Stop losing eye contact while reading a script.

Almost all of it is geometry. Not confidence, not charisma, not "being bad on camera". Where the text sits relative to the lens, how far away you are, and how wide the lines are. Get those three right and most of the problem disappears without you performing differently at all.

What the viewer is actually seeing

Two separate things read as "not looking at me", and they have different causes.

Gaze offset. Your eyes are pointed somewhere other than the lens. This is what happens when the script is on a second monitor to the left, or on your laptop screen below a camera mounted on top of it. The viewer sees you addressing a spot slightly off to one side of them, permanently.

Gaze motion. Your eyes are roughly on the lens, but they are moving — sweeping left to right along a line, jumping back for the next one, or darting ahead to check what is coming. This is the one people underestimate. A static offset can read as thoughtful. Constant motion never does. It reads as reading.

Fix them in that order, because gaze offset is cheap to fix and gaze motion is where the real work is.

Fix one: distance does most of the job

The angle between the lens and the text is what matters, not the physical gap. That angle shrinks as you move back. So the cheapest improvement available to anyone is to sit further from the camera.

At arm's length from a laptop, text on the screen below the webcam is a large angle and it is obvious. Push the same setup back a couple of metres and the same physical gap becomes a small angle that a viewer will not resolve. You have changed nothing except where you sat.

This is why studio setups look effortless and desk setups do not. It is not the camera. It is that the subject is further away with a longer lens, so the prompter can sit a hand's width from the lens and still be invisible.

Practically, for a room you already have:

• Back up as far as the room allows, then zoom or crop in to reframe rather than moving closer.
• Put the text directly under or above the lens, never beside it. Vertical offset is far more forgiving than horizontal, because everyone's eyes move vertically when they think. Sideways glances read as evasive.
• Raise the screen so the text sits just below the lens rather than down on the desk. A stack of books is a legitimate piece of equipment.
• If you use a second display for the prompter, put it under the camera and turn the laptop away, rather than leaving both in view. Two text sources is two reasons to glance.

Fix two: narrow the column, enlarge the text

Once the offset is handled, the remaining giveaway is motion, and motion is a typography problem.

Wide lines force horizontal sweeps. Every sweep is a visible eye movement, and there is one per line. Narrow the column until each line is a short phrase and the sweeps mostly disappear, because your eye can take the whole line in without travelling across it. This costs nothing and it is the single highest-return setting in any prompter.

Bigger text helps for the same reason: you can read it with less precision, from further away, without leaning in. Leaning in is itself visible.

And break the script into short lines by meaning, not by paragraph. One idea per line, blank space between blocks. A wall of prose makes you hunt for your place every time you look back; short lines let you land back in the right spot instantly.

A useful test: film 30 seconds, then watch it with the sound off. If you can see your eyes working, the column is too wide or the text is too small. Sound off is the important part, because with sound on your brain forgives the eyes.

Fix three: the part geometry cannot reach

Here is where honest advice runs out, because there is a category of eye movement that no amount of distance or column width will remove.

If the text is scrolling on a timer, you are permanently checking your position against it. That check is an eye movement, and it is the specific one that reads as anxious rather than thoughtful. When you are slightly ahead of the line you look down to find where the machine has got to. When you are behind it you look ahead to see how much is coming. Neither is a reading movement — both are monitoring movements, and they happen even when your column is perfect and your lens is two metres away.

You can feel this yourself. Set a fixed scroll, record a minute, and count the moments where you were not reading but checking. That number does not go down with practice. It goes down when there is nothing to check.

Which is the argument for a scroll driven by your voice rather than a clock. If the text advances because you said the words, your position in the script is never in question, so the monitoring glances stop existing. What is left is ordinary reading, in a narrow column, near the lens — which is what looks like eye contact. That is what Mirrento does, and if it is the monitoring glances you recognise rather than the geometry, the 7-day trial is enough to film the same paragraph both ways and watch the two back with the sound off.

The caveat, stated plainly: this fixes one class of eye movement. If your text is on a laptop screen beside you at arm's length, voice tracking will not save you, and no prompter will. Geometry first, always. Software only helps with what is left.

What does not work

Three things I have tried or been told, which do not survive contact with footage.

"Just look at the lens more." You cannot read text you are not looking at. Consciously forcing eye contact while reading produces a worse artefact: long unblinking stares broken by panicked glances back to find your place.

Memorising to avoid reading entirely. It works, and it costs hours per video, and the recall effort produces its own upward-and-left thinking glance. There is a middle path that is cheaper than both, which I wrote about in why learning your lines is usually the wrong investment.

Buying a beam-splitter rig first. A proper prompter mirror puts the text on the optical axis and it genuinely is the best answer to gaze offset. But it fixes offset only. Bought before you have narrowed your column and sorted your pace, it will not do what you hoped, and you will conclude that you are simply bad on camera. You are not — you have three problems and you solved one of them.