Troubleshooting
Video tracking problems
When the path on screen isn't the path in the pool.
Video tracking works by contrast, so most tracking problems come back to light, colour or the camera. The original notes said it plainly: black pools "cause tracking problems, because they reflect lights. But it can be done." Here are the questions people ask most when the track goes wrong.
Questions people ask
The platform is plotted in the opposite position, or paths are mirrored. Why?
The pool edges were clicked in the wrong order during calibration; recalibrate in the order top, bottom, left, right. 1) Recalibrate: North (top), South (bottom), West (left), East (right). 2) Test with a light object at a known position. 3) For rectangular arenas, click top-left first, then bottom-right.
More detail
Calibrating East and West in reverse puts the target in the mirrored position; the order must be North (top), South (bottom), West (left), East (right). Recalibrate and check with a test object before trials. In rectangular arenas, clicking the bottom corner before the top corner reverses north and south values. Already-run files can sometimes be corrected by HVS (keep originals separately).
Going deeper
To avoid it, follow the white on-screen instruction at each click, and test with an object at a known position after every recalibration. Keep a backup of the calibration file once it is right.
The platform isn't shown, shows in the top-left corner or as a purple marker when running. What is wrong?
The platform number entered isn't one you calibrated. 1) The first calibrated position is platform 0 (usually the centre). 2) With the suggested positions, 1–4 are the quadrant centres (1 = NE, then clockwise). 3) With one calibrated platform, enter 0. 4) Enter the number in the platform number box (bottom left).
More detail
Numbers refer to calibrated positions in order: first = 0, then 1, 2… (up to 99). With the standard set, 0 is centre and 1–4 are NE, SE, SW, NW, so the maximum is 4; with eight manual positions use 0–7. An uncalibrated number is not shown, or is drawn at the top-left (0,0) or as a purple marker. A magenta disc shows when the trial ends automatically at the platform; a light-blue circle otherwise. If no positions were calibrated, recalibrate, adding platform positions after the four pool extremes (follow the white text at the bottom of the calibration screen), then Save.
Going deeper
Platform numbers are not quadrant numbers unless you calibrate centre first and then one per quadrant from NE, which is common practice. In a .txp, the platform number selects whichever calibrated position has that number.
My black pool or dark water isn't tracking white animals. What should I change?
Tick 'black' when calibrating a black pool; calibration defaults to white. 1) Recalibrate with the black box ticked: you should see a black mask round the pool. 2) Lower the light intensity above the pool to reduce reflections. 3) If using colouring, make the water dark enough for strong contrast.
More detail
Calibration assumes a white pool unless black is ticked; afterwards you should see a black mask around a black pool (white around a white one). HVS Image shows a red tracking cross for white animals on dark backgrounds. Black food colouring in a white pool works if the surface looks dark enough; if tracking is poor, lower the light above the tank and use more black colouring (not mixed primary colours). Reflections attract the cross: dim, diffuse lighting and a threshold adjusted until tracking is consistent.
Going deeper
Green cross = dark animal on light background; red cross = white animal on dark background.
The tracker sticks to the platform, or follows my hand at the start of the trial. How do I stop it?
The tracker scans from the top-left and locks on to the first dark object it finds, so keep your hands and darker objects out of its way. 1) Work from the south (bottom of screen) side. 2) Wear a white cap and sleeve when reaching to hard start points. 3) Adjust lighting so the platform is no darker than the water. 4) Set the threshold with a mouse in the south of the pool.
More detail
Because tracking starts from the north-west (top-left), an experimenter reaching from that side, or a platform darker than the water, can capture the cross. Fixes confirmed in support: reaching from the south-east wearing a white cap; rotating the camera so the hardest start point is at the top (never mid-experiment); adjusting lighting so the platform is not darker; and setting the threshold with a mouse in the south of the pool so it is tracked even if the cross started on the platform, then checking round the pool. After a camera change, use Check Lighting and send HVS a screenshot, stating dark-on-light or light-on-dark.
Going deeper
Start trials by clicking the remote as the animal enters (avoids false starts from hands); automatic starts can be wrong.
Why do I get straight lines on path plots, frame drops or a frozen trial timer?
Usually background programs or uneven lighting. 1) Close all other programs; schedule antivirus or security scans outside tracking. 2) In Task Manager, end unneeded tasks and disable unneeded start-up apps. 3) Make lighting even and diffuse (uplighters). 4) Run Check Lighting and send HVS a screenshot if it continues.
More detail
Straight lines towards the top-left, a freezing timer or dropped frames mean the PC paused (security scans, other software). HVS can configure settings and schedule scans outside tracking, and can liaise with IT. Straight lines within the area mean the cross is flicking between areas because of uneven light or shadows, or nothing dark enough to track: aim for even, indirect light with four uplighters around the pool and no dark pool edges. 'Multiple or interpolated points at start' is usually lighting or poor contrast. If an old PC still lags, try another computer.
Going deeper
To show HVS your set-up, take a screenshot of Check Lighting under trial lighting and send it with a short description.
There's no video picture, or it's black, or the wrong camera shows. What should I check?
Check the room lights, the camera connection and that no other program is using the camera. 1) Switch on the tracking lights before setting up (a dark room gives a black picture). 2) Close the camera maker's software: only one program can use the camera. 3) Unplug and replug into another USB port, restart the software (or PC). 4) Disable any built-in webcam in Device Manager.
More detail
Swapping different cameras between tests can cause connection errors: the software expects one camera per test (except 3D, which uses two). A USB camera can be disabled in Device Manager; enable it. After Windows updates some older webcams were not recognised.
Why do I see the laptop's built-in webcam instead of my tracking camera?
Disable the built-in webcam and restart HVS Image. 1) Open Windows Device Manager. 2) Disable the built-in webcam. 3) Check the tracking camera is plugged in. 4) Restart HVS Image. This is a quick fix.
More detail
If the picture comes from a built-in web cam, disable that web cam in the computer's Device Manager and restart the software. Also make sure the tracking camera was plugged in before starting; if it was not, restart the software after plugging it in.
Going deeper
Camera guidance: the HVS Image LabCam is recommended, but a wide range of USB web cams work. For long runs use a 10 m USB Active Repeater Hi Speed extension in a USB 3 port.
Some .hvs files give odd results (short latency, points outside the pool, rogue jumps). Can I fix them?
Use the original files or backups made at recording; corrupted copies lose data. 1) Corrupt copies (e.g. via USB) give short latencies, points outside the pool or no plot. 2) For a few rogue jumps, edit a copy: remove bad X,Y points and add the same number of estimated points to keep latency. 3) Always keep originals.
More detail
Yes, usually: odd results like these come from corrupted data or tracking jumps in the .hvs file, and you can often repair a copy of the file by hand. Corruption (garbage characters) causes shorter latency (lost end data), points outside the pool (altered numbers) or no plot (missing start data). Jumps to shadows show as sudden X/Y changes in the .hvs; remove them in a copy and insert the same number of estimated points between good points so latency is preserved; long stuck periods cannot be reconstructed, so use judgement. (0,0) is the top-left. Save as .hvs.
Going deeper
Analyse with unadjusted originals where possible and keep originals in separate subfolders.
If you're still on an older tracker, see whether to upgrade from earlier versions of HVS Image.
Spotted a mistake, or know a better source? Tell us and we'll fix it and say so on our corrections page.