Volumetric capture is the process for capturing human performances in real-time to bring them into virtual worlds. This is achieved by using multiple accurately synchronized cameras to capture the environment from many angles or vantage points. The captured data is then processed to form a 3D avatar of the subject that can be viewed from all angles.
As a premier supplier of volumetric capture cameras, accessories, and software Emergent’s products have been used by numerous tier 1 companies who have created capture studios and laboratories with as many as 500 cameras in a system.
EXCELLENT IMAGE QUALITY AT HIGH-SPEED
Our high-speed cameras utilize the latest sensor technology from Sony and Gpixel. These industrial grade sensors are designed for the most demanding VR, AR, and MR applications. Sensors from the Sony Pregius S family and the Gpixel GMAX and Gsprint series offer excellent image quality to capture every life-like detail, for example hair or skin texture, to render hyper realistic models. They also provide high-frame rates to reliably capture every movement of the model in real-time. All feature a global shutter CMOS to eliminate motion distortion.
With resolutions from 0.5 to 100MP+ and frame rates to 3462fps, we offer a wide range of cameras to suit your specific application requirements. Emergent can also provide turnkey performance camera systems tailored to individual application needs.
HIGH BANDWIDTH, HIGH PERFORMANCE
Because VR, AR, and MR applications are data intensive, cameras with high-bandwidth digital output interfaces are most commonly used. We are the award-winning pioneer for 10GigE, 25GigE, and 100GigE. Based on SFP+, SFP28, and QSFP28 interfaces, our cameras outperform competing interfaces like CoaXPress (CXP) with up to 12,500 Mbytes/s bandwidth available.
In addition, we take responsibility for the complete solution by developing technologies to keep the server-side CPU utilization and latency to incredibly low levels. This results in less than 1% CPU utilization and approximately only 100μs added latency.
ACCURATE MULTI-CAMERA SYNCHRONIZATION
Volumetric capture relies on multiple cameras to capture image data of a model. Emergent provides several options to tightly control the timing synchronization between cameras including Precision Time Protocol (PTP), and hardware triggering, delivering reliable microsecond accuracies with minimal effort.
Fig 1:The IEEE 1588 Precision Time Protocol (PTP) enables the precise synchronization of cameras in multi-camera systems.
SIMPLIFIED EASY LOWER COST INTEGRATION / SET-UP
Our cameras are GigE Vision and Genicam compliant. This means they can be used with standard hardware, cables, and any GigE Vision compatible software for simplified and low-cost integration. Emergent’s own software solution (eCapture Pro and eSDK) supports Windows 10 and Ubuntu, RedHat, and CentOS Linux. Because our cameras are based on SFP+, SFP28, and QSFP28, we offer several options to support cable lengths of up to 10kms.
One system which was recently shipped involved 3 servers and 48 Emergent Bolt HB-25000-SB-C 25GigE cameras running at 25MP and 60fps. See the video in-line for the initial software experience. This system records and saves 200Gbps (25GB) per server for a total of 600Gbps (75GB) across all 3 servers with ZERO data loss thanks to Emergent’s ZERO copy technology. All cameras are also synchronized to sub 1 microsecond through IEEE 1588 PTP v2. Our eCapture Pro software and eSDK are the glue to hold it all together. Simply connect the cables, power up, and run the software. In just a few clicks you will see your data come to life. Another such system has a single server with 36 Emergent HR-12000-SB-C 10GigE cameras running at 12MP and 60fps. We even provide stands and lighting where desired. Another video shows a system developed to perform high performance H.265 compression for storage of images to small SSDs as well as supporting RTMP for sending images to RTMP clients on the internet.
The above technology and services is also applicable to lower end systems such as NVIDIA Orin, Xavier, Clara AGX, etc. The above is also application to OEMs, system integrators, end-users, etc.
See the video in-line below for the initial software experience.
One can use our eSDK with all the same performance benefits. For inference applications, you can train your models with PyTorch or TensorFlow and then with a few clicks test them with an Emergent camera in eCapture Pro using our unique plug-in feature.