Looks cool, but from my personal experience, I’ve tested a lot of AI models like this (because I love technology), and they never really do what’s shown in the promo videos. It always feels more like a vision of the future. That last video with the low poly scene is probably a more accurate representation of what it can actually do, compared to the top ones.
Well I guess these things have to start somewhere but is is not much use as it is. Part of the problem is what they are aiming for and for who?
If they set a target like merely making low poly game useable results, they might make something of use to that group. People with zero ability or interest in Blender basically, just wanting things to give an AI game making version the items to make the game for them with. Getting to the point of a game player being able to tell an AI make me a game in this genre and then just play it. All the industry disposed of. Which is not a bad thing in itself, that is what progress has done to every craft, industry, or work, or we would be out there in the fields still. Or dragging big rocks about to praise the gods!
Nice idea. That could be a great entry point for some people. Right now, we’re still in the phase where flashy models are used to grab attention and attract investment money. When things slow down, maybe we’ll see more useful models that are actually helpful. There are already some promising tools out there, like magnific.ai. The Indie 3D team talked about it at the Blender Conference, how they train their artists to improve scene lighting quickly. Also curious to see what the Blender team comes up with. If I remember right, they mentioned working on something AI-related to help artists.
Personally, I would never buy a game made by AI, but some people do. I think to get the best of both worlds, anything AI was involved in should be clearly labeled, like “Made with AI.” That way, everyone can choose for themselves.