Startup renders 3D product models for top retailers


Creating a retailer’s online catalog used to require expensive physical photo shoots to capture products from every angle. A Tel Aviv startup saves brands time and money by turning those camera clicks into mouse clicks.

Hexa uses GPU-accelerated computing to help businesses transform their online inventory into 3D renderings that shoppers can view in 360 degrees, animate, or even try on virtually to support their purchasing decisions. The company, which recently announced a $20.5 million funding round, works with brands across fashion, furniture, consumer electronics and more.

“The world is going 3D,” said Yehiel Atias, CEO of Hexa. “Just a few years ago, the digital infrastructure for this was so expensive that it was more affordable to organize a photographer, models and lighting. But with the advances of AI and NVIDIA GPUs, it is now possible for retailers to use synthetic data to replace physical photo shoots.”

Hexa’s 3D renderings are used on major retail websites such as Amazon, Crate & Barrel and Macy’s. The company creates thousands of renderings each month, reducing the need for physical photoshoots for every product in a retailer’s catalogue. Hexa estimates it can save customers up to £300 in carbon emissions for each product depicted digitally rather than physically.

From physical photo shoots to AI accelerated renderings

Hexa can reconstruct a single 2D image or a series of low quality 2D images into a high fidelity 3D asset. The company uses varying levels of automation for its renderings, depending on the complexity of the shape, the amount of visual data to be reconstructed, and the object’s similarity to Hexa’s existing dataset.

To automate elements of its workflow, the team uses dozens of AI algorithms developed with the deep learning framework PyTorch, running on NVIDIA Tensor Core GPUs in the cloud. For example, when one of Hexa’s artists reconstructs a 3D toaster, an algorithm can identify similar geometries the team has created in the past to give the creator a head start.

Another neural network can scan a retailer’s website to determine how many of their products Hexa can support with 3D renderings. The company’s entire rendering pipeline also runs on NVIDIA GPUs, which are available through Amazon Web Services.

“Accessing compute resources through AWS gives us the ability to use thousands of NVIDIA GPUs at any time,” said Segev Nahari, lead technical artist at Hexa. “If I need 10,000 frames by a certain time, I can request the hardware I need to meet the deadline.”

Nahari estimates that rendering on NVIDIA GPUs is up to three times faster than relying on CPUs.

Expand beyond retail, into the omniverse

Hexa developers are constantly experimenting with new methods of 3D rendering and are looking for workflow improvements in pre-processing, object reconstruction, and post-processing. The team recently started working with NVIDIA GET3D, a generative AI model from NVIDIA Research that generates high-fidelity three-dimensional shapes based on a training data set of 2D images.

Sneaker generated by GET3D
By training GET3D with Hexa’s shoe dataset, the team was able to generate 3D models of novel shoes that were not part of the training data.

In addition to its work in e-commerce, Hexa’s research and development team is investigating new applications for the company’s AI software.

“It doesn’t stop at retail,” Atias said. “Industries from gaming to fashion to healthcare are finding that synthetic data and 3D technology is a more efficient way to, for example, digitize inventory, create digital twins and train robots.”

The team describes its membership in NVIDIA Inception, a global program that supports cutting-edge startups, as a “great asset” in improving the technology used by Hexa.

“Being a part of Inception opens doors that outsiders don’t have,” Atias said. “For a small company trying to navigate the vast array of NVIDIA hardware and software offerings, it’s a door opener to all the cool tools we wanted to experiment with and understand the potential they could bring to Hexa.”

Hexa is testing the NVIDIA Omniverse Enterprise platform – an end-to-end platform for building and running Metaverse applications – as a tool to unify its annotation and rendering workflows used by dozens of 3D artists around the world become. Omniverse Enterprise enables geographically dispersed development teams to customize their rendering pipelines and collaborate to create 3D assets.

“Each of our 3D artists has a different software workflow that they are used to – so it can be difficult to get a consistent result and still be flexible in terms of the tools that each artist uses,” said Jonathan Clark , CTO of Hexa. “Omniverse is an ideal candidate with great potential for Hexa in this regard. The platform will allow our artists to use the rendering software they are familiar with while allowing our team to visualize the final product in one place.”

learn more about it NVIDIA Omniverse and next-gen content creationregister for free NVIDIA Terms and Conditionsa global conference for the Age of AI and the Metaverse taking place online March 20-23.

Images and video courtesy of Hexa

Source link

Related Articles

Leave a Reply

Your email address will not be published. Required fields are marked *