Doly Generates 3D Product Videos on iPhone

French startup AniML’s Doly generates 3D product videos directly on your mobile device, making it easier for online vendors to advertise their products.

AniML, the French startup that developed the new 3D capture application, Doly, aspires to create the PhotoRoom of product recordings. Doly assists individuals engaged in the sale of footwear through an online marketplace or the creation of Instagram advertisements for their direct-to-consumer products in converting 3D models created on their mobile devices into product videos with an expert appearance.

Although video production is notoriously tricky, 3D model generation is even more challenging. As a result, the AniML team has prioritized experience simplification. They intend to integrate 3D capture into an iPhone application to popularize the technology.

Doly’s 3D capture functionality is as follows: To capture the product in three dimensions, the user physically maneuvers around it while pointing their phone camera at it. The application captures and uploads static images to the cloud in the background. AniML has constructed a reconstruction pipeline employing Gaussian splatting to generate a realistic three-dimensional model from these images.

Traditionally, 3D models are constructed using a collection of coordinates in three-dimensional space, a 2D texture superimposed on these surfaces, and lighting effects. Gaussian splatting is a novel rendering pipeline in which a pre-trained AI model estimates a 3D point cloud from 2D images.

“A technological discovery served as our impetus: artificial intelligence had recently entered the realm of a three-dimensional environment. “Therefore, Facebook employees, and Google employees in particular, conducted research on NeRF and published a relatively significant research paper on the subject,” Rémi Rousseau, co-founder and CEO of AniML, explained to TechCrunch. “It is a paradigm shift to attempt 3D reconstruction through machine learning.”

“You are operating in neural-based 3D as opposed to polygon-based 3D,” he further stated.

Gaussian splatting differs slightly from NeRF, but according to Rousseau, it is a descendant of 3D modeling technology.

This concludes the technical aspect. AniML subsequently concentrated on identifying a use case that could immediately attract users. E-commerce enterprises were the logical selection for a 3D model creation application.

What additional features does the application provide? After capturing a 3D model, Doly users can select a 3D scene for their object to be integrated into by navigating a template library. In terms of marketing staging, this could be as straightforward as a 3D rotation featuring an empty backdrop, or it could be more theatrical, as evidenced by the camera gradually approaching the subject with varying angles.

Customers can purchase and obtain the video from the app for future use if they are satisfied with the outcome.

Rousseau previously established two virtual reality (VR) enterprises, one of which was Mimesys, a startup subsequently acquired by Magic Leap in 2019. Pierre Pontevia, one of the company’s co-founders, has an intriguing history, having sold two businesses to the 3D tools behemoth Autodesk and the 3D content development platform Unity, respectively.

AniML has raised $2 million to date, with Adjacent serving as the seed round leader. Additionally, the venture took part in AI Grant, an accelerator program founded by Nat Friedman and Daniel Gross. Several angel investors and Kima Ventures are also investing, including Hugging Face’s Julien Chaumond, Stupeflix’s Nicolas Steegman and Francois Lagunas, Sketchfab’s Alban Denoyel, Bertrand Schmitt, Thibaud Elziere, and Vincent Nallatamby. Bpifrance reportedly contributed a portion of this round through a grant as well.

Whether or not e-commerce professionals, second-hand resellers, and significant brands adopt 3D-rendered videos for upcoming campaigns and online listings will be intriguing to observe. However, artificial intelligence may eliminate the need for a professional video recording studio to produce captivating product visuals.