Search options (Page 68 of 136)

Le besoin existe car les distorsions de la caméra posent des problèmes pour la précision des mesures et donc impactent toutes les fonctions d'analyse quantitative…
Avec la GoPro3 qui filme en 720p à 120fps ça vaudrait le coup de pouvoir rectifier les images.

Un peu sur le même principe, un stabilisateur de vidéos filmées sans trépied est également possible (en tout cas pour les vidéos sans pan ni zoom). Même topo : la fonction peut être implémentée sans trop de difficultés sur une suite d'images données mais pour faire du temps réel c'est plus compliqué.

I concur on what Chas said regarding zoom and shutter speed if your camera support changing these parameters. If you are using a webcam or similar device, place it so that what you want to see fills most of the image.

If you plan on making measurements, another important aspect is the calibration of the space. To make accurate measurements you should either make sure the camera is perfectly perpendicular to the plane on which you make all your measurements, or use the perspective grid to map a physical rectangle and make the measurements on that plane.

I'll try to do a more in depth post to compare the errors that can be expected from both calibration methods.

Yes. The point tool (cross marker) is now using the same tracking system as the other tools. To get back the trajectory tool you need to directly right click the object you want to track and choose "Track path". Do not add the cross marker.

Great !

The sad thing is that there actually was a similar feature in the ancestor of Kinovea back in 2005 (screencap from the help files). You could interactively add/remove positions to the composite.

In all those years I never got around to reimplementing it (Massive limitation : it worked only for perfectly fixed camera views. I want it to work even when there is camera motion, but it's much more involved).

Particularly interesting in the first images you posted is that the composition is selective. You see all the ball's positions, but only a few of the racket's ones. This makes for a much clearer image if all you want to see is the ball trajectory.

Hi,
I just tried and yes, you're right :-)

You need to set the .Owner property of the form to be the main Kinovea form for it to work properly.
If your menu handler is inside Kernel.cs, you can simply do this:

form.Owner = MainWindow;

If it's outside, in another module or elsewhere, you can call a similar function through this :

using Kinovea.Services;

Form form = new Form ();
DelegatesPool dp = DelegatesPool.Instance();
if (dp.MakeTopMost != null)
    dp.MakeTopMost(form);
form.Show();

Interesting… The timestamp part might be covered by the existing naming scheme feature. Have you tried that ? From the capture tab in the main preferences you can have the snapshots named automatically with a timestamp. If it needs to have more precision it would probably be simple to integrate.

Then the second part would be to collect the list of files/timestamps in a text log, csv or spreadsheet file.
This could be done entirely in an external program I guess, monitoring the output directory and collecting info on the fly. Or it could be added as an internal feature, but it sort of feel a bit specific…

Another way that might be interesting would be to be able to define an external command to be executed after each snapshot is saved. That would leave the possibilities open and allow for many other uses. The external program would be fed the file name, time stamp and other variables as parameters.

It's more advanced on the user side and would need a bit of programming to get it working but we can probably create scripts for common scenarios as the requests come.