Exactly how helpful can AI be to games development?
A Reader’s Feature offers a very optimistic view of how generative AI might be used in video game development in the future.
One reader tries to imagine the best case scenario for using AI to make video games, cutting down the time and effort needed by developers.
There are many out there who are terrified about the possibility that AI will take over a large segment of the workforce, but I argue that the gaming industry (as well as many others) is struggling with the cost of making games; we’ve seen many studios having to lay off staff and ‘restructure’ to cut costs. And the biggest cost by far is just creating the world!
With growing hardware power, it has become more and more difficult to create a graphically beautiful world and characters. Hundreds of development hours are spent creating a simple landscape, realistic physics, and trying to make sure there are no glitches, like holes in the universe players can fall into.
More and more often we’ve seen AAA games come out that are so glitchy they may as well be beta releases. This has led many studios to start releasing games still in development as early access versions, allowing players to hunt out the glitches without backlash from the community.
Some even turn to releasing retro style games which are incredibly low resolution and low power so the devs can spend more time just creating a good story, side quest lists, and multiple endings based on player choices (in some cases the multiple endings are only slightly different, if at all).
Now we’ve seen generative AI create photorealistic videos in under a minute, so imagine one designed to take a small amount of basic input and generate or populate an entire landscape, then the developer either flies over and/or runs through the area to see how it looks and picks up on anything that doesn’t match their vision, removing anything they don’t like or highlighting sections they want changed and getting the AI to modify the section.
Or imagine the developer creates the basic outlines of a village or city, then feeding the AI some concept images of that they want the village to look like and getting the AI to completely create it. Then, as before, the developer can go through the village and make alterations where needed or wanted.
I can imagine a day when developers with VR headsets can literally create a world while standing in it, looking at a group of buildings and asking the AI to turn one into a tavern, then changing the building material to logs, then changing the roads from cobble stone to dirty and make it muddy.
Now show me what it will look like at night! Right, add some touches outside the tavern. Decrease internal lighting. OK, back to daytime and show me some characters that match the era. OK, make their clothes more dirty and tattered. OK, that’s good, now populate the tavern, roughly 30 …read more
Source:: Metro
