61 Reply by joan

(1 replies, posted in General)

If you are not using the center of gravity point then it doesn't really matter, you can place the joint points exactly where you need them to be according to your needs. The CoG placement is based on segment weights coming from old anthropometric tables. You can build an alternate human model by modifying the existing one if you need more customization.

Go to forum: General

62 Reply by joan

(1 replies, posted in General)

Hi, there should be one or more files in the log folder with names that start with "Unhandled exception", could you paste the content of the most recent one here as well? This will contain more precise information for debugging.

But more importantly this is from an old version published in 2011, could you try and see if you reproduce the problem in the latest version? A lot has changed in the comparison/synchronization system.

Go to forum: General

63 Reply by joan

(18 replies, posted in Cameras and hardware)

It appears the issue with FLIR cameras is that when you have two of them of the same model connected at the same time, the identifier we use doesn't differentiate between them and this causes a crash. Single camera should work.

Go to forum: Cameras and hardware

64 Reply by joan

(2 replies, posted in Bug reports)

Hi,
What version of Kinovea are you using? There were some important changes in 2024.1, it's possible there is a regression I haven't noticed in my testing.
Is the camera static? Are you using camera motion estimation? When you say a digitized point do you mean a new static marker or a new point along a trajectory?
Indeed a given point on a given frame shouldn't change coordinates unless the coordinate system is tracked or the calibration object is tracked, is it your case?

Go to forum: Bug reports

65 Reply by joan

(18 replies, posted in Cameras and hardware)

That's great to hear!

Yes the Spinnaker SDK has the "FLIR_GenTL_v140.cti" file which is a bridge between the actual device driver and the GenICam libraries. (In the case of user aroe79 above they already had that installed based on the log so that wasn't the issue).

I'll add a little paragraph to the first post to clarify.

Go to forum: Cameras and hardware

66 Reply by joan

(1 replies, posted in Español)

En el menu Herramientas > Calibración del tiempo (puede que sea otro nombre según la version de Kinovea). En este dialogo se puede cambiar el eje del tiempo. Si quieres que el video se reproduzca a 120 fps, ignorando el valor escrito en el archivo, puedes cambiar en la parte inferior.

Go to forum: Español

67 Reply by joan

(1 replies, posted in General)

No, at the moment you cannot do any tracking in the capture screen on a live camera feed.

Go to forum: General

68 Reply by joan

(2 replies, posted in Ideas and feature requests)

Check via Tools > Scatter diagram. It shows the positions of all the markers and has a "Copy to clipboard" function that copy the data as CSV.

Go to forum: Ideas and feature requests

69 Reply by joan

(3 replies, posted in Cameras and hardware)

The compatibility table at the moment is the following:
- Kinovea 2024.1 + plugin Kinovea.GenICam-1.1.0 = Basler Pylon any version.
- Kinovea 2023.1 + plugin Kinovea.Basler-1.3.0 = Basler Pylon 6.1
- Kinovea 0.9.4 and 0.9.5 + plugin Kinovea.Basler-1.2.0 = Basler Pylon 6.0

Kinovea 2024.1: https://www.kinovea.org/en/forum/viewtopic.php?id=1959
Kinovea.GenICam-1.1.0: https://www.kinovea.org/en/forum/viewtopic.php?id=2005

Go to forum: Cameras and hardware

70 Reply by joan

(18 replies, posted in Cameras and hardware)

So far I have received feedback from 3 persons with Teledyne FLIR cameras. It worked for two of them and didn't work for the third one, but I'm not sure what the problem is. So any additional feedback helps.

Go to forum: Cameras and hardware

71 Reply by joan

(4 replies, posted in Cameras and hardware)

Certain machine vision cameras go in this direction of the trade off since for industrial inspection they also often need more resolution and don't care that much about frame rate. I see devices listed with 5496×3672 at around 5 fps for example or even 5320×4600 at 15 fps. But they can be quite pricey.

I wonder if at a certain point, maybe 0.5 or 0.1 fps, the best approach would be to have something that takes regular photos and save them to a folder, and then something else is taking these images and produces an MJPEG stream out of it.

If the images are saved directly to the computer it could even bypass the stream/capture step entirely. Kinovea can open a folder of images with sequential names and load it as a video. I have not tested this with very large images, the explorer will be quite slow to create the thumbnails which isn't necessary for this. I think for 8K images at 0.5 fps or less this might be an interesting option.

There is also latency and lens quality to think about.

Go to forum: Cameras and hardware

72 Reply by joan

(1 replies, posted in General)

For versions before 2023.1 (including 0.9.5) the export size depended on the current view port size so it will change depending on your screen resolution. For 2023.1 and newer it only depends on the original size so saving dual FHD videos should result in 3840x1080 or 1920x2160 if saved vertically.

Go to forum: General

73 Reply by joan

(3 replies, posted in General)

For others: you can change it on a per-drawing basis but if you would prefer the default for all drawings to always be visible you can change the default visibility in Options > Preferences > Drawings > Opacity, and check Visible for the entire video.

Go to forum: General

74 Reply by joan

(1 replies, posted in Bug reports)

It's a bug I'm also seeing it in various videos but haven't had time to investigate more deeply yet.

Go to forum: Bug reports

75 Reply by joan

(18 replies, posted in Cameras and hardware)

Thanks for testing and sorry it's not yet working correctly. Thanks for the log by mail. I don't think SwingCatalyst is interfering. I'll investigate.

Go to forum: Cameras and hardware