Delay between input and real-time output — critical in motion capture, virtual production, VR scenes. Above 40ms noticeable, beyond 100ms unusable.
Anyone shooting motion capture or working on a virtual production set needs to understand latency – otherwise, it's a disaster. The delay between a physical movement and its real-time representation in the digital system determines whether an actor can perform fluidly or feels like they're moving through quicksand. Below 40 milliseconds, the eye barely notices anything. But from 100ms onwards, every gesture becomes torture – the actor sees their virtual avatar reacting on the monitor with a noticeable lag, the synchronization between body and feedback collapses.
In a motion capture studio, latency is everywhere. Cameras capture markers, tracking software calculates position and rotation, the host computer interpolates the skeleton, the real-time engine renders the geometry, and the video is sent to the monitor. Each station costs frames. Professional systems stay below 16ms – that's the standard at 60fps. But as soon as you add complex rigging, cloth simulation, or many virtual cameras, it can quickly become 40–80ms. And then the actor sits there, looks at the monitor, and can't maintain their timing.
Virtual Production exacerbates this further. The physically filmed actor must be synchronized in real-time with the live rendering engine of the LED wall. If the camera pans and the world renders a frame too late, you see it immediately – the parallax is wrong, the perspective breaks. This isn't subjective discomfort; it's technical failure. That's why VP systems run on specialized hardware stacks, optimal networking, and dedicated GPU architecture.
Practical advice: Measure latency, don't guess. Work with timing tests – a blinking LED in the motion capture room, a stopwatch, a frame counter on the monitor. 20ms is acceptable, 30ms is okay, 40ms is noticeable, 60ms is problematic. For VR applications (see: Immersive Production), the requirements are even stricter – the human vestibular system registers latency from 20ms as discomfort. That's why VR headset manufacturers are working on sub-20ms systems.
For the DoP and the VFX supervisor, this means concrete steps: calibrate systems before shooting, adjust buffer times and frame buffer configurations, and always plan for a buffer margin during long takes with complex simulations. With Virtual Production, you can also compensate technically – with predictive rendering, meaning speculatively rendering the next frames in advance. But that's a band-aid. Ultimately, it only works if the entire pipeline is designed for minimal latency.