The room already has an audience. IMAG is what lets all of it see.
The term is exactly what it looks like — a live camera feed, magnified onto a screen, for the people already sitting in the room.
IMAG is Image MAGnification, and the trade press has used the term since well before streaming made cameras familiar equipment on a corporate stage. It solves a narrower problem than a broadcast does. A general session’s sound system reaches the back row at close to the same level it reaches the front, so a presenter at a lectern is heard clearly from every seat in the room. Seeing them does not scale the same way — past a certain distance a face stops reading as an expression and starts reading as a shape, and no amount of additional sound pressure fixes that.
A stream is a second production for a second audience — people who are not in the building, watching on a device with its own screen and its own speaker, off a feed carrying its own separate mix. IMAG can share the same cameras a stream would use, and sometimes the same switcher, but its only audience is the one already in their seats. Nobody watching from home ever sees the IMAG feed, and nobody in the room, glancing at the screen beside the stage, is watching a stream.
Two cameras is the floor. Each one is doing a different job.
A locked-off wide is the shot that is always right. An operated tight is the shot doing the actual magnifying.
A tight camera set up close to the stage crowds the presenter’s own space and blocks a sightline somebody paid for a seat to have. The standard fix runs the other way: further back on a longer lens rather than closer on a shorter one. Broadcast trade coverage of IMAG has documented the reason for years — a long lens is generally not built to focus on a subject closer than about twenty feet, and a stadium-scale lens can run to eighty or a hundred times magnification and beyond, mounted on a tripod built to carry the lens rather than the other way around.
That still leaves the camera somewhere. Its platform is the same floor-plan decision a tight shot on a lectern always is — low enough and far enough back not to block a view, high enough to clear the heads in front of it, which usually means it stands where seats were going to go, or on a riser built for the job.
- The wide
Set once, covering the whole stage, left alone for the rest of the program. It is the frame a director cuts to whenever anything unplanned happens — a stumble, a walk to a different mic, a handoff — because it is never wrong about where the presenter actually is.
- The tight
Operated, and following the presenter rather than the stage. This is the camera doing the magnifying the term is named for, and it is the one an audience actually watches once the program is underway.
A wall sized for the slide deck is answering a different question.
The 8:1 rule sizing a general session’s wall is real — and it is calibrated for a different task than reading a face.
The 8:1 rule — divide the furthest seat’s distance by eight for the minimum comfortable image height — is correct, and it is the right number for what it was built to measure: a viewer who needs to recognize a slide and follow a video without studying either one. AVIXA, the standards body behind the contrast-ratio standard a general session’s wall is also measured against, publishes a companion figure for image size, and it does not treat every viewing task as the same distance problem. Divide the farthest seat by eight for passive viewing of a moving image. Divide it by six for basic decision making — recognizing a face is AVIXA’s own example of this category. Divide it by four for analytical work, reading a spreadsheet or a CAD drawing in detail.
A general session wall is usually sized to the 8:1 figure, because that is the right test for the content it spends most of the program showing — slides, a sizzle reel, a speaker’s name on a lower third. The moment a live camera feed of a face goes up on the same wall, the task the back row is doing changes, and a distance that was comfortable for a slide is not automatically comfortable for a face. Basic decision making’s 6:1 threshold is the tighter number IMAG has to clear, not the 8:1 figure the rest of the program is sized to.
A cut has to read as instant. Most of the delay budget is not in the switcher.
A director calls the cut over an intercom circuit; the box that executes it is rarely where a delay problem actually starts.
Every camera feeding one switcher runs off a shared timing reference, called genlock, so a cut from one source to another lands as a clean frame rather than a roll or a tear. A director watches the wide and the tight on a preview monitor and calls the cut over an intercom circuit that also reaches the camera operators — the same discipline as any live cut-based production, run at whatever scale the show needs, down to one operator calling their own shots on a single-camera job.
The switcher itself is not usually where a delay budget goes. Blackmagic Design’s own published specification for its ATEM line states a processing delay of under 10 lines when a source is genlocked and in the same format as the switcher. A 1080-line HD frame runs 1,125 total lines top to bottom, blanking included — the count the format is built on — so 10 lines is under 1 percent of a single frame’s scan. Whatever delay shows up between a presenter’s mouth and their image on the wall is accumulating somewhere else in the chain: a source that has to be scaled to match the switcher’s native format, and the wall or projector’s own image processor at the far end.
What that delay has to clear before it is a problem is a published number, not a guess — the same broadcast sync tolerance already worked through frame by frame against a live orchestra timed to picture: audio and image can drift apart by tens of milliseconds before a trained viewer notices. The fix on a corporate stage is the one a broadcast truck already uses: delay the audio to match the video path, using the same delay control every digital mixer already carries.
The camera missing the presenter, and the monitor nobody budgeted.
Two different failures, and both come from treating IMAG as one line item instead of two separate jobs.
An operated tight camera is only as good as the operator’s read of where a presenter is about to go, and a presenter who was never walked through their own blocking will drift off the mark the camera was set for — toward a screen to point at something on it, off the riser to greet somebody in the front row, out of the tight shot’s frame entirely. The fix is the same rehearsal discipline a general session runs for any other reason: mark the walk, the gesture, the moment they step down, before the room fills.
The screen the audience watches and the screen a presenter needs are not the same piece of equipment, and a plan that budgets for one and calls it done leaves the presenter with no way to check their own slide, their own notes or their own place in the deck without turning around to find it on the wall behind them. A confidence monitor — small, facing the presenter rather than the room — is the actual fix, and it carries the presentation, not the IMAG feed: putting a presenter’s own magnified face in front of them while they are talking is one of the fastest ways to put them off the sentence they are in the middle of.
Sized and cut correctly, nobody in the room thinks about the camera.
IMAG works when it disappears — when the back row reads an expression without registering that a camera put it there. Getting there is two cameras doing two different jobs, a screen sized to the task the audience is actually doing rather than the one it was doing five minutes earlier, a cut that lands clean because the delay was budgeted rather than discovered, and a presenter who has somewhere of their own to look that is not their own face.
Where the figures come from
- Long lenses generally not focused closer than about 20 feet; stadium lenses at 80x, 100x and beyond — documented broadcast-camera practice for IMAG coverage, published by Streaming Media (Anthony Burokas, "What Is IMAG?"). Used in §02 to explain why a tight camera sits back on a long lens rather than close on a short one.
- 8:1 for passive viewing, 6:1 for basic decision making, 4:1 for analytical work — AVIXA’s published display-image-size standard, the companion figure to the AVIXA/ANSI contrast-ratio standard already cited on the LED-versus-projection piece. Basic decision making is AVIXA’s own category for recognizing a face; §03 uses it to show why the 8:1 figure that sizes a slide deck is not the number IMAG has to clear.
- Processing delay: under 10 lines when genlocked and in the same format as the switcher — Blackmagic Design’s own published technical specification for its ATEM switcher line.
- 1,125 total lines in a 1080-line HD frame — SMPTE 274M, the standard defining the format — the count §04 divides the switcher’s own delay against to show it is under 1 percent of a single frame.