity would be less problematic if these engines were engineered to match the underlying reality of graphics APIs/drivers, but they're not. Neither of these can systematically fix the shader stuttering they are causing architecturally, and so essentially all games built on these platforms are sentenced to always stutter, regardless of hardware.
Both of these seem to suffer from incentive issues similar to enterprise software: They're not marketing and selling to either end users or professionals, but studio executives. So it's important to have - preferably a steady stream of - flashy headline features (e.g. nanite, lumen) instead of a product that actually works on the most basic level (consistently render frames). It doesn't really matter to Epic Games that UE4/5 RT is largely unplayable; even for game publishers, if you can pull nice-looking screenshots out of the engine or do good-looking 24p offline renders (and slap "in-game graphics" on them), that's good enough.
The shader stutter issues are non-existent on console, which is where most of their sales are. PC, as it has been for almost two decades, is an afterthought rather than a primary focus.
No, that's not the reason.
The shader stutter issues are non-existent on console because consoles have one architecture and you can ship shaders as compiled machine code. For PC you don't know what architecture you will be targeting, so you ship some form of bytecode that needs to be compiled on the target machine.
5 replies →
Any search on game console reviews on YouTube will show otherwise, even though it isn't as bad as PC.
Just a quick search,
https://www.gamesradar.com/games/the-ps5-stutter-issue-is-re...
https://forums.flightsimulator.com/t/stutters-on-xbox-series...
1 reply →
If anything I think PC has been a prototyping or proving grounds for technologies on the roadmap for consoles to adopt. It allows software and hardware iterations before it's relied upon in a platform that is required to be stable and mostly unchanging for around a decade from designing the platform through developers using it and recently major refreshes. For example from around 2009 there were a few cross platform games with the baseline being 32bit/DX9 capabilities, but optional 64bit/DX11 capabilities, and given the costs and teams involved in making the kind of games which stretch those capabilities I find it hard to believe it'd be one or a small group of engineers putting significant time into an optional modes that aren't critical to the game functioning and supporting them publicly. Then a few years later that's the basis of the next generation of consoles.
You know the hardware for console so you can ship precompiled shaders.
Can't do that for PC so you either have long first runs or stutter for JIT shader compiles.
2 replies →
Imagine living in a reality where the studio exec picks the engine based on getting screenshots 3 years later when there's something interesting to show.
I mean, are you actually talking from experience at all here?
It's really more that engines are an insane expense in money and time and buying one gets your full team in engine far sooner. That's why they're popular.
Pretty much it.