One camera can't see your room.
Two or three, directed properly, can. All 7 emirates.
Presenter camera. Audience camera. Content camera. And the part every brochure skips — the director that decides which one the meeting sees, second by second. Vector Digital Systems designs multi-camera rooms for Teams, Zoom and BYOD, installs them, and proves the switching on a live call before handover. Since 2009, from Dubai stock.
- PTZ & tracking cameras, counted per zone
- A director — processor, AI core or platform AI
- Voice or vision switching, tuned on site
- Dedicated camera cabling, planned before ceilings close
- Teams Rooms Pro configuration done right
- Live-call handover with the switching proven
Half your room is invisible on video calls — here's why
You have seen the recording. The trainer moves to the whiteboard and vanishes off the edge of the frame. The remote attendees watch an empty lectern for four minutes. Or the boardroom version: the camera sits under the display, twelve people around the table, and the six with their backs to it spend the whole meeting as haircuts.
The instinct is to buy a wider lens. The instinct is wrong. Wider means smaller faces, and the people who were already hard to see become impossible. A room with more than one zone that matters — a presenter and an audience, a table and a whiteboard, a stage and a floor — needs more than one camera. That much, most suppliers will tell you.
What almost none of them explain is the harder half: a video call only uses one camera at a time. Buying a second camera without deciding who directs between them buys you a dropdown menu, not a better meeting. This page is about the directing — the three ways to do it, what each honestly costs, and the 2026 platform facts we verified against Microsoft's own documentation this month, because this capability has been changing quarterly and stale advice here is expensive.
A video call shows one camera at a time — so who switches them?
Plug four cameras into a meeting PC and Teams or Zoom will list four devices. It will use one. The other three wait in a dropdown nobody opens mid-meeting. That is the entire multi-camera problem in one sentence: the cameras are easy; the decision of which one the meeting sees, second by second, is the product you are actually buying.
Something has to make that decision. It can be a human with a control panel — which works until the human is the trainer, mid-sentence. It can be hardware: a processor that listens, watches and cuts between cameras automatically. Or it can be the platform itself, with Microsoft's AI composing what remote people see. Those are the three architectures below, and choosing between them is the real decision on this page. Everything else — camera count, placement, cabling — follows from it.
Three ways to make cameras switch automatically — and which fits your room
Method 1 — The ecosystem processor (AED 45,000–110,000 for typical two-to-three camera rooms). A native Teams Rooms or Zoom Rooms system whose AV processor is the director. Cameras connect to it over dedicated links; it selects the shot based on who is speaking, and the platform receives one clean, already-directed feed. Yealink's AVHub sitting under an MVC system is the reference design; Poly and Logitech build the same logic into their large-room ecosystems. This is the corporate answer: one-touch join, the switching invisible to users, the whole thing managed like any other Teams Room. If your organisation lives in Teams or Zoom and the room is yours, start here.
Method 2 — The AI production core (AED 38,000–95,000). A central matrix tracking processor — the class AVer's MT500 defines — takes every camera, watches and listens with on-board AI, and outputs a directed feed to whatever the room is running: Teams today, Zoom tomorrow, a visiting laptop on Thursday. FaceFocus-style vision switching picks the camera that shows the speaker's face most clearly, not just the loudest microphone. Dante and NDI carry everything over the network, and the same core records or streams the production. Training providers, lecture halls, government chambers and any venue serving outside organisations: this is your method, because it belongs to the room, not to a platform.
Method 3 — Platform-native intelligence (licence cost, not hardware cost). Microsoft Teams Rooms on Windows can now do real multi-camera work by itself — up to four camera feeds shown to remote participants, or AI-composed IntelliFrame views that cut the room into individual face tiles. It is genuinely impressive, genuinely improving, and genuinely bounded — the boundaries are specific enough that we gave them their own section below, verified against Microsoft's documentation as of this month. Small and medium rooms with the right licence get remarkable results from this method plus good cameras. Large rooms usually pair it with Method 1: hardware directs, the platform's intelligence rides on top.
One honest line before the details: many rooms that ask us for multi-camera need one better-placed camera and a tracking model. The survey answers that question before any quote does — and when the answer is one camera, we say so.
Two quotes for the same second camera, thousands apart — here’s what the cheap one skipped
You asked three companies to add a second camera. The quotes came back thousands of dirhams apart — for the same camera model. Before choosing the cheap one, check what it actually includes. A camera in a box and a camera working in your ceiling are two different products.
Four things turn a boxed camera into a working one. The cable. A dedicated run from the camera position to the processor — through the ceiling, terminated and tested. In a finished office, that run often costs more than the camera itself. The brain it plugs into. Every camera takes a port on the room's processor, and processors have limits — sometimes the second camera is cheap, but the bigger processor it forces is not. The tuning. An afternoon in your room, with your people, teaching the system when to switch and when to hold. Skip it, and the room cuts to the coffee machine every time it hisses. The finish. The right mount at the right height, power planned — not a bracket and a trailing cable.
The cheap quote is usually missing one of the four — most often the cable run or the tuning. The missing item doesn't disappear. It comes back later as a variation order, or as a room that cuts to the wrong shot in your first townhall. Our quotations list all four as separate lines, with the tuning hours written down, so the comparison you make is real. If another quote is thousands lighter, put it beside this list and ask which line is missing. That one question pays for this page.
Where each camera goes — and why wrong placement makes everyone look away from you
Television solved multi-camera seventy years ago and the grammar transfers straight into meeting rooms. Three shots cover almost every room. The audience shot: a camera at the display end, facing the people, so remote attendees see the room the way a person standing at the screen would. The presenter shot: a camera at the back or side, facing the stage or whiteboard, usually with tracking, so the presenter can move and stay framed. The content shot: a camera or capture path on the whiteboard, demo bench or document — the thing being talked about, not the person talking.
Then the rule that separates professional from painful: eyelines. Each camera lives next to the thing people look at from its zone. The audience camera sits at the display because the audience looks at the display — so on the far end, they appear to be looking at you. Put the audience camera in a corner and every remote attendee spends the meeting watching a room full of profiles. It costs nothing to get this right at design stage and a cable-pull to fix it later.
Crossfire is the last check: two cameras facing each other will each film the other camera. Offset angles, height discipline — cameras at seated eye level, not ceiling-corner security height — and the survey drawing that marks every camera's field before a single bracket goes on a wall. This is the twenty-minute conversation that decides whether the room feels like broadcast or like CCTV.
How the system knows who is speaking — and switches before the sentence ends
Automatic switching runs on one of two senses. Voice-triggered systems follow the microphones: whichever zone is speaking gets its camera on air. Simple, robust, and occasionally fooled — a cough at the back, a door, the projector fan — which is exactly what threshold tuning during commissioning is for. Vision-triggered systems watch the video itself: AI on the processor or the camera detects faces, movement and gesture, and picks the shot that shows the active speaker most clearly. It handles the two-people-talking-at-once moment better, and it is where the newest cores and cameras are pouring their intelligence.
Either way, judge a switching system on two numbers nobody prints on the box. Cut latency: the gap between someone starting to speak and their camera going live. Under a second feels natural; two-plus seconds means every remote attendee hears the first sentence from the wrong shot. Hold time: how long the system stays on a shot before it is allowed to cut again. Too short and the meeting becomes a music video; too long and interjections happen off-screen. These are tuned values, not spec-sheet facts — set in your room, with your people, during staging. It is why our commissioning is priced in days and demonstrated on a live call, not promised in a paragraph.
The moment this is really for: audience Q&A. Someone in the fifth row asks a question. In a single-camera room, remote attendees hear a faint voice over a wide shot of nobody. In a properly built multi-camera room, the ceiling microphone array does two jobs at once — it captures the questioner's voice cleanly with no microphone passed down the row, and it tells the directing system exactly where in the room that voice is coming from, so the nearest camera frames the person speaking. Remote participants see a face ask the question, the same way the room does. No runner with a handheld, no "can you repeat that." It is the difference between remote people attending your townhall and remote people overhearing it.
Microsoft Teams multi-camera in 2026 — the verified facts
This section exists because Teams' camera capability has been changing quarterly, and half the advice online describes last year. The following is checked against Microsoft's current documentation, this month.
Multiple camera view is real and native. Teams Rooms on Windows lets remote participants see up to four camera feeds from the room at once, each in 16:9, arranged in the room's video tile — with view switching that shipped in January 2026 and support in webinars and town halls when the organiser manages what attendees see. It is a presentation of multiple feeds, not automatic directing: remote viewers get the angles; nothing is cutting between them for you.
IntelliFrame is the AI alternative — and the two are mutually exclusive. Enable multiple camera view and the IntelliFrame toggle greys out; a room runs one or the other. IntelliFrame comes in two grades: Cloud IntelliFrame, which takes an ordinary camera's wide shot and composes individual face tiles in Microsoft's cloud, and Multi-Stream IntelliFrame on certified AI cameras, which adds people recognition — name labels under faces, names in the roster, who-said-what transcription — once staff enrol voice and face profiles.
The boundaries, in Microsoft's own words. Everything above requires the Teams Rooms Pro licence — Basic gets active speaker and panoramic views only. Cloud IntelliFrame shows its smart feed at nine or fewer people and is flagged by Microsoft as not ideal for large spaces, because distant faces blur under digital zoom — the same optical-zoom physics our large meeting rooms page explains. Which is the honest conclusion: in big rooms, platform AI rides on top of hardware directing rather than replacing it.
And one buying note ahead of the market: Microsoft's roadmap — updated this very week — previews enhanced IntelliFrame using certified edge-AI cameras with on-device processing, targeted for late 2026, exclusive to hardware carrying the new intelligent-camera certification. If you are specifying cameras this quarter for a room you will run for five years, that certification line on the datasheet is worth asking about. We are tracking it so you don't have to.
Camera cables go in before ceilings close — or you pay twice
Every multi-camera architecture moves video from camera positions to a director, and none of them do it over ordinary office Wi-Fi. Ecosystem processors take their cameras on dedicated point-to-point links over Cat6 — the manufacturer's own transport, uncompressed and in sync, never on the data LAN. Production cores take NDI over a managed network with the bandwidth planned, plus Dante for audio. Either way the runs cross the room: back wall to front rack, ceiling position to credenza. In a fit-out, those routes are drawn before containment closes. In a finished office, they are the single biggest cost variable in the quote — which is why our survey walks the ceiling before our quotation prints a number. A multi-camera room designed after the ceilings shut is a room paying twice for its own cables.
Boardroom, training room, townhall, classroom — the exact camera setup for each
The long boardroom (two cameras). Audience camera at the display; a second camera mid-table or at the far end so the near half of the table stops being haircuts. Method 1 or Method 3 with good glass. The most common upgrade we install — often onto an existing system, keeping the bar as the near camera.
The training room (two to three cameras). Presenter-tracking camera on the trainer, audience camera on the class, optional content camera on the whiteboard. Method 2 shines here: the same core that switches the meeting records the session for the learning library, over NDI, without a second system.
The townhall (three to four cameras). Stage camera with tracking, one or two audience cameras, a wide safety shot. Verified and worth knowing: Teams webinars and town halls now accept the room's multiple camera views on the live stage when the organiser manages the layout — your quarterly all-hands can cut between stage and audience natively. Ceiling arrays over the audience give Q&A both its audio and its camera targeting — a questioner is framed the moment they speak. Handheld microphones still apply for roaming presenters, with the staging validation our divisible rooms page explains.
The hybrid classroom (two cameras plus capture). Instructor tracking camera, student camera, content capture — with recording as a first-class requirement, not an afterthought. Education buyers: this is where AVer's tracking cameras built their reputation, and where the production-core method earns its keep on day one.
The multi-camera brands, compared honestly
Yealink — the reference native ecosystem: MVC S-series rooms with the AVHub directing multiple cameras over dedicated links, SkySound ceiling audio feeding the voice triggers, the sharpest AED for multi-room rollouts. HP Poly — the E70's twin-lens tracking plus the G7500 ecosystem; the choice where boardroom optics are a stated requirement. Logitech — Rally ecosystems with RightSight intelligence and the cleanest global-standardisation story. AVer — the education and venue specialist: CAM570 tracking cameras and the MT500 production core that directs, records and streams platform-free. Neat — design-forward rooms whose built-in intelligence covers small and medium multi-view needs without external directing. Shure — because voice-triggered switching is only as good as the microphones feeding it; MXA arrays are the trigger source in our most demanding rooms.
All stocked in Dubai — which is why the recommendation can be honest: your platform and zones choose the method, the method shortlists the brands, and the survey picks the winner in your actual room.
Send two photos on WhatsApp — get your camera count and budget the same day
Everything from the large-room survey applies — depth, ceiling, glass, HVAC, containment — plus this page's layer: the zone map (where presenting, sitting and content actually happen), the eyeline check (what each zone looks at), camera fields drawn against crossfire, the director's home (rack position and the cable routes to it), and your platform reality — Teams Pro, Zoom, guests, or all three — because Method 1 versus Method 2 is decided there, not in a catalogue.
Before any visit: WhatsApp us four things. A photo from the display wall toward the back. A photo from the back toward the display. One line on who presents and from where. One line on your platform. That is enough for a method recommendation and a budget range the same day, 8am–6pm Mon–Sat.
Multi-camera systems, exported from Dubai
Vector Digital Systems exports complete multi-camera packages — cameras, directing processors or production cores, cabling schedules, switching presets and commissioning documentation — from Dubai stock to 45 countries across the Middle East, Africa, South Asia, Central Asia and the Caucasus. Universities, broadcasters-turned-corporate and integrators receive pre-configured systems with the tuning notes that make on-site commissioning a day, not a week; air freight typically lands in 3–7 days. For export enquiries, WhatsApp our export desk.
What multi-camera clients say
"Our board table seated fourteen and the far six were invisible on every call. Vector added a second camera and the switching just works — whoever speaks, the right camera is already on them. Nobody in the room touches anything, which is the whole point."
"They tuned the switching with our actual trainers in the room — the camera follows the instructor to the whiteboard and cuts to students when they ask questions. And every session records through the same box. Two problems, one system."
"We ordered a three-camera townhall package for our Nairobi office through their export desk. Arrived pre-staged with the presets loaded and the cable schedule labelled. Their engineer walked our team through the tuning on WhatsApp video. First all-hands ran clean."
Multi-camera conference rooms — FAQ
Does Microsoft Teams support multiple cameras in one room?
Yes — verified against Microsoft's current documentation. Teams Rooms on Windows supports multiple camera view: remote participants can see up to four camera feeds from the room simultaneously, each in 16:9 within the room's video tile, with switching between views. It requires admin configuration on the room system and works in regular meetings, and in webinars and town halls when the organiser manages the attendee layout. Note it presents the feeds; it does not automatically direct between them.
What is the difference between multiple camera view and IntelliFrame — and can I use both?
They are different philosophies and a room runs one or the other — enabling multiple camera view disables the IntelliFrame toggle. Multiple camera view shows remote people up to four raw angles. IntelliFrame uses AI to compose the room into individual face tiles: Cloud IntelliFrame does it from an ordinary camera's wide shot, and Multi-Stream IntelliFrame on certified AI cameras adds people recognition — name labels, roster entries and attributed transcription after voice and face enrolment. Choose per room based on whether you want angles or a composed gallery.
Do these Teams camera features need a special licence?
Yes — the Teams Rooms Pro licence gates them. Multiple camera view, IntelliFrame and people recognition are Pro features; a room on the Basic licence gets active speaker and panoramic views only. It is a per-room annual line item and belongs in the budget from day one, exactly like the two-licence truth on our divisible rooms page. We state licence position explicitly in every quotation.
Can Cloud IntelliFrame handle a large room instead of extra cameras?
Microsoft's own guidance answers this: Cloud IntelliFrame shows its smart feed at nine or fewer people and is flagged as not ideal for large spaces, because faces far from the camera blur under digital zoom. That is the same optical-versus-digital physics that governs all large rooms. Small and medium rooms get genuinely good results from it; past roughly ten people or ten metres, hardware directing with real optical reach does the work and platform AI rides on top.
If I connect four cameras to the meeting PC, will the call use all of them?
No — this is the one-camera truth. A Teams or Zoom call consumes one video device at a time; extra cameras appear as a dropdown list, not as a directed production. Something must choose the live shot: an ecosystem processor selecting by voice, an AI production core selecting by vision, or the platform's own multi-view and IntelliFrame features within their limits. Buying cameras without deciding the director buys a menu, not a meeting.
How does the system know which camera to show?
By voice or by vision. Voice-triggered directing follows the microphones — the speaking zone's camera goes live — and is tuned with thresholds so coughs and door slams do not steal the shot. Vision-triggered directing watches the video with AI, detecting faces and gestures and choosing the camera that shows the active speaker most clearly. The best current systems blend both. Either way the behaviour is tuned in your room during staging, with hold times set so the meeting neither freezes on one shot nor cuts like a music video.
Can a second camera be added to our existing video bar system?
Very often, yes — it is our most common multi-camera retrofit. Modular native ecosystems accept additional cameras onto their processor within its port limits, and the existing bar usually survives as the audience camera while a tracking camera takes the presenter. The survey confirms three things: the processor's spare capacity, the cable route to the new position, and whether your platform configuration exposes the second feed the way you expect. If the existing system is a sealed all-in-one with no expansion path, we say so before quoting anything.
What does switching latency mean and why does it matter?
Latency is the gap between someone starting to speak and their camera going live. Under a second feels natural; beyond two seconds, remote attendees hear every first sentence over the wrong shot, which reads as a broken room. Its partner value is hold time — how long a shot stays before the system may cut again. Both are tuned during commissioning with people in the room, which is why serious multi-camera quotes price commissioning in days and prove it on a live call, not in a paragraph.
Can a multi-camera room record and live-stream as well as run meetings?
Yes — and it decides your architecture. Production-core designs do it natively: the same box that directs the cameras records to USB or network storage and streams over NDI, so a training session becomes a library asset with no second system. Native Teams and Zoom rooms rely on the platform's own recording of the meeting. Venues that need broadcast-grade outputs, overflow rooms or lecture capture lean to the core method for exactly this reason.
Can we mix camera brands in one multi-camera room?
Within one directing method, usually yes — production cores in particular are built to take multiple manufacturers' cameras over standard connections, and platform multi-view accepts any camera the room system recognises. Inside a native ecosystem, the processor's own camera family keeps the dedicated-link benefits and the cleanest tracking handoffs, so we mix only where there is a reason. The survey states which combination your chosen method supports before anything is ordered.
How many cameras does my room actually need?
Count the zones that matter, then stop. One seating zone within camera reach is one camera. A presenter plus an audience is two. Stage, audience and whiteboard is three. A wide safety shot makes four in townhalls. More cameras than zones adds cost and cutting noise, not quality — and some rooms that ask for multi-camera genuinely need one better-placed tracking camera, which we will say plainly when the survey shows it.
Two multi-camera quotes are far apart for the same room. What moves the price?
Five things: the cable runs (a dedicated link per camera through real ceilings — the biggest variable in finished offices), the director's class (processor ports or core channels, and the step-up when a camera count crosses a limit), the licence position (Teams Rooms Pro disclosed or buried), commissioning scope (threshold and hold-time tuning in days, proven on a live call), and mounts-and-power finished properly. A second camera priced like a webcam has skipped the run or the tuning, and the gap returns later as a variation order.
What does a multi-camera conference room cost in Dubai?
Supply and installation for typical two-to-three camera rooms: native ecosystem designs with a directing processor run roughly AED 45,000–110,000; AI production-core designs roughly AED 38,000–95,000; platform-native setups carry licence cost rather than directing hardware, on top of the cameras themselves. Adding one camera position to an existing compatible system typically lands in the AED 15,000–30,000 range installed and tuned — the cable run and commissioning, not the camera, set the figure. Room depth, ceiling works and display strategy move all of these; a survey converts the range into a fixed quotation.
Should we wait for the new edge-AI Teams cameras before buying?
Know about them; do not freeze for them. Microsoft's roadmap previews enhanced IntelliFrame on certified edge-AI cameras with on-device processing, targeted late 2026 and exclusive to hardware carrying the new intelligent-camera certification — existing USB cameras keep standard IntelliFrame. Practical buying rule: if you are specifying now for a five-year room, ask whether the camera line has that certification on its path; if your need is this quarter, current certified hardware already delivers the features on this page. We track the certification list so your quote reflects it.
What should we send on WhatsApp for a same-day recommendation?
Four things: a photo from the display wall toward the back of the room, a photo from the back toward the display, one line on who presents and from where, and one line on your platform — Teams, Zoom, or mixed guests. That is enough to recommend a directing method, camera count and budget range the same day, 8am–6pm Mon–Sat. The survey then draws the zones, eyelines and cable routes and fixes the quotation.
When someone in the audience asks a question, can remote participants see their face?
Yes — this is the Q&A moment a multi-camera room is really built for. The ceiling microphone array does two jobs at once: it captures the questioner's voice cleanly without a handheld being passed down the row, and it localises where in the room the voice comes from, so the directing system frames that person with the nearest camera. On Microsoft Teams, certified AI camera setups can also present in-room speakers as individual face tiles. The behaviour is tuned during commissioning — pickup zones mapped over the seating, thresholds set so a cough does not steal the shot — and demonstrated on a live call at handover.
Which areas of the UAE do you cover for multi-camera room installations?
Vector Digital Systems serves businesses across all seven UAE emirates — Dubai, Abu Dhabi, Sharjah, Ajman, Ras Al Khaimah, Fujairah and Umm Al Quwain — supplying, installing and supporting multi-camera conference room systems. Within Dubai, active client sites include Business Bay, DIFC, Jumeirah Lake Towers (DMCC), Barsha Heights (TECOM), Dubai Internet City, Dubai Media City, Dubai Silicon Oasis, Downtown Dubai, Sheikh Zayed Road, Jebel Ali Free Zone (JAFZA), Dubai Investment Park (DIP), Dubai World Trade Centre, Dubai Healthcare City, Al Quoz, Deira, Bur Dubai, Al Barsha, Dubai Marina and Dubai South. Across the Northern Emirates and Abu Dhabi: Musaffah Industrial City (ICAD), Khalifa Industrial Zone (KIZAD), Masdar City, Al Ain, SAIF Zone (Sharjah), Hamriyah Free Zone, Sharjah Industrial Area, RAKEZ, Fujairah Free Zone and Ajman Free Zone. On-site installation, configuration and fault response are available across all emirates; remote support is available UAE-wide.
Do you export multi-camera room systems from Dubai?
Yes. We export pre-configured multi-camera systems from Dubai stock to 45 countries: Saudi Arabia, Oman, Qatar, Kuwait, Bahrain, Iraq, Jordan, Kenya, Ethiopia, Tanzania, Uganda, Rwanda, Burundi, Djibouti, South Africa, Mozambique, Zimbabwe, Zambia, Mauritius, Seychelles, Nigeria, Ghana, Senegal, Côte d'Ivoire, Cameroon, Angola, Togo, Benin, Congo DR, Gabon, Egypt, Morocco, Algeria, Tunisia, Pakistan, Bangladesh, Sri Lanka, Kazakhstan, Uzbekistan, Turkmenistan, Kyrgyzstan, Tajikistan, Georgia, Azerbaijan and Armenia — pre-staged with switching presets loaded, cable schedules labelled, and full export documentation, air freight in 3–7 days.