Microsoft has created an AI-generated version of Quake
-
This post did not contain any content.
-
System shared this topic
-
This post did not contain any content.
-
This post did not contain any content.
Nobody wants this
-
This post did not contain any content.
The tech demo is part of Microsoft’s Copilot for Gaming push, and features an AI-generated replica of Quake II that is playable in a browser. The Quake II level is very basic and includes blurry enemies and interactions, and Microsoft is limiting the amount of time you can even play this tech demo.
Microsoft is still positioning Muse as an AI model that can help game developers prototype games. When Muse was unveiled in February, Microsoft also mentioned it was exploring how this AI model could help improve classic games, just like Quake II, and bring them to modern hardware.
Okay, here's a much-less ambitious use of existing AI technology that I think would be vastly-more-useful than whatever they're off doing: how about just going out and using existing AI upscaling techniques and limited human interaction to statically-upscale the textures by maybe 2x to 4x, take advantage of more VRAM on newer hardware?
-
This post did not contain any content.
But, we already have quake.
-
The tech demo is part of Microsoft’s Copilot for Gaming push, and features an AI-generated replica of Quake II that is playable in a browser. The Quake II level is very basic and includes blurry enemies and interactions, and Microsoft is limiting the amount of time you can even play this tech demo.
Microsoft is still positioning Muse as an AI model that can help game developers prototype games. When Muse was unveiled in February, Microsoft also mentioned it was exploring how this AI model could help improve classic games, just like Quake II, and bring them to modern hardware.
Okay, here's a much-less ambitious use of existing AI technology that I think would be vastly-more-useful than whatever they're off doing: how about just going out and using existing AI upscaling techniques and limited human interaction to statically-upscale the textures by maybe 2x to 4x, take advantage of more VRAM on newer hardware?
I mean the demo capability isnt bad, could test out 100 ideas before you commit. But thats not what this is going to be used for. Its going to be a weapon against developers eventually.
-
The tech demo is part of Microsoft’s Copilot for Gaming push, and features an AI-generated replica of Quake II that is playable in a browser. The Quake II level is very basic and includes blurry enemies and interactions, and Microsoft is limiting the amount of time you can even play this tech demo.
Microsoft is still positioning Muse as an AI model that can help game developers prototype games. When Muse was unveiled in February, Microsoft also mentioned it was exploring how this AI model could help improve classic games, just like Quake II, and bring them to modern hardware.
Okay, here's a much-less ambitious use of existing AI technology that I think would be vastly-more-useful than whatever they're off doing: how about just going out and using existing AI upscaling techniques and limited human interaction to statically-upscale the textures by maybe 2x to 4x, take advantage of more VRAM on newer hardware?
https://store.steampowered.com/app/993090/Lossless_Scaling/
That's kind of what this is for.
-
Nobody wants this
Wait until Preston shows up with his AI generated radiant quests before you judge.
-
Nobody wants this
-
The tech demo is part of Microsoft’s Copilot for Gaming push, and features an AI-generated replica of Quake II that is playable in a browser. The Quake II level is very basic and includes blurry enemies and interactions, and Microsoft is limiting the amount of time you can even play this tech demo.
Microsoft is still positioning Muse as an AI model that can help game developers prototype games. When Muse was unveiled in February, Microsoft also mentioned it was exploring how this AI model could help improve classic games, just like Quake II, and bring them to modern hardware.
Okay, here's a much-less ambitious use of existing AI technology that I think would be vastly-more-useful than whatever they're off doing: how about just going out and using existing AI upscaling techniques and limited human interaction to statically-upscale the textures by maybe 2x to 4x, take advantage of more VRAM on newer hardware?
There is a lot of potential for generating on the fly skins and specs. Everyone seems to want to do stupid flashy junk, but a well configured agentic setup could easily get constrained in ways that alter geometry and interactions more dynamically than the broad scope nonsense I keep seeing.
-
This post did not contain any content.
“We’ve talked about game preservation as an activity for us, and these models and their ability to learn completely how a game plays without the necessity of the original engine running on the original hardware opens up a ton of opportunity.”
No, I don't think that you're talking about preservation then. Not even game emulation. You're talking about game hallucination.
-
This post did not contain any content.
Thank god, I've been waiting for this my whole life /s
-
https://store.steampowered.com/app/993090/Lossless_Scaling/
That's kind of what this is for.
That's frame scaling in real-time, rather than offline texture scaling.
-
This post did not contain any content.
We have Quake at home
-
This post did not contain any content.
Hit detection doesn't work, going into pure darkness makes the AI hallucinate to the point it distorts the map layout, npcs will teleport around ot just disappear... runs at like 640p at maybe 20 fps, textures are a blurred indistinct mess...
Oh and it requires basically a super computer to run this.
Brilliant.
-
“We’ve talked about game preservation as an activity for us, and these models and their ability to learn completely how a game plays without the necessity of the original engine running on the original hardware opens up a ton of opportunity.”
No, I don't think that you're talking about preservation then. Not even game emulation. You're talking about game hallucination.
Yeah it's the difference between playing an old game today and remembering what it was like to play in the past. Not the same thing at all.
Plus a big part of old games these days is decompiling it, so you can recompile to run with higher framerates, higher resolution and without emulation. It's also possible to add nice QoL features and entire new game modes.
Just look at what Ship of Harkinian has done with OOT. It looks great, it feels like OOT still, but has the nice quick buttons. And if you want to experience the game like it's brand new, there is the randomizer. And similar projects exist for other old games.
And there's also people going through the code, figuring out glitches. And how certain mechanics worked, nobody understood very well back in the day. Discover Easter eggs that were never found.
That's game preseveration, not some AI fever dream if you squint a bit it kinda sorta looks like the old game.
A lot of the AI stuff I've seen from Microsoft also sucks hard and they know it. But they operate under the assumption these LLM systems will get better and better. Like this game thing they admit it sucks now, but imagine what it could be one day. However the reality seems to show more and more the point of rapidly diminishing returns has been reached. Throwing more data and processing at the thing isn't going to make it a lot better.
They are also so busy inventing new AI features nobody wants. Putting new flashy buttons everywhere and doing awful tech demos. They completely forgot to make actual useful features. For example a thing that happens a lot when working with less computer capable people, is people sending screenshots of Excel data. How awesome would it be if instead of helping write a new signature, the AI would go: "Wow what an asshole, sending a screenshot like that. Here is the original data so you can copy paste.". Or when trying to send an email without the attachment that really should have an attachment, it warns you. It already does this, but I think it just triggers on certain keywords like attach. This would be an excellent use case for an LLM, where it doesn't even matter much if it's wrong some of the time.
For me personally "AI" in the form of LLM can fuck all the way off. It certainly has it's uses, but this all in use it everywhere for everything has made me hate it. And the misleading marketing making people think it's basically AGI is wrong on so many levels.
-
This post did not contain any content.
Copy+paste is still a pain in the ass in Microsoft Teams. Why don't you work on that instead?
-
Yeah it's the difference between playing an old game today and remembering what it was like to play in the past. Not the same thing at all.
Plus a big part of old games these days is decompiling it, so you can recompile to run with higher framerates, higher resolution and without emulation. It's also possible to add nice QoL features and entire new game modes.
Just look at what Ship of Harkinian has done with OOT. It looks great, it feels like OOT still, but has the nice quick buttons. And if you want to experience the game like it's brand new, there is the randomizer. And similar projects exist for other old games.
And there's also people going through the code, figuring out glitches. And how certain mechanics worked, nobody understood very well back in the day. Discover Easter eggs that were never found.
That's game preseveration, not some AI fever dream if you squint a bit it kinda sorta looks like the old game.
A lot of the AI stuff I've seen from Microsoft also sucks hard and they know it. But they operate under the assumption these LLM systems will get better and better. Like this game thing they admit it sucks now, but imagine what it could be one day. However the reality seems to show more and more the point of rapidly diminishing returns has been reached. Throwing more data and processing at the thing isn't going to make it a lot better.
They are also so busy inventing new AI features nobody wants. Putting new flashy buttons everywhere and doing awful tech demos. They completely forgot to make actual useful features. For example a thing that happens a lot when working with less computer capable people, is people sending screenshots of Excel data. How awesome would it be if instead of helping write a new signature, the AI would go: "Wow what an asshole, sending a screenshot like that. Here is the original data so you can copy paste.". Or when trying to send an email without the attachment that really should have an attachment, it warns you. It already does this, but I think it just triggers on certain keywords like attach. This would be an excellent use case for an LLM, where it doesn't even matter much if it's wrong some of the time.
For me personally "AI" in the form of LLM can fuck all the way off. It certainly has it's uses, but this all in use it everywhere for everything has made me hate it. And the misleading marketing making people think it's basically AGI is wrong on so many levels.
LLMs have been hijacked by fucking snake oil salesmen. This is so tremendously tragic.
The good thing is that they will give up eventually if the demand won't settle at high enough level because of the upkeep costs.
The bad thing is that forcing the demand seemingly began taking root. It's anecdotal, but I've met several non tech people who prefer LLM generated babble over looking for answers on sites/manuals/docs by themselves. It's scary, but I hope that soon hallucinations will be too common to rely on LLMs even with unimportant questions.
-
This post did not contain any content.
-
This post did not contain any content.
Old school gamer here. Headline should definitely say Quake II.
There might not seem to be much difference to a casual observer, but from that standpoint there's not much difference between either and any other FPS. Even Minecraft to some extent.
Speaking of which, the Minecraft equivalent to this had all the same problems outlined in other comments here. Interesting as a proof of concept, but there are almost certainly better ways of using AI.
