Volumetric Capture is a new technology which captures a three-dimensional space, such as a location or performance utilising a mixture of multiple cameras and depth sensors with custom geometric and pixel mapping algorithms. At SpeedAR, with our volumetric video production and Augmented Reality film production services, we employ a number of different solutions depending on the complexity of each project at hand.

A scene or action captured as a volumetric video can be turned into a 3D model mesh. Once in volumetric mesh form, a user can view it’s virtual, holographic style form from any angle and position. Volumetric videos are platform agnostic, but work best with the latest AR headsets such as Magic Leap and Hololens.

Volumetric Video Capture

Complex Volumetric Video Capture

The actor is placed in the middle of a large range of capture cameras and IR depth sensors. Often this is done within a green-screen surround, although this is not necessary. The studio lighting is designed to provide equal light balance to the capture model which is important for colour consistency. The resulting 3D mesh models have the lowest level of depth noise, however take a very long time to produce and process. This form of capture is limited to short takes.

Simple Volumetric Video Capture

Our simpler method of producing volumetric video animations is to use a lower number of sensor units with depth sensors and cameras feeding data into a lower post pipeline handling lower density models. By reducing the amount of capture information, the post pipeline is streamlined and the data size more manageable. There is more flexibility with this style of Volumetric Video Production, however models can be noisy or feature subtle warp from frame to frame.

Basic Volumetric Video Capture

The basic method we use for Volumetric film capture is to use one or two depth sensor units with custom software. This provides a lower quality model, however the turnaround time and flexibility for longer takes is vastly improved. This form of capture is suited to giving actors and directors freedom for multiple takes and run-throughs in a short space of time. The amount of data recored also remains relatively low and can be processed more quickly into the developer workspace.

Photogrammetry Model Capture

Our method of photogrammetry turns static objects into 3D mesh models for use in augmented reality headsets and on mobile devices with our augmented reality app development. Using a series of images taken with single camera, multi-camera, or multi-camera, multi-depth sensor, we can create many common objects of varying sizes for use as 3D model assets within augmented reality. This can even be done with static models of humans.

The resulting models can be used as assets in any Augmented Reality scenario and platform, or even Virtual Reality scenes.


Photogrammetry Model Capture

There are generally two methods of photogrammetry we consider for each project at SpeedAR. The more complex setup uses a large range of cameras configured in a studio with balanced lighting. This results in the densest models with consistent lighting and colour textures. The models often need retopology to reduce the overall polygonal count whilst maintaining quality.

We also employ simpler methods for quicker turnarounds for more basic textured models using a single camera with multiple views of the object. This method is slow and less consistent, but quicker to setup.