SvennoJ said:
Soundwave said:
I think what they will try in the long term future is a new setup of graphics generation, the graphics are no longer polygons/textures but based on AI picture data that AI TMUs build an environment and characters from. For now they're going to basically price people out of physical hardware.
Maybe only the immediately character you're controlling is rendered and a few enemies is rendered locally on your device.
That environment then is effectively "streamed" to your device allowing for what they'll term as "photorealism". If you have a bad internet connection, you'll get environmental artifacts and lag but the main gameplay should still be alright because it's rendered locally. But in this scenario, you no longer own the hardware yourself and have to pay essentially for GPU/TMU usage.
Purists will probably have a shit fit, but it will allow things like smartphones even to have games that look photorealistic "enough". |
That would be another death blow to PC gaming. The biggest draw of PC gaming is the freedom to spice things up with mods next to cheap sales. The whole PC gaming is affordable spiel comes from cheaper games. Not renting them from the cloud.
Streaming has already been tried on PC and rejected.
And purists won't like the extra lag, streaming artifacts etc. |
Well it "failed" when local hardware was affordable. When local hardware becomes $3000-$5000 just for the GPU, that prices out a lot of people.
The mod thing would be interesting on AI neural based games because they could allow the user to alter game elements on the fly as they please without even having to sit around waiting for modders, even making games based on prompts.
Although probably a ways down the line, but if your graphics pipeline is no longer polygons/textures created by people but rather just training data/photos/videos, you can alter those things much more easily.
