VJing is the part of the club night nobody in the crowd names but everybody feels: the visuals behind the booth living or dying on whether the software can keep up in real time. Resolume just closed a real gap for that job. Version 7.28 of Avenue and Arena, out September 22, adds computer vision as a native building block, and critically, it never leaves the machine.

What actually changed under the hood?

The update introduces two new low-level nodes in Wire, Resolume's node-based patching environment: Human Segmentation and Depth Estimation. Human Segmentation isolates a person from their background frame by frame; Depth Estimation builds a depth map out of an ordinary 2D camera feed, no depth sensor required. Built on top of those two nodes, Arena and Avenue ship six new ready-made effects: Background Removal, Depth Blur, Depth Field, Depth Displace, Depth Pixelation and Depth Shift, plus a new Depth blend mode for compositing layers by apparent distance rather than plain alpha. Because every effect is itself a Wire patch, a VJ can open it, rewire it, or build an entirely custom effect from the same two nodes, per CDM's rundown of the release.

Depth Estimation carries temporal filtering and smoothing specifically to cut down flicker between frames, the failure mode that has made real-time depth mapping unreliable for live use until now. The release also adds Group FFT, letting audio-reactive effects pull frequency data across multiple layers at once, and a right-click-to-default reset inside the Wire node editor.

Why does running locally matter more than the effects themselves?

Any of these effects could, in theory, be farmed out to a cloud AI service, plenty of video tools already do exactly that. For a VJ mixing live in a club or on a festival stage, that is a non-starter: a network hiccup mid-set means a frozen frame or a dropped feed in front of thousands of people, and sending a camera feed off-site raises its own privacy questions when that feed includes a crowd. Resolume's approach keeps segmentation and depth estimation running entirely on the VJ's own hardware, with no round-trip and no data leaving the rig. That is the actual news here: not that background removal exists, but that it exists without the latency and dependency tax that comes with cloud AI.

Running the models locally means the visuals stay live even if the venue's internet does not.

Who does this actually change things for?

Any VJ or visual artist running Arena or Avenue for club, festival or installation work now has real-time human segmentation and depth effects without buying a depth camera or writing custom computer-vision code. It also narrows the gap between Resolume and the wave of AI-powered visual tools that have leaned on cloud inference, while keeping the reliability guarantees that live production actually needs.

Why it matters

Visuals are increasingly part of how house and techno events sell themselves, from festival main stages to warehouse parties with projection mapping, and the software running that show has quietly become as consequential as the DJ software next to it. On-device computer vision removes one of the last real excuses for visuals crashing mid-set.

What we think

This is the kind of update that never trends but genuinely changes a workflow: a VJ can now pull a performer out of a background or fake volumetric depth from a flat webcam, live, with zero internet dependency. That is a bigger deal for the club floor than another synth plugin, because when visuals glitch, the whole room feels it.