inclementimmigrantOwlcat is using generative AI for The Expanse Osiris Reborn, but the final game will be "100% human made"
My question is if everything is going to be human made in the end why bother using at all? You won’t even get any of the much vaunted time savings at that point.
It is conceivable (though I certainly understand skepticism) that they use it for concept and placeholder art, proofs of concept and the like.
As always, the question should be whether the final product is any good.
The question should be whether the final product is worth what was sacrificed to make it. That line is different for everyone, but it’s important to keep that in mind. Plenty of companies I boycott make acceptable products but are supporting a genocide.
I don’t think generative AI use is worth it however it’s employed and regardless of the quality of the final product. If enough people agree maybe they’ll stop using it.
For instance. You can try things out without first creating them by hand. Then you pick and choose and make the final version by hand.
kazernielEvery other commenter under this seems to forget that stock assets exist and worked fine for decades without involving AI slop.
Stock assets (at least if you need more than the absolutely basics) cost quite a bit. Programmer art can work, but if you want something close to the tone of the finished product, still takes time and thus money. Slop is quick and free.
Frankly, given the fact that placeholder assets are literally meant to be utilitarian, disposable, “just good enough” work, it’s actually not a terrible use case. Placeholders are meant to be slop either way, so not much is lost by automating it, so long as it is actually removed after.
Placeholder assets are generally better if they look out of place because then you don’t forget to replace them 😅
AI art generation is trained to be just good enough to fly under the rader if not looked at too closely…
It depends on what you are using placeholder assets for. If you want to use it to gauge how a scene would look before setting out to build it, then placeholders that stand out get in the way. You would need a way of tracking all the slop, but then you could have a build tool track how much slop is still in the game to make sure you catch it all before release.
Depends on the use case. If its just to be a piece to fill the spot and nothing else, yes. That said, assets impact tone and gameplay, and if you’re trying to judge how something will feel or play, then sometimes you need something closer to the given use case. For example, if you have a survival horror game and are trying to judge the ambiance and visibility of an in-progress level, using wildly out of place assets will mess with the tone, and may result in difficulty in judging factors like the visibility of gameplay elements.
Like was said before, the same role as stock assets and programmer art.
That s not entirely true , you may want to see what it would look like and big cube purple and black are not ideal for that.