A 2025 preprint measured LED-wall latency against a named threshold

The stage

Render, tracking and display latency is a research question, not a single event, so this entry is anchored to one opened paper: The Stage Comes to You: A Real-Time Tele-Immersive System with 3D Point Clouds and Vibrotactile Feedback, by Takahiro Matsumoto and six co-authors, posted to arXiv, 8 October 2025. The paper documents a system capturing performers as 3D point clouds via synchronized LiDAR-and-camera units, transmitting the data, and rendering it on a large 3D LED wall at a remote venue with a haptic floor. Its headline result is an end-to-end latency, capture through render, averaging 81.3 milliseconds at Expo 2025, linking sites 20 kilometres apart for a live dance performance.

What the documents establish

The paper's methodology states its figure precisely: capture units, each with three LiDAR sensors and a global-shutter camera synchronized by precision time protocol, feed a GPU host fusing point clouds before transmission, and the authors measured 81.3 milliseconds end-to-end, compared against the 150 ms one-way threshold of ITU-T G.114, a telecommunications standard for acceptable delay. That is the paper's own method, not this site's estimate. This is a tele-immersive performance, not a conventional in-camera-visual-effects stage; its LED wall displays a remote performer's point-cloud stream, not a virtual environment behind actors. A second paper, Azzarelli, Anantrasirichai and Bull's 2024 review, Reviewing Intelligent Cinematography, states virtual production has the greatest potential to impact other production, driven by growing interest in LED volumes for in-camera visual effects, confirming ICVFX latency sits inside an active research area without itself supplying a figure.

The department view

Render and tracking latency is the brain bar's problem first: keeping the engine, tracking system and LED processing chain synchronized so a camera move appears on the wall without perceptible lag. The paper's approach, measuring latency against a named, external threshold rather than an internal target, is the discipline a volume's team needs even though the system differs: a sourced number checked against a standard, not a round figure asserted without method. The trade-off it documents, sub-100-millisecond latency via synchronized capture, shows the budget a real-time pipeline needs.

What to check before you build

This is an editorial application of the paper's method, not a claim about film production. A production should ask vendors for a latency figure measured against a named methodology, not an unverified marketing figure; confirm what threshold a figure is compared against, since ITU-T G.114's 150ms figure was designed for voice, not camera-to-wall perception; and treat any number as illustrative, not a benchmark.

• Has our tracking and render vendor supplied an end-to-end latency figure, and against what named methodology was it measured?
• What perceptibility threshold is a quoted latency number actually being compared against, and does that threshold apply to our use case?
• Is a cited latency figure from a film ICVFX study, or, as here, an adjacent LED-wall application?

Render latency for LED-wall systems remains an active research question, with one opened paper stating a measured figure and the standard checked against it, a stronger basis than an unsourced rule of thumb, even from a tele-immersive stage rather than a film set.