Volumetric capture: the journey to easy to use volumetric video setup?

Posted by EFEVE on March 26th, 2020

Many of us are used to seeing large volumetric capture studios made of hundreds of cameras. I am talking about Microsoft or Intel. Also, you have seen studios which use around 40 cameras such as 8i. Note, the last company tried to find a market for such a studio but now are starting to pitch smaller portable studio. This is what we've been doing at EF EVE since the start. The portability issue is huge and at the same time it raises the cost of hardware to such heights that close to 0 can afford it. However, no one can deny that the quality of volumetric video produced is great. But then what if the people just can't have it. 

We have been working with 3D sensors for volumetric capture since 2014 and believe me it is not an easy task. All of the sensors including Kinect V2, Intel Realsense and the newest Azure Kinect produce a lot of point cloud “noise”. It comes with factory calibration which can not give precise RGB and Depth matching. The sensors require loads of power and not to mention that volumetric video nature is to give enormous amounts of data. 

We as a team at EF EVE ™ everyday struggle with it. However, I am truly happy about our achievements to date. I would like to share with you some of them.

First of all we have tested, developed, failed and finally succeeded to have proprietary point cloud cleaning algorithms which greatly reduce the noise.

Second of all, we have developed a system which allows us to support 2 sensors on one PC. This greatly reduces the cost of volumetric capture set up.

Thirdly, we managed to compress the data so much that our proprietary depth codec allows us to stream live volumetric video into the web and VR, AR.

Some of the newest addition to this list is automatic sensor calibration. This has been an extremely difficult process and probably the one which constantly proves that there can't be enough precision. The saying that the more you eat the hungrier you get very well applies here. 

However, it is well worth it. We currently allow people to have automatic calibration of up to 10 sensors on 5 computers. In other words get 4 sensors and 2 computers and you have the most affordable volumetric capture to date. 

We know that not everyone is looking for precision and therefore released another of our products EF EVE ™ Creator. I compare it to Adobe Premiere when it just started but for volumetric video. The software allows you to edit VV, manipulate quality and has in-built VFX as well as OBJ or PLY sequence export. Also, we offer Platform with VR functionality and embeddable 3D web container for VV.

I am sure that I have forgotten to mention so many other things. However, when you put this all together it's quite a lot for such a small team. If any of you have questions or want to learn more about our journey please visit here.

Link To Directory
Top Searches - Trending Searches - New Articles - Top Articles - Trending Articles - Featured Articles - Top Members

Copyright 2020 Uberant.com
739,067 total articles and counting.