Ai can now generate 3d models on adobe

its crazy to watch AI developments just trivialize asset/content creation

where this train stops, i dont know

The end of scarcity itself. People free to pursue their desires and achieve more than ever for the benefit of all society. A utopia we never dared dream of.

b8426afd6110433ba846a2440d4f95ab.gif
 
its crazy to watch AI developments just trivialize asset/content creation

where this train stops, i dont know
Hopefully, it leads to more, better games in shorter periods of times as humans can concentrate on the creative aspects, without spending years on repetitive tasks.

Or it could lead to soulless games.

Likely, it will ultimately be a mixed bag of good and bad.
 
Not a chance, these are gaussian splats, not meshes, it's basically a point cloud with colour. 3D meshes are built out of vertices.
And even you were to somehow convert it to a mesh, the topology would be awful.
Epic's boss seems interested so we could see something similar from them, they already have some kind of generative environment in the engine
 
There's no mesh, so they are useless in 99% production pipelines. They are good for previz and that's about it because any motion will have tons of artifacts.

Also, photometry and stock models exist, so these type of tools are kinda pointless.

Just anther video overhyping AI and causing more misconception to the masses.
 
In 10 years this industry will look a lot different.

I really don't think AAA titles taking over half a decade to come out is going to be a thing in a few years tbh.
 
In 10 years this industry will look a lot different.

I really don't think AAA titles taking over half a decade to come out is going to be a thing in a few years tbh.
I don't mind AI if it's used carefully and not overbearing. If AAA development time can be shortened with AI than thats more than welcome to me.
 
Looks good. If it can improve quality, speed, and cut down on costs all the better.

I'd like to see how many of those artists who complain about AI avoid using it and stick to grassroots drawings all by themselves. I bet all the digital artists use Photoshop or whatever graphics program already. So this is really just an extension.
 
There's no mesh, so they are useless in 99% production pipelines. They are good for previz and that's about it because any motion will have tons of artifacts.

Also, photometry and stock models exist, so these type of tools are kinda pointless.

Just anther video overhyping AI and causing more misconception to the masses.
Not an expert here, but there was a recent video posted here about AI creating DOOM on the fly, without any 2.5 or 3d structual basis. Meaning that meshes arent necessarily required to create flowing 3d imagery. Of course doom is very low level tech in this example, but the precident for this becoming a full fledged technology could be in the future.
 
While i'll always be a tradionalist, its crazy how fast you can iterate ideas with gen. AI. I was messing around with some shit the other day; gen'd a few wolf man images then gen'd a 3d model from that. Topo is shit - which i expect to resolve over time - but it made an excellent base. Just "trace" with a box model for proper topo. Even with my old man, get off my lawn perspective, i can see pipeline and process usage cases.
 
Imagine doing any creative study 😂😂
I think there are two aspects with this:

1. Social media content (viral stuff)/low-effort media to make a quick buck without too much effort.
the above is going to destroy generative A.I. (this is already happening)

2. People using A.I to be even more competitive with higher quality and output: produce more (still requiring talent, effort, and knowledge)
the bar is rising exponentially fast, it's overwhelming, schools are way behind. "generalist" will rise again.
 
Still not implemented but I think this can be used for game assets
My experience is that it's not very viable for game assets or 3D printing, because the actual models don't look great without the textures. It uses a kind of photogrammetry so it has the same issues as photogrammetry models.
 
it's so bleak that the things that we thought Ai would be bad at doing like art ended up being the thing that it's the best at doing.
 
I'm all in for AI. Give it 10 years and a small team will be able to produce a game on Rockstar levels because the GPU go brrr.

To all the artists out there. Don't be Johh Henry, be the rock drill operator instead. Artists have a leg up because they understand composition and colour theory.
 
Extremely impressive. Hard to predict what and when, but the next 10 years are going to be nuts for game development.
 
Not an expert here, but there was a recent video posted here about AI creating DOOM on the fly, without any 2.5 or 3d structual basis. Meaning that meshes arent necessarily required to create flowing 3d imagery. Of course doom is very low level tech in this example, but the precident for this becoming a full fledged technology could be in the future.
the doom AI demo is not really a game engine. It is more of a simulation of the doom gameplay experience by generating and predicting frames, using previous frames as context. You still need an actual copy of doom to train it. It can't create a game out of thin air.
 
Top Bottom