3D Display Videos: Technical Explainer

3D display videos are video assets that encode separate left-eye and right-eye views so that an autostereoscopic (glasses-free) display can deliver a stereo image to a viewer without head-mounted optics. This explainer frames the concept, walks the architecture, and sets expectations for review workflows built around 3D video content.

Concept Frame: What 3D Display Videos Mean

For professional 3DV audiences, the term 3D display videos is used narrowly. It refers to video sources that carry or can be mapped into a stereo view pair, rendered against a display whose optical layer separates those views per eye. The category sits inside the broader 3D Spatial Display product line and is a workflow question as much as a content question.

It is important to separate this from related but distinct terms:

• Glasses-free 3D videos – any video intended for autostereoscopic playback, regardless of how the source was captured.
• Naked-eye 3D videos – the popular consumer phrasing for the same target, often used in out-of-home signage contexts.
• Autostereoscopic video – the engineering term, emphasizing the optical layer that makes the view separation possible.
• Stereoscopic 3D video – the broader category, which includes glasses-required formats (active shutter, passive polarized) that 3DV displays do not use.

3DV displays are architected for the narrow, glasses-free interpretation. A 3D video that requires active-shutter glasses or polarized eyewear is not, by itself, a 3D display video in the workflow sense — it would need to be re-rendered or re-encoded into a stereoscopic input acceptable to the display pipeline.

Source Formats: SBS, Interleaved, and Stereo-Ready Video

Source format is the first gating decision. The most common input formats for glasses-free 3D playback are:

• Side-by-side (SBS) – left and right views occupy the left and right halves of a single frame. Widely used because it is simple to author, store, and stream. Tools like the 3DV Spatial Player / SBS player validate this input form against the spatial display pipeline.
• Top-bottom (over-under) – left and right views stacked vertically. Used when horizontal resolution is more constrained than vertical, or by camera pipelines that natively output this layout.
• Frame-interleaved / column-interleaved – pixel rows or columns are swapped between views. Acceptable to some optical layers but more sensitive to scaling and compression artifacts.
• Multi-view / MV-HEVC – multiple views encoded in a single stream. Useful for content that needs to drive several viewing zones, but requires an authoring and decode pipeline that not all professional workflows yet support.
• Application-rendered stereo – the view pair is generated live by a 3D engine (Unity, Unreal, WebGL, custom OpenGL/Vulkan) using a stereo camera rig. This is the dominant path for CAD review, medical visualization, and industrial inspection content.

A practical rule: if the source cannot be expressed as a left-eye / right-eye view pair at the frame buffer, the display pipeline cannot recover a usable stereo image. Plain 2D video is legitimate input — it just displays as a 2D image on the same panel, which is why 3DV Pro models ship with a 2D / 3D workflow switch.

Architecture Notes: How Autostereoscopic Displays Decode Video

Decoding a 3D video into a perceived stereo image is an architectural problem, not a single-step operation. The main building blocks are:

1. Input ingest and view separation. The pipeline identifies the stereo layout (SBS, top-bottom, interleaved, or application-rendered) and routes the left and right view buffers separately. See the Autostereoscopy primer for the optical-layer grounding.
2. Eye tracking. A structured-light sensor monitors the viewer’s position. The tracked eye coordinates feed the dynamic view mapping stage so that the left and right view frustums stay aligned to the actual eyes, not a fixed nominal position.
3. Dynamic stereo view mapping. The FPGA or display-side processing pipeline warps and crops the view pair in real time so the active optical element can deliver the correct sub-image to each eye. This step is what makes eye-tracked, glasses-free 3D viewing feel stable across natural head movement.
4. Optical layer. A lenticular or parallax barrier array refracts or masks the sub-pixel pattern so that each eye sees only its intended view. Different optical layers have different crosstalk budgets and viewing-zone counts, which in turn constrain the video source framing.
5. Refresh and sync. The panel drives the mapped sub-image at full panel refresh. For head-tracked 3DV displays, view mapping runs continuously, so the input frame rate and the view-mapping update rate should be matched within the pipeline’s supported range.

Two implications for video authors. First, the view pair is the contract; the pipeline does not recover what was not recorded or rendered. Second, the optical layer dictates the geometry of the final sub-image pattern, so tightly cropped frames, fine text, and small UI elements placed near the panel edges can end up partially in one view only, which is read as crosstalk or blur.

Workflow Implications: Preparing Source Content for Review

For a 3D video to perform on a glasses-free display, the source must be shaped for the architecture. The core workflow implications are:

• Author in a known stereo layout. SBS remains the safest default for pre-recorded content. Application-rendered content should expose a stereo camera rig and a per-eye view buffer.
• Match the viewing distance to the optical layer. Parallax barrier and lenticular designs are tuned for a nominal eye-to-panel distance. Designing content at the right depth scale for that distance avoids the “cardboard cutout at the screen” or “object floating far in front” failure modes.
• Plan framing for the active viewing zone. Eye-tracked displays follow one viewer in the active zone. Content meant for shared viewing needs either a multi-view source or a controlled viewing arrangement.
• Reserve headroom for the view-mapping warp. Tracked eye movement causes the pipeline to crop and re-project the view pair. Putting critical UI outside the safe frame area invites clipping at the panel edges.
• Use the 2D / 3D switch for hybrid review sessions. Pro models accept the same panel for high-quality 2D work and 3D review. Switching mid-session lets reviewers compare a flat reference image against the stereo rendering of the same source.

For teams that already run a 3DV Spatial Display, the path from a 2D video asset to a 3D display video is usually one of: re-export as SBS from a stereo-capable source, capture a stereo render from a 3D engine, or accept that the asset will play as 2D on the same panel.

Limits and Failure Modes

3D display videos are subject to limits that are easy to underestimate:

• Single-viewer assumption. Eye-tracked autostereoscopic displays are tuned for one viewer in the active zone. Multi-viewer content requires either multi-view source content or a workflow that rotates reviewers through the active zone.
• Crosstalk at frame edges. High-contrast elements close to the panel edges can leak across views, producing ghosting. This is an optical-layer property, not a content property.
• Depth budget. Comfortable depth on a given panel size is bounded. Pushing content far in front of or behind the panel quickly becomes uncomfortable for the viewer.
• Format mismatch. A 2D video routed into the 3D path will not synthesize depth on its own. The pipeline expects a stereo input; absent that, the result is a 2D image on the 3D panel.
• Compression sensitivity. SBS and other stereo layouts double the effective data per frame compared to a 2D stream. Aggressive compression can introduce inter-view artifacts that look like depth noise.

These are not reasons to avoid 3D video; they are constraints to design into the workflow.

Implementation Notes for Review Teams

For evaluation and deployment planning, the following notes are useful for technical buyers and integration teams:

• Start with content, not panel. Confirm whether the planned content exists as a stereo view pair or can be rendered as one. The pipeline extends that content; it does not invent depth.
• Use the Spatial Display Simulator and Player for early validation. The Spatial Display Simulator and the 3DV Spatial Player / SBS player are the fastest path to check whether an SBS asset or a rendered view pair is even well-formed before commissioning a panel.
• Plan around the active viewing zone. If the review is one-on-one (a specialist at the panel), the eye-tracked workflow is direct. If it is a small group, structure the session so the active viewer is the one making the call.
• Keep the 2D path for legacy content. 2D video and image review remain legitimate uses of the same panel. A 3D video pipeline that fails to fall back to 2D is harder to operate than one that does.
• Loop in the application layer. For content rendered from CAD, DICOM, or 3D engines, the stereo camera rig is the contract with the display. Confirm that the chosen viewer or SDK exposes a per-eye view buffer before commit.

3D display videos are best understood as a contract between content and architecture. The display pipeline is engineered to deliver a stereo image to a tracked viewer without glasses, but it can only deliver what the source provides. Teams that align their acquisition, encoding, and review workflow to that contract get the most out of the spatial display; teams that treat 3D video as a drop-in replacement for 2D video usually find that the bottleneck is upstream of the panel.