Explaining VR cameras

A VR camera is a set of two or more synchronized camera units capturing a 180º to 360º visual range that would realistically match human perception once viewed in a VR headset. That’s where “immersion” comes from - viewers experience things as if being there. VR is cool as it adds another level of intimacy between what happens on the screen and the viewer. POV videos offer an actual person’s perspective, with things seen at a human scale. Along with binaural audio it allows immersion into action based on the nature of the experience. A different mindset is required when producing VR than with conventional flat videos. 

Get the camera right and things will start to work effectively. VR post-production and rendering are covered by separate posts.

This post will be updated as new VR cameras come out

180º or 360º 

The stereoscopic 180º video is preferred for the current generation of VR. These cameras replicate human sight as each unit corresponds to the left and right eyes. Spaced apart at around 65 mm to mimic the distance between human pupils (Interpupillary distance or IPD), they are capturing the same scene from slightly different angles. The differences in captured images or the so-called parallax effect allows the human brain to recreate spatial 3D environments (stereopsis) like we experience in real life. The parallax effect accounts for depth perception and thus the 3D effect. Once in VR the person only sees what was shot on the camera, therefore making the camera work as someone’s eyes.

Fish-eye lenses are used for capturing a wide field of view ranging from 165º to a 220º field of view (FOV). Each camera records videos separately. Later these two videos are stitched together into a single file that is projected onto a virtual sphere inside the DeoVR video player. 

The more the viewer can “look around” within the video the better immersion is experienced. However, 360º videos are still rare as it’s much easier to produce 180 and 360º cameras fall short compared to 180º cameras. This is a very different type of technology involving four to 16 or even more cameras. When moving between the cameras, objects get often distorted. 360° cameras require an additional layer of complexity. Unless there’s a major shift in 360º technology, having so much video from behind the user is not highly valued yet doubles the already big filesize. It also requires well recognized visual or audio cues to change the point of view. 

180º stereo is an awesome hack to get a great VR with little effort. We recommend watching what’s being filmed in VR while recording to make sure the scene is properly set.

Camera specifications

There are some good professional, semi-professional and amateur cameras capable of VR 180° production. The standard specs are: 6K 180º 60FPS stereo video with 8K 60FPS on the horizon. 

Most cameras have 2900x2900px sensors per unit capturing from 165º to a 220º FOV. Once you put the two sets of footage side by side into a single output file it results in 5800x2900px. DeoVR fish-eye stitching technique utilizes full FOV and sensor resolution. Otherwise it is cut to 5400x2700px to fit an equirectangular 180º projection.

Often there’s a tradeoff between resolution and framerate, like 2880p@30FPS or 1440p@60FPS. If you have to choose, shoot at the highest resolution at a cost of lower framerate for better image quality. Frame rates can be later advanced using interpolation techniques

Keep in mind that some producers practice upscaling the footage up to 8K for marketing reasons, which is not beneficial at all from a quality perspective and is absolutely not recommended. 

There are some true 8K 60FPS cameras coming to the market that will give producers a significant advantage.

Professionals will be interested to know that there’s a trade-off - either sensor resolution is used for higher FOV at lower pixel density, or lower resolution at higher pixel density thus better image quality also resulting in sharper stereo.

Advanced producers should also consider the color depth of a camera. There are 8 bit cameras recording 16.7 million colors and more advanced 10-bit cameras with 1.07 billion colors (64 times more). We are still waiting for HDR headsets to arrive so we can fully enjoy the advantages of 10 bit color. In the meantime, 10 bit color gives immediate gains in color grading. 

Recorded video should be edited and stitched. Some manufacturers offer proprietary stitching and editing software, but professionals might like MistikaVR or the more advanced Mistika Boutique editor. We will be releasing a Mistika tutorial and presets soon.

Recording in h.265 could be seen as an advantage as many decoders are not capable of processing h.264 videos over 4K, h265 also offers better quality at a lower bitrate. 

It’s highly beneficial to stream a camera in VR or at least check the picture in VR before hitting the record button. 


Choosing the camera

 

Resolution

Lens

Framerate

Codec

Color

Live streaming

Stitching

Price

VuzeXR

5800x2900

210º

30

h.264

8 bit

+

Proprietary, free

$400

Zcam K1 Pro

5800x2900

165º, 200º

220º

30

h.264

8 bit

+

Proprietary, free

$4,000

Zcam K2 Pro

5800x2900

200º

60

h.265

10 bit

Extra setup required

Professional, required

$6,000

 

VuzeXR has proven to be a great amateur VR camera. It’s a good start for filming 6K VR videos. Proprietary stitching software makes it easy to get the right video with little investment. It also allows you to live stream in VR headset what is happening in front of the camera to assure the right scene.

Cons: Small sensor, requires a lot of light, otherwise might not look good. Due to the handle, it’s hard to create POV scenes with it. 

Vuze XR rig setup is coming soon...

ZCam K1 Pro is a semi-professional camera. 5800x2900px sensor @ 30FPS, recording in 8 bit, h.264 with 220º VRCA, MKX200 and MKX220 (recording in 165º) lenses. Equirectangular projection results in 5400x2700px 180º video.

One man team can make a great deal of it. Make sure to record in 2880p@30FPS. Then recorded fisheye videos from cameras need to be interpolated to 60FPS. The difference between two consecutive frames is algorithmically calculated then one frame in between is generated. Interpolated videos look a lot smoother than original. 

It comes with proprietary stitching software Wonderstitch making it an easy process.

Cons: Records in h.264, poor performance in darkness, 30fps.

Learn how to shoot with the Zcam K1 Pro


The ultimate professional camera available on the market is ZCam K2 Pro.

It’s a cinematic camera consisting of two ZCam E2 units. Although the K2 Pro sensor resolution is the same as the K1 Pro, the K2 Pro image quality is better. 10 bit gives more opportunities for color grading. While there are no VR headsets with 10 bit screens on the market to fully benefit from it, it’s still preferable to 8 bit recording. It also shows good performance while filming in the darkness. Also, the K2 records in 60fps and does not require interpolation.

All recent SLR Originals videos are shot with the K2 Pro. 

Recommended stitching software: Mistika Boutique.

Cons: 2.5kg heavy, additional setup required for live streaming. Pro only.

Zcam K2 Pro rig setup is coming soon…

 
Read also:
DeoVR now supports spatial audio. This allows the user to experience sound in VR as you would in real life when moving and rotating your head. This means you will experience much deeper immersion, wit...
Start shooting professional VR videos with the ZCam K2 Pro camera. Follow @deovr for more  The ZCam K2 Pro is a professional VR camera with 5800x2700px sensor @ 60FPS, 10-bit color reco...
A VR camera is a set of two or more synchronized camera units capturing a 180º to 360º visual range that would realistically match human perception once viewed in a VR headset. That’s ...