Guide · 6 min read

Hand control

The four modes, every gesture, calibration and the camera permissions.

Updated

Munu can watch your hands through your webcam and let you drive the pointer and windows without touching the mouse. You can move the pointer, click, drag, scroll, right-click, grab and throw windows, give a silent thumbs-up “yes”, and call Munu with a raised fist.

The camera is never recorded. Video is read live on your computer, only to work out where your hands are. Nothing from it is recorded or saved, and it does not leave your computer on its own. The only time a picture leaves is when you ask Munu to look at something (see Look at this), and that sends a single picture, never a recording.

Hand gestures need a Mac with Apple silicon (M1 or newer) or a Windows PC. An Intel Mac cannot use them, and Munu says so: “Hand gestures need a Mac with Apple silicon.”

The Hand gestures step of Munu’s first-run window, with the camera switch and a Calibrate now button
The hand gestures step in Munu’s first-run window.

Turning it on

Hand gestures are on by default, with Let gestures move the pointer on too. An Intel Mac leaves them off quietly. Any of these turns them off or back on.

  • First run. Setup has a Hand gestures step, with Use the camera for hand gestures ticked by default. Keep it ticked and Munu asks for the camera there and then. Untick it to turn hand gestures off. You can turn them on later.
  • Settings. Settings > General has the same switch, plus which camera to use and Let gestures move the pointer. Turn that second one off to use only the talk gesture.
  • Tray menu. A Hand control item you can tick or untick at any time.
  • By voice. Say “Munu, start hand control” or “Munu, turn on hand gestures”. “Munu, stop hand control” or “Munu, turn off hand gestures” turns them off. Munu answers “Hand control is on.” or “Hand control is off.”

Turning hand gestures on does not hand over your computer. Munu only watches your hands at first (see the four modes). Turning them off, by any of the ways above, stops the camera and immediately lets go of anything a hand was holding.

Permissions

  • Mac: Camera. The first time you turn hand gestures on, macOS asks to let Munu use the camera. If you said no: System Settings > Privacy & Security > Camera, then switch on Munu.
  • Mac: Accessibility. Moving the pointer and windows needs one more permission: System Settings > Privacy & Security > Accessibility, then switch on Munu. Without it, Munu still tracks your hands and the talk gesture still works, but switching to PC control tells you to allow Accessibility first.
  • Windows: Camera. Settings > Privacy & security > Camera: allow camera access and allow desktop apps to use it.
  • Windows: admin windows. Windows does not let an ordinary program control a window that is running as administrator, so hand gestures cannot move, resize or click into one. Everything else works as normal.

If no camera can be found, or another app is already using it, Munu says so and keeps trying every few seconds. Nothing else is affected.

The four modes

Hand gestures start in Watching: your hands are tracked and drawn on screen, but nothing you do moves the pointer or a window. Switch modes with a two-hand pose, held still for about 0.6 seconds. Crossed arms is held a little longer, about 0.7 seconds, so it does not trip by accident.

The four modes and the pose for each
PoseModeWhat it does
Both thumbs upPC controlYour hand drives the pointer and windows.
One thumb up, one thumb downWatchingHands are tracked, nothing is controlled. This is where it starts.
Both thumbs downGestures offHands are seen but nothing acts.
Cross your forearmsPausedEverything on screen hides. Cross again to resume.

Every switch shows a short message at the top of the screen for a couple of seconds and plays a small tone. Switching into PC control also shows a reminder of the gestures for a few seconds. Switching modes always lets go of anything you were holding or clicking. While paused, crossing your arms again resumes. You can also say “Munu, hide the overlay” to pause and “Munu, show the overlay” to resume.

The gestures

The first hand Munu sees becomes your pointer hand. It keeps the pointer even if your other hand appears, until you lower it. The pointer follows that hand around an area in front of the camera that maps to your whole screen. Moving the pointer, clicking, scrolling and moving windows need PC control. The mode poses, calling Munu and the thumbs-up work in Watching too.

Gestures and what they do
Do thisWhat happens
Move your handThe pointer follows it
Pinch thumb and index finger, let go quicklyClick
Pinch, hold and move, then let goDrag
Pinch thumb and middle fingerRight click
Two fingers out together, then moveScroll, like a touchscreen
Make a fist over a windowGrab the window
Open your fist while moving fastThrow it: sideways snaps to half the screen, up maximises, down minimises
Both fists on one window, move apartResize the window
Raise one fist and hold it for about 2 secondsMunu starts listening
Thumbs up after Munu asks you somethingA silent “yes”

Two quick clicks make a double click. If Let gestures move the pointer is off in Settings, or Accessibility has not been allowed, PC control gestures are not available, and Munu says so when you try to switch to it.

Not sure how a gesture goes? Ask “Munu, show me how to scroll” (or click, grab a window, switch to PC control, pause) and Munu shows it on the glass stage with an animated hand. It needs no camera and works with hand control off. Ask “Munu, which hand gestures are there?” for a card listing them all.

Two windows at once

Each hand can carry its own window. Make a fist over one window with one hand and a fist over a different window with the other, and you can move, throw or snap each one on its own. If both fists land on the same window, that grab becomes a resize instead: move your hands apart or together to resize it around its centre, between about a quarter and four times its original size. Opening either fist ends the resize and leaves the window where it was.

Calling Munu with a gesture

Raise one fist, high up and roughly in the middle of the picture, and hold it still for about 2 seconds. Munu opens its microphone to listen, the same as choosing Talk now from the tray. This works in Watching and PC control, not while paused or off, and it interrupts Munu if it is already speaking.

Thumbs up means yes

Hold one hand up with just the thumb pointing up for about half a second. Munu takes it as a silent “yes”, and a gold flash lights up around that hand’s dot. It only counts right after Munu has asked you something and is waiting for an answer. A thumbs-up never confirms a risky action. For that, Munu only accepts “Munu confirm” spoken out loud.

What you see on screen

While hand gestures are on, a small, transparent overlay that you can click through shows:

  • a dot at each visible hand’s aim point: a small ring in PC control, a larger soft dot in Watching, and a hollow grey dot in Gestures off. The ring fills in as a pinch closes, and there is a small ripple on each click;
  • an outline around a window you are holding, and a brief message when it snaps into place (“Snapped left”, “Snapped right”, “Maximized”, “Minimized”);
  • a gold flash around a hand’s dot when a thumbs-up “yes” registers;
  • the mode message (“PC CONTROL”, “WATCHING”, “GESTURES OFF”, “PAUSED” or “RESUMED”) and, when you switch into PC control, a short reminder of the gestures.

The aim point is where your pinch closes. While you pinch it is the point between your thumb tip and index tip. While your hand is open it is the spot in front of your fingers where the pinch will close. So the dot does not jump when you pinch.

By default the overlay never shows the camera picture itself, only the dot. If you want to see your whole hand, turn on Show a mesh of my hands in Settings > General, or say “Munu, show my hands”. It is off by default. To see the camera, use the voice commands in the next section. None of this changes what leaves your computer: the picture is still only looked at, live, here.

While paused, none of this is shown after the “Paused” message has had its couple of seconds. The dots, the hand mesh, any camera view and the webcam window all disappear.

Voice commands for hand control

Voice commands for hand control
SayWhat happens
“Munu, start hand control” / “Munu, stop hand control”Turns hand gestures on or off.
“Munu, open the stage” / “Munu, show the overlay”Shows the glass stage, and un-pauses hand control if it was paused. It never turns the camera on.
“Munu, close the stage” / “Munu, clear the stage”Hides the glass stage (its cards stay for next time) / takes every card off it.
“Munu, hide the overlay” / “Munu, minimize the overlay”Pauses hand control. The same as crossed arms.
“Munu, close the overlay”Turns hand control off entirely.
“Munu, show my camera” / “Munu, hide my camera”Shows or hides a faint, mirrored camera view behind everything else.
“Munu, clearer camera” / “Munu, fainter camera”Makes that camera view easier or harder to see.
“Munu, show myself” / “Munu, hide myself”Shows or hides the whole camera picture, with your hand dots on top and no tracking lines.
“Munu, show my webcam” / “Munu, hide the webcam”Opens or closes a small mirrored preview window in the bottom-right corner of the screen.
“Munu, show my hands” / “Munu, hide my hands”Shows or hides a faint drawing of your whole hand. It never turns the camera on.
“Munu, calibrate my hands” / “Munu, recalibrate my hands”Opens hand calibration, and turns hand control on first.

You do not have to get the wording exactly right. Only an explicit camera request turns hand control on: “start hand control”, “show my camera”, “show myself”, “show my webcam”, or the same in your own words, such as “put me on the screen”. Everything else, such as “clearer camera”, “fainter camera”, pausing or any other adjustment, only works while hand control is already on. If one of your own shortcuts uses the same words as a phrase above, your shortcut wins.

Calibrating your hands

Gestures work out of the box, but a short calibration fits them to your hands, your camera, your screen and how far you sit from it. It takes about 30 seconds. Start it in any of these ways:

  • Calibrate my hands in Settings > General (in the Hands section);
  • Calibrate now on the first-run Hand gestures page, which runs it right there, before the free trial;
  • by saying “Munu, calibrate my hands”.

It turns hand control on if it is off. The window fills the screen, and each step starts after a short countdown.

  1. Point (10 seconds). Five amber dots light up one at a time, in the middle and near the four corners of the screen, for 2 seconds each. Look at the screen, pinch your thumb and index finger where the dot is, as if you were touching it, and hold until the white ring closes. A dot turns green when it caught your hand.
  2. Rest (2 seconds). Hold your hand still. Munu learns how far you sit from the camera.
  3. Pinch (about 5 seconds). Pinch and open, three times. Munu learns how closed your pinch is.
  4. Fist (about 5 seconds). Make a fist and open, three times. A fist that does not quite close is accepted, within limits.
  5. Throw (about 5 seconds). Grab and throw to the side, twice. Munu learns how fast your throws are.

Steps 2 to 5 show a blue hand to copy next to your own hand as the camera sees it. While the steps run, hand control holds still, so nothing gets clicked, grabbed or thrown. At the end, the window says in one sentence what changed, lists the numbers hand control now uses next to the usual ones, and saves it. The numbers are in plain words: speed (how far the dot moves for each centimetre of hand movement), reach (how many centimetres of hand movement cross the whole screen) and two-hand distance, plus the pinch, fist and throw levels. The centimetres assume an average palm, about 10 cm from wrist to knuckles, so they are a guide, not a measurement.

If a step did not work, for example no hand was in view, that part keeps the usual setting, the window says so, and the rest still counts. Cancel keeps your previous calibration. Reset calibration in Settings > General goes back to the usual settings. Only the resulting numbers are saved, never a picture.

Look at this

Ask Munu to look at something, such as “look at this”, “what am I holding” or “read this”. It takes one picture from the camera and sends it to answer, the same way any other question goes. This only works while hand control’s camera is already running. Asking Munu to look does not turn the camera on by itself, and Munu will not look while hand control is paused. Nothing is saved from that picture, and the log only notes that one was sent, never what was in it.

The glass stage

The glass stage is a see-through board on your screen where Munu puts cards: a note, an answer, a picture, a video, a 3D model, some numbers or a file browser. Say “Munu, close the stage” to hide it, “Munu, open the stage” to bring it back and “Munu, clear the stage” to take the cards off. These gestures work in Watching and in PC control, as long as hand control is on and not paused or off. In PC control they only apply when your hand is over a card. A pinch here is thumb and index finger together.

Some glass stage gestures
GestureHow to do itWhat it does
TapPinch on a card and let go quickly, without moving your hand.Opens or closes a note, plays or pauses a video, or brings a card to the front.
Move a cardPinch on it, hold and move, then let go.Carries the card and drops it where you let go.
Fling awayPinch a card, swipe fast and let go.The card flies off the screen. Ask Munu to show it again.
Make a card bigger or smallerPinch the same card with both hands, then move apart or together.Resizes it between a quarter and four times.
Dismiss a cardTap it with thumb and middle finger.Removes the card.
Clear the stageHold both hands flat, fingers up, well apart. Bring them together like a clap.Clears every card.

Tips for good tracking

  • Light. Aim for even light on your hands. Do not sit with a bright window or lamp behind you.
  • Distance. Sit about an arm’s length from the camera. Too close or too far is harder to see.
  • Framing. Keep your hands well inside the picture, not at its edge, with the whole hand showing. Do not hide your fingers behind each other or behind a cup or a phone.
  • One hand for the pointer. The first hand Munu sees keeps the pointer. A second hand only matters for two-hand poses, such as mode switches and resize.
  • Slow and clear beats fast. Make each gesture on purpose and hold it briefly. Two-hand poses need to stay still for a moment.
  • Your mouse always wins. Moving the mouse or trackpad hands the pointer back to it. Hand gestures take over again about half a second after you stop.
  • Pointer cannot reach the edges, or gestures keep missing? Calibrate your hands.
  • Not sure Munu sees you? Say “Munu, show my camera” to see the picture, and “Munu, hide my camera” to hide it again. The tray’s Hand control menu also has Show gesture details, a small live panel with the numbers behind the pose you are holding.
  • Turn it all off with the Hand control switch, or say “Munu, stop hand control”. The camera stops at once and anything a hand was holding is released.

More fixes are in the troubleshooting guide.

Say its name. It answers.

Munu is a desktop voice assistant for Windows and Mac, with memory you own and hand control through your webcam.