Comment on Anti-doom-and-gloom post
monotremata@kbin.social 4 months ago
I think it's going to get even better in the next few years, too. The tools for 3d modeling are poised to improve in a way that makes it dramatically easier to create very high quality graphics. Nanite is one component of this, reducing the need for multiple levels of detail in polygon-based rendering. But 3d reality capture is improving too, both thanks to hardware like depth sensors and software like Gaussian splatting and NeRFs.
Indie games are just going to keep getting better, basically. As will AA games. I think the days of the AAA blockbuster may be numbered.
CleoTheWizard@lemmy.world 4 months ago
I think the only way that large studios continue to justify their existence is in motion capture acting and detailed open world design. Everything else is now within reach of indies I think.
Don’t just think of graphical improvements either. We can’t leave AI out of this discussion as it is a valuable tool for indie devs and I expect they’ll take advantage of it in the next few years. They aren’t going to replace main stories but I think AI will take the burden off of indie studios for simple dialogue options. Could even be trained to voice act for lesser roles or do some texture work.
monotremata@kbin.social 4 months ago
I kind of disagree about AI, I guess.
I do think it's a valuable tool, but honestly there's not a ton that it does that you couldn't already do with an asset store. And there's a fair amount of risk associated with using AI in the near term. Folks already have a lot of qualms about the ethics of how those AIs were trained. And the first games that come out that rely heavily on AI are likely to be really janky--there are devs who will have tried to entirely replace a role on the team with AI, and the quality will suffer as a result. So I think in the near term there's going to be a pretty severe backlash against AI-generated stuff in games. Folks will say it all feels generic and low-effort; it'll be the new "asset flip."
Long-term, I think it will have a place in the workflow for sure, the same way that store-bought assets do; you'll just need to adapt them to fit in with the feeling you're going for in your game, and hand-revise some things. But near-term, I think there will be a lot of folks who lose interest in a game if they find out there's AI involved. And that goes triple for AI voice acting. A bad human voice actor can at least be interesting, but AI has that uncanny valley quality that really turns people off once they notice it.
CleoTheWizard@lemmy.world 4 months ago
You’re not wrong about a lot of these points but I think it’s worth pointing out that each AI application will basically be a separate tool.
For texture work, these tools will be used immediately and probably only have beneficial effects. It will even give premade assets a nice feel AND you can use AI to turn one hand-made texture into a large amount of unique assets. Which helps indies by avoiding the copy/paste feel of their assets. Same is true of models but will take more time for that.
As for voices, what you said is true. A bad human voice is still mostly better than AI. However if you have a lot of dialog or just want to play test, the AI voices are great for that before you ask a human to do the lines.
And as for story and script dialog, AI is already pretty good at this but the application is important. Games that overuse AI to write their story or characters will stand out as really bad. But when used correctly, AI can change generic filler into interesting plot. Or make people say unique things every time you see them. Combine that with voice generation once it’s good and we’re in for some very cool applications.
monotremata@kbin.social 4 months ago
Yeah, I agree about the textures, but I think you're overestimating the existing LLMs. I think folks are already starting to recognize the style of the current LLMs and finding it off-putting. I think that's only going to increase as people try to apply them in even more places.