Library · Devices and formats
Spatial video and MV-HEVC
Spatial video is Apple's commercial name for stereoscopic video. It is recorded with iPhone and Apple Vision Pro; both eyes are stored in a single file in MV-HEVC format. Device information on this page is as of November 2026.
Recording
| Device | Per eye | Frame rate | Interaxial |
|---|---|---|---|
| iPhone | 1920 × 1080 | 30 fps, SDR | 19.2 mm |
| Apple Vision Pro | 2200 × 2200 | 30 fps, SDR | 63.76 mm |
- iPhone models: iPhone 15 Pro and 15 Pro Max, the iPhone 16 family (except 16e), iPhone 17, 17 Pro and 17 Pro Max. iPhone 15, 15 Plus, 16e and iPhone Air do not record spatial video.
- The two eyes on iPhone: the main image comes from the wide (1×) camera; the second eye is cropped from the ultra-wide camera's image.
- Interaxial: 19.2 mm on iPhone, about a third of the interocular distance (hypostereo). The Vision Pro value was read from the metadata of its own files (Swanson 2024).
- Spatial photo: a multi-image HEIC file. It contains the left- and right-eye images, a group that defines them as a stereo pair, and spatial metadata.
MV-HEVC
MV-HEVC (Multiview HEVC) is the multiview extension of the H.265/HEVC standard (Annex G; the common multi-layer structure is Annex F). Both eyes sit in a single video track at the same resolution:
- Base layer: the left eye, coded as ordinary HEVC. A player that does not know stereo decodes only this layer and plays the video in 2D.
- Second layer: the right eye, coded by prediction from the left eye (inter-view prediction). Because the two eyes are very similar, sources report files about 25–35% smaller than two separately coded streams (NVIDIA 2025; 3GPP TR 26.966).
Metadata
A spatial video file also carries stereo information so that the player can show the image correctly:
| Field | Unit | Meaning |
|---|---|---|
| Baseline (interaxial) | micrometers | distance between the optical centers of the two cameras |
| Horizontal field of view | thousandths of a degree | horizontal angle covered by the image |
| Horizontal disparity adjustment | ten-thousandths of image width | horizontal shift the player applies to the two images |
| Projection | – | projection type of the image |
Disparity adjustment and HIT
Apple also calls this value the convergence adjustment. The player shifts the left and right images horizontally in opposite directions, each by half the value. It is a HIT applied at playback:
d' = d + H H = disparity adjustment
where: d is the parallax (fraction of image width), d' the parallax at playback
- A positive value pushes the scene back, a negative value brings it forward; the sign is the same as HIT on this site.
- Apple states that +2% typically provides a good default. Values read from files: iPhone 15 Pro +2%, Vision Pro +2.93% (Swanson 2024).
- Convergence is set at playback, not during shooting. In post-production, depth grading can be done by changing this value.
Tools
- Final Cut Pro 11 (November 2024): imports and edits spatial video, and exports it as MV-HEVC (8 or 10 bit) for Apple Vision Pro.
- Apple developer tools: AVFoundation and VideoToolbox encode MV-HEVC; Compressor can also export it.
- AWS Elemental MediaConvert: creates MV-HEVC from two inputs, one for the left and one for the right eye.
- ffmpeg decodes MV-HEVC since version 7.1. Two-view encoding is available in the x265 4.0 encoder; the feature is enabled separately when the encoder is built (ENABLE_MULTIVIEW).
- Meta Quest plays spatial video since software v62 (February 2024). Videos are uploaded with the Meta Quest phone app, converted for Quest and stored in the cloud.
When shooting
- The interaxial is fixed. The depth range is controlled only by the camera's distance to the scene and by how the scene is arranged.
- Display size: spatial video is mostly watched in a window on a headset or on a virtual screen. The target screen is the apparent width and distance of that window. See Viewing devices.