I took a horizontal video from my Smartphone Huawei P30, 1920*1080@60 fps .mp4. Kinovea is 0.9.3 beta
edit: I repeated with a horizontal 1920*1080@50 fps mp4-clip from my Gopro Hero8 - same results..
Then i made a trackpoint on a point that is moving vertically strong and horizontally weak. The tracking is clean.
I did'nt touch the coordinate system in this little test.
So i get in linear kinematics a vertical displacement from -101 to +12 pixels, horizontally -2 to +24 pixels. Seems to be OK.
Then i added on the same frame a vertical line and scaled it to 50 mm.
After doing that, i get completely wrong values in linear kinematics. While Acceleration seems to be correct, i get now
a vertical displacement from -1 to +15 mm and a horizontal displacement from -7 to 58 mm - so the orientation has obiously switched.
After that, i made the coordinate system visible. The Zero point moved from the middle of the image to the starting point of the scaled line - OK. I can see the orientation of the coordinate system is'nt 0/180° anymore but a bit tilted. I cannot recognize, if this tilt is 5° or 95° as i see no difference between x- and y-axis.
Regarding the wrong values in linear kinematics, i suppose it's rather 95°.
The light tilt of the coordinate system does'nt seem to follow the movement direction of the tracked point and not the direction of the drawed line wich was strictly vertically, it seems rather to be randomly.
Again: I did'nt touch the coordinate system and i did'nt track it.
I repeated this little test by drawing the calibrated line not vertically like in the first test but now horizontally. In this case, the plotted horizontally and vertically curves in linear kinematics seem to be more plausible. So obviously, the direction of a calibrated line influences the rotation of the coordinate system but not exactly 90°. I think, it should not do this?
Hope, this short story can help.
Kind regards, Michael
edit 2: Just did the same procedure in Kinovea 08.27: Here, defining a calibrated line, does'nt affect the coordinate system, independent of the orientation of the line. Not the zero point and not the rotation. In 8.27, the graphs shown in linear kinematics seem to be pretty ok. So the effect came oblivioulsly between 8.27 and 9.93
When i additionally try to track the coordinate system, Kinovea 8.27 crashes when calling "linear Kinematics", so i can't say if this will affect the graphs.
Crash report says:
"unknown exception in the software"