ChatGPT-maker brings AI-powered Point-E, text-to-3D image creator

Open AI, the maker of the popular Artificial Intelligence (AI) powered bot ChatGPT has come up with a new AI model that can churn out 3D images with text commands.

The company is calling it 'Point-E' and kind of apt name, given how the images have squarish pointy ends.

The interesting thing about the Point-E 3D image generator is, it uses less power compared to the current crop of AI models. The latter needs several Graphics Processor Units and takes several hours of processing to come up with a 3D image, with the Point-E, the Open Ai team was able to create an advanced model, that needs just one GPU and a few seconds to convert text to 3D images.

3D image samples created from text prompts. Credit: Open AI research paper

"Our method first generates a single synthetic view using a text-to-image diffusion model and then produces a 3D point cloud using a second diffusion model which conditions on the generated image," the Point-E team noted in the research paper.

Though it is not as realistic as 2D pictures generated by DALL-E, Point-E delivers 3D images. With improvements in machine learning algorithms, it can be refined to create believable 3D models.

3D images created by the Point-E model with just text prompts. Credit: Open AI

Must read | AI photo generator DALL-E churns out frightening end-of-the-world selfies

With such technology, creative artists and filmmakers can create more realistic images in movies, 3D gaming and animations. Also, it can be put to good use in virtual platforms such as Metaverse and even in scientific studies and medical education too.

Also read | ChatGPT storms social media with its smart repartee

Get the latest news on new launches, gadget reviews, apps, cybersecurity, and more on personal technology only on DH Tech.