Point-E is a machine learning system that creates a 3D object given a text prompt in one to two minutes on a single Nvidia V100 GPU, in contrast to text-conditional 3D object generation, which can take multiple GPU-hours to produce a single sample!
This is the kind of tech that will radically accelerate collaboration and expression — by greatly accelerating the the process of converting text prompts into 3D object approximations.
What can you do with these? Do these produce files that can be fed into a 3D printer? (Or is it not at that stage yet?)
This is the first time I've seen a Product Hunt launch whose website link was an arXiv paper…
Report
Impressive! Point-E is a revolutionary machine learning system that enables quick and efficient 3D object generation from text prompts, drastically reducing the time it takes to produce a single sample compared to traditional methods.
Report
It's amazing how much faster Point-E can produce 3D objects compared to text-conditional generation. Congratulations on your launch!
Great job on the launch of Point-E! This machine learning system is definitely a game-changer and will save a lot of time and effort for those in the 3D object generation field.
Report
Congrats on the launch! This will be interesting to follow, the 3D images produced by nVidia's own model were really impressive!
Osaurus
Versus
SocialBu
InterSub