The field has progressed quite a lot, there are now open source models that can do this generation locally. I'm experimenting with Hunyuan3D-2 which was trained by Tencent, Microsoft has Trellis.
one could 1. take a picture/draw a picture 2. tell another AI system to "rotate 360" getting a image for each frame 3. use those images for photogrammetry 4. use this AI for modeling over that photogrammetry so much work so quickly done, its dangerous
I recently tried meshy to generate some D&D minis, but in the end I decided to just stick to Hero Forge.
What did they use to coat the penguin and give him a shiny finish? 5:53
I have found that Tripo is pretty amazing at turning 2D models into 3D
Is this a paid promotion for meshy? It really feels like it. The two samples -- a shark and a penguin -- are kind of laughably simplistic as test cases. And FWIW this looks like a wrapper of Tencent's Hunyuan3D v2.
I recommend simply creating the image you want to make using text to img like Bing Creator then img to 3D like Makerlab
It's "great" if you don't care what you get - which is kinda rare tbh. A novelty at best.
I like the older penguin, that's so funny
"We are almost to the the point where the A.I. will create absolutely perfect 3D models of anything in a matter of minutes. We are entering an exciting future..." How is that exciting? When AI is making everything within minutes nothing is going to be cool anymore. Got an amazing model you made with AI and want to show it off? No one will care because every other man and his dog is doing the same thing. Same with images, films, music etc. Just more cheaply made generic AI trash to flood the internet with. Making something with AI is about as exciting as doing maths with a calculator.
Love this tools and how fast they progress. I hope they will move as fast as image and video AI learned, it's for sure getting better and better. As an end user I plan to use them to the moon and back - the entitlement and aggressiveness of "real artists" that I have seen is astonishing to the point it became off-putting (in all fairness they are right to be afraid about being replaced or demand going from bad to worse - the same happened to illustrators)
Great video.
Glad my first degree was in IT and I'm good with electronics (soldering, reballing, chip programming, etc)... my post-grad in CAD is in real trouble!
Woot woot!
Pure slop, coming to a another asset store near you.
I'm not sure I quite understand the whole complaint about AI was trained using copyrighted material. Humans look at copyrighted material all the time and get inspiration from it or learn from it. So if it's wrong for AI to learn from copyrighted material, then by that logic, humans should also stop looking at anything copyrighted as well because we might be inspired to create something similar ourselves right?
Gave meshy a try over the last few weeks, and it got better, but not for my usecases. Buildings all look very round and fantasy-like and are not generated very clean. Thats why its good for fantasy stuff I guess, why think about what creature you want for your new game if you can just type in "whale" and you get a whale with mutations
Neat! When will they implement it into printers themselves? Soon I bet!
There's no use for 3D printing and printers for majority of people, better to pay for end products that you need in some rare occasions instead of playing with 3D printing like a toy.
@nickrowan