It's heading in the right direction, but it's clearly not there yet.
That process uses solid colored images. If you attempt to provide something more complex like a photo you get very iffy meshes. For example: https://www.youtube.com/watch?v=HO1LYJb818Q
But it's definitely heading there. Think of any style asset that has a well-fleshed out collection of existing examples. Like painterly game crates.
Feed a database of a hundred painterly style game models to a machine learning algorithm, tell it to make a new one, and boom you have a new asset!
yeah! these are so cool! and think of all the new game design possibilities and UX. Also, I would love to have a 3D program that takes 2D concept art and block it out in 3D to be refined and iterated further. going back between sketching and 3D mockups will be so fast!
Replies
That process uses solid colored images. If you attempt to provide something more complex like a photo you get very iffy meshes.
For example:
https://www.youtube.com/watch?v=HO1LYJb818Q
But it's definitely heading there. Think of any style asset that has a well-fleshed out collection of existing examples. Like painterly game crates.
Feed a database of a hundred painterly style game models to a machine learning algorithm, tell it to make a new one, and boom you have a new asset!