Photography has always been a medium for a lot of people to reflect on their past as well as to see the world from someone else’s viewpoint. In attempts to make the medium much more life-like, Kharpann has implemented a complex computer vision algorithm to create 3D videos from 2D images that show the beauty of Nepal.

Kharpann Enterprises Pvt. Ltd. is a privately-owned AI company in Nepal which provides outsourcing AI solutions to its clients in different domains such as Computer Vision, Time-series analysis, Natural Language processing, etc. The company states this project as an effort to make online media more interesting with the help of algorithmic processing.

Kharpann’s YouTube Video – Creating 3D videos from 2D images

Kharpann has recently published a video on YouTube presenting the results of their implementation of a Computer-Vision algorithm known as ‘3D Photography using Context-aware Layered Depth Inpainting’ (Meng-Li Shih, Shih-Yang SuJohannes Kopf, Jia-Bin Huang). The Computer-Vision algorithm was published in early 2020 with results comparable to Facebook’s 3D photo generation algorithm.

The video shows a compilation of a total of six different photographs with their 3D counterparts, and the results look oddly satisfying.

Have a look at the results here:

According to the paper linked in the video description, ‘3D Photography using Context-aware Layered Depth Inpainting’ shows results better than that of Facebook’s 3D photo in terms of the smoothing behavior of the algorithm. Moreover, the algorithm does not leave an empty area or stretch the image as produced by depth-warping.

Kharpann 3D video - 3D Photography using Context-aware Layered Depth Inpainting
Result comparison of various algorithm’s against (d) 3D Photography using Context-aware Layered Depth Inpainting

The algorithm uses Layered Depth Image with explicit pixel connectivity as underlying representation and presents a learning-based inpainting model that synthesizes new local color-and-depth content into the occluded region in a spatial context-aware manner.

Source: 3D Photography using Context-aware Layered Depth Inpainting

Create your own 3D videos with Kharpann

Kharpann is currently looking to collaborate with online media and photography enthusiasts of Nepal and is accepting photograph entries to be featured on their next 3D conversion video. If you are a photographer and are interested in adding life to your photos in three dimensions, send it over at [email protected].

Kharpann's YouTube video snippet
Kharpann’s YouTube video snippet

In Conclusion

Looks like 3D photos and videos will be seen more frequently now in social media as well as on online news portals. This is a wonderful infusion of photography with algorithmic processing and we are excited to see what new changes this technology will bring in media all around the world.

What do you think about the video? Feel free to let us know in the comments. If you liked this article, you might want to have a look at Kharpann’s findings on the ‘Foodmandu Case Study: A Complete Analytical Report of 2019‘.

LEAVE A REPLY

Please enter your comment!
Please enter your name here