361

(18 replies, posted in General)

I made a fix and uploaded a new version of the Daheng plugin at the following location:

Kinovea.Daheng-1.0.1-0.9.5-x64.zip.

It's built against 0.9.5 but it should also work in 0.9.4 as-is. Let me know if this works.

362

(18 replies, posted in General)

The problem appears to be a vertical flip. The rotation makes it look like horizontal mirroring.
I can reproduce the issue when I switch to "Raw" stream mode instead of "RGB". The problem seems to be with any monochrome stream.
And do you also have the issue on the thumbnail?
Thanks

363

(18 replies, posted in General)

Great, thanks for the video this is very helpful. I'll investigate what's going on.

364

(13 replies, posted in General)

What type of metrics are you interested in in the kinematics plots?
At the moment there is no kinetics measurements as there is currently no way to assign weights to points.

365

(13 replies, posted in General)

Hello everyone!

This is Kinovea 0.9.5.
Please report any issues as soon as possible.

This is a relatively small incremental release over 0.9.4 since the main effort was to update the reference manual which hadn't been updated since version 0.8.15.

The new manual is here: Kinovea manual - 0.9.5
This version also include a number of bug fixes, new logo, and the installer is now signed with my digital certificate.

Thanks!

Don't hesitate to post feedback, questions, feature requests and bug reports! You can do this either in this thread or create new dedicated threads.

366

(18 replies, posted in General)

I just tried it without reproducing the problem. Some questions:

Do you have mirroring active on the camera side?
Image rotation?
Do you have a default playback KVA file that is loaded with every video that has mirroring maybe? (Preferences > Playback > General > Default annotation file).

Maybe the mirror flag was active at some point and it's stuck in the KVA file somehow.
Can you locate the last captured file and see if there is a .kva file of the same name, open it, and look in the first lines for <Mirror>true</Mirror>.

In any cases the capture side and the playback side should be the same, either mirrored or not.

The difference between compressed and uncompressed is puzzling. Is the camera streaming MJPEG or something else?

I hope they didn't remove support for the .NET wrapper. Don't really have time to rewrite the Basler plugin from scratch at the moment. I see they removed the older SDK from the download area so it's not even possible to get 6.1.x without contacting them…

Most probably the difference is coming from the extra smoothing pass which is done for high speed footage for velocity and acceleration. This was added because the small time intervals amplifies the digitization noise on the derivatives.

For the next version I have added an option to disable this in the preferences so you can get the raw data. Disabling the main smoothing will also now impact how curves are drawn in the graphs and tracks.

edit: Actually this extra smoothing wasn't implemented for angular kinematics so I'm not sure where the difference could be coming from.

Most probably the device that recorded the 1000 fps video marked the file as 20 fps, this is common. You should go to menu Video > Configure video timing, and in the "High speed camera" section, set the frame rate to 1000.

370

(1 replies, posted in General)

Hi,
At the moment the data is not exported. It's not too hard to add I think, I'll look into it.

You can use right click at the end to choose if you want to keep or discard the last point. ("Finish polyline here" or "Finish polyline at previous point"). Once it is committed you can't add or remove points from the point list.

372

(18 replies, posted in General)

Hi,
I can't reproduce the problems at the moment.

For the first one make sure you re-download the version 0.9.4 as the original one had a bug in it related to opening the configuration dialog on tracks.

For the second problem when the object is tracked it will have solid opacity during the tracking period, it should only fade out at each end, maybe it's this?

The problem was that the endpoint of the reference axis was set to be trackable, but since this is an abstract axis that is always defined in relation to the stationary arm it shouldn't be tracked. The inconsistent behavior comes from a conflict between the manipulation constraints system and the tracking.

I fixed it in the development version. It's also possible to fix it on your end by modifying the goniometer tool file manually:

<Kinovea program files>\DrawingTools\Custom\1 - Goniometer.xml

At line 45, remove trackable="true" (or replace it with trackable="false").

Interesting! thanks, I can reproduce the problem.

I re-released 0.9.4 with a fix for this as it was fairly critical.