It will be a full circle. U feed AI this
and get this
Than cash in.
PS: Starfield is 1st fully AI generated game, thats why its a soul-less husk, it was generated in 6 months, after Beth scratched their 7,5years of work, because it was a total flop.
Incorrect.
You will feed AI this:
"You are at the start of a hallway lit by torches and candle light. There are spider webs in corners. Some walls have rounded arches in them which may lead to secret passages. There's a very small window at the end through which a bit of daylight comes through. The stone walls have some damage in them with rocks scattered around"
I'm not sure I understand the point of this. The more you leave AI to figure things out on its own, the less persistent things become. It would kinda feel like living a dream when you look at something and it slowly distorts as you keep walking. Or as another example, every time you walk into a room, it looks a little bit different. Spoonfeed the AI too much and there's not much benefit in using it.
Meanwhile, it still has to render the assets, so, I don't quite get how it would reduce the computational load on the GPU.
@schmidtbag I actually think ai as co-pilot to a 3d artist can be a great tool .... instead of say spending hours to make a lake with lush vegetation .. you just tell the ai to make that and then you add in the small details you want .... making scenery in games in a lot of cases it is already for a lot of cases auto generated with out ai .... if anything ai will make it less repetitive if used right.... obviously there are devs that will turn the lazyness dial to the max !
@schmidtbag I actually think ai as co-pilot to a 3d artist can be a great tool .... instead of say spending hours to make a lake with lush vegetation .. you just tell the ai to make that and then you add in the small details you want .... making scenery in games in a lot of cases it is already for a lot of cases auto generated with out ai .... if anything ai will make it less repetitive if used right.... obviously there are devs that will turn the lazyness dial to the max !
Aren't there already tools that do this? Sure, maybe they're not quite to the same depth, but world development is still pretty fast.
Incorrect.
You will feed AI this:
"You are at the start of a hallway lit by torches and candle light. There are spider webs in corners. Some walls have rounded arches in them which may lead to secret passages. There's a very small window at the end through which a bit of daylight comes through. The stone walls have some damage in them with rocks scattered around"
Dont come with "incorrect" to me mate... If u think DLSS10 will be only about "text to image" than whatever. Games still need some structure, esp if u want 1 000 000 players to have similar experience. So you make a layout of a dungeon, simple, as shown in my image. Than you maybe put in some description like you said, and the result will be made by AI... So we are both right. Dont bring "incorrect" here...
Anyway, who knows. Hand crafted and AI generated, both will have its place.
Game maker:" Nvidia, make me a game that generates millions in revenue every year."
Nvidia DLSS 10 creates another Fortnite/ Minecraft/ Call of DUty clone...
Loot boxes and paid cosmetic items included of course.
DLSS will do the complete rendering process in a game.
I do not get how that works.
The point of DLSS is to fill in without spending time rendering, how is it ever going to fill in when there is no source material to work from?
It can not really make matching frames when there is nothing to match.
If it just builds a hallway from stock photos, then that is already possible, it is just done by the game dev at this point in time.
No matter what, the hallway still needs to be rendered on the screen and that still needs GPU power, unless it just makes a static non moving 2D hallway and changes it every time the player or something else moves.
I'm not sure I understand the point of this. The more you leave AI to figure things out on its own, the less persistent things become. It would kinda feel like living a dream when you look at something and it slowly distorts as you keep walking. Or as another example, every time you walk into a room, it looks a little bit different. Spoonfeed the AI too much and there's not much benefit in using it.
Meanwhile, it still has to render the assets, so, I don't quite get how it would reduce the computational load on the GPU.
I would think the AI does main work, but that devs continue to add their input or modify it to reflect their intended vision of the game. Its not like their work is done by simply giving the AI commands then calling it a day.
If good games can be made with an AI I dont mind.
A fucking shitton of bad games have already been made, with no help from a AI.
Bad games have been around as long as gaming.
Aren't there already tools that do this? Sure, maybe they're not quite to the same depth, but world development is still pretty fast.
Yes but this is literally the next step ! The evolution of it if you like you can give it more details ask for things more specific ! Ai will still give generic looking stuff but less generic if that makes sense !
Edit: obviously this is a tool and it can be used badly or be used to greatly increase quality and productivity!
I would think the AI does main work, but that devs continue to add their input or modify it to reflect their intended vision of the game. Its not like their work is done by simply giving the AI commands then calling it a day.
Venix:
Yes but this is literally the next step ! The evolution of it if you like you can give it more details ask for things more specific ! Ai will still give generic looking stuff but less generic if that makes sense !
Edit: obviously this is a tool and it can be used badly or be used to greatly increase quality and productivity!
I get the impression we may be talking about different things. I got the impression that this AI is to effectively replace traditional rasterization, as opposed to world design. So for example, everything you see is generated by AI in realtime where the details are artificially generated rather than rasterized though fixed models. Perhaps I'm wrong about that, but this is why I'm confused because it would require an exceptionally advanced (and power-intensive) AI to render a coherent, consistent, and persistent world in realtime.
As for what both of you are talking about (having an AI pre-generate the environment), sure, that's much more realistic and possible since you can just permanently save whatever the AI decides to do.
@schmidtbag I see , well my interpretation was like that cause I can not see anytime soon an ai model making a full fledged aaa game and come anywhere close to being good and not at best a bizarre mediocrity !
The AI developer combo will be great, particularly for indy devs. It takes a lot of effort to generate the assets and make a pretty game. For example a dev making an aliens game can basically describe what they want from say an alien and then point the AI at some Giger pictures for inspiration and out pops super high def realistic aliens.
and get this
Than cash in. PS: Starfield is 1st fully AI generated game, thats why its a soul-less husk, it was generated in 6 months, after Beth scratched their 7,5years of work, because it was a total flop.