ViewAR: AR visualizations and shoes try-on for the Metaverse

ViewAR is our new product for simple visualization of any 3D asset in Augmented Reality secured by the ICP with an additional feature that allows for displaying 3D models of shoes in a virtual try-on.

Comment 2

Inspiration

Digitization of assets and 3D scanning (creation of digital twins of physical items) is getting easier and simpler. But the Metaverse ecosystem is still missing a solution for a few problems, which ViewAR will solve by:

  • simple visualization of 3D objects in WebAR,
  • virtual try-on for fashion assets (like shoes, caps, glasses and ultimately clothes),
  • the ownership of digital assets,
  • missing usability and functionality of fashion NFTs.

We specialize in development for the fashion industry which is struggling with the problem of high return rates due to a mismatch in size or look. Solving this problem was initially the main goal of WEARFITS. We are now looking at other industries more broadly as well, and we are also taking a step forward from Web2 to Web3.

We believe that Metaverse and AR will be the next big thing, so we focus on solutions for XR/AR utilising deep tech like machine learning and blockchain.

What it does

ViewAR allows for AR visualization of any 3D asset with digital rights and ownership secured. There is also an additional feature for the fashion industry that allows for displaying 3D models of shoes in a virtual try-on. We help in getting to the Metaverse :)

How it works:

  • sign in using Internet Identity
  • drag and drop any 3D file to ViewAR
  • the file is converted to WebAR for both platforms: iOS and Android
  • optionally make simple 3D edits (e.g. texture, size etc.) before publishing to AR
  • get URL to AR preview and QR-code
  • all files are stored on the Internet Computer blockchain with the publisher's digital rights and ownership of 3D asset secured

For 3D objects representing footwear, ViewAR creates a virtual try-on link, allowing to try on shoes in front of the real mirror and on your own feet, just with the use of a smartphone or laptop camera (without any additional app).

The solution has huge social and ecommerce potential and solves a real world problem with contactless purchases (as well online as offline - you can also try-on shoes in physical stores virtually). The MVP is limited (e.g. storage) but fully usable. It utilizes core IC functionalities which protect publishers. We are going to commercialize it and add more functionalities (e.g. simple creation of NFTs, automatic creation of 3D models from photographs, the most accurate shoe-foot size fitting).

How we've built it

We have made an MVP implementing the Internet Identity and persistent file storage on the canister. Most of the computing job (like file conversion, machine learning) is done on the client's side to optimize speed and the usage of cycles. We have a module responsible for conversion and optimization of 3D files and publishing it via URL or QR-code. For the shoe try-on, we have trained our own ML model with over 15,000 hand-annotated foot images, which allows us to position and visualize the 3D model of the shoe in real time.

Challenges we've ran into

Training the ML model with more than 15K images. Optimization for real time use through a web browser. Poor and limited persistent storage functionality of the ICP.

Accomplishments that we're proud of

The results of implementation on the ICP are quite satisfying. Our ML model is very accurate. And the solution works quite well :)

What we've learned

We have learned a lot about the Internet Computer and ICP. We are also using machine learning much more intensively. The project proved to us that using Internet Computer for Web3 makes sense and it has a chance to be the future of it.

What's next

For the production release of ViewAR we should:

  • implement ICP payments
  • solve problems with storage limits
  • develop some admin page for publishers

ViewAR is useless without 3D models. Therefore, we have also developed a technology that automatically creates 3D models of shoes from photos. It will allow us to sell it on a large scale. We'd like to integrate it with IC as well to offer a complex solution for Web3.

In the next few days we'll launch a shoes try-on for a large customer.

Currently we are working on a footwear size fitting solution using our ML model to calculate accurate measurements of feet. We plan to release it in the following months.

This year we'd like to raise a seed round for sales and marketing activities as well as for further development of our virtual try-on and size fitting solutions for the fashion industry (both - apparel and footwear) and ViewAR.

We'd like to develop our new solutions parallely for both Web2 and Web3.

+ 1 more
Share this project:

Updates