Hi everyone, the graphs of the motion tracking show the information of the speed of each tracker, in the X axis we can see the frames, in Y axis the speed, I want to know the units of speed used (m/s) or blender units per frame, thanks
the graph is based off a 2D image (so no depth)… and the speed is measured in pixels… so you would have to define the DPI of the image for it to be converted into m/s
To get the sped in m/s, you need to convert those frame-by-frame pixel locations/offsets into “world space” locations/offsets. There is a reasonably good tutorial for how to do this in the Blender Cloud and some on YouTube. Once you’ve done that, you can scale the solved scene to ensure that a fixed size object in the shot is the right size (say a 1x1m box is 1x1 blender units) and then read the velocities from empties parented to the moving mocap markers.
If it’s a fixed camera shot, there is a fixed-size object at the same depth (give or take a small margin) in the view, and not a huge amount of distortion due to the camera lense - you could get the pixel size of that fixed-size object and use that to calculate the “world distance per pixel” at that area of the shot. Multiply that by the pixel offsets to get the approximate velocity of the marker.
Thanks for the quick response, then the speed is in pixels per frame, but I don’t understand very well about the conversion to m/s, especially this part “frame-by-frame pixel locations/offsets into “world space” locations/offsets.” Any link of a video or a tutorial will be very appreciated, thanks.
each plot on that graph, is the difference in pixels location on a per frame basis. as in when it spikes up on the red… the track point has moved in the positive x direction on the 2d Image.
In regards to videos or tutorials
There is a reasonably good tutorial for how to do this in the Blender Cloud and some on YouTube. Once you’ve done that, you can scale the solved scene to ensure that a fixed size object in the shot is the right size (say a 1x1m box is 1x1 blender units) and then read the velocities from empties parented to the moving mocap markers.