Lots of new results on 3D model generation

New developments in ML for 3D model generation

This week I wanted to share some work I found cool in 3D model generation ML research. I also present some cool results I got by animating cyberpunk mugshots using an ML model that transfers motion onto images.

NVIDIA’s new AI model quickly generates objects and characters for virtual worlds | Engadget

GET3D could make it easier for developers to make games and VR experiences..

Engadget

Text to 3D

“Text to 3D model” algorithms let you type in a description and then get a 3D model of that thing back. This work was actually done by my friend Ajay Jain.

DreamFusion: Text-to-3D using 2D Diffusion

DreamFusion: Text-to-3D using 2D Diffusion, 2022.

DreamFusion: Text-to-3D using 2D DiffusionBen Poole

Animating Midjourney generations

In his recent tweet, my friend Blair Renaud shared a teaser for his upcoming VR game LOW-FI. He had previously generated hundreds of mugshots for inhabitants of his Noir Blade Runner inspired city and I used a motion-transfer AI model to animate them. They came out super cool and creepy.

That’s all for this week! If you find anything cool this week that you think I should see, please share it with me!

-John