Triple Tracking

#1

I have a moving object that is obstructed on both ends, not in a static on constant way, but moving.

So I need to track the object, the border where the left obstruction covers the object, and the right obstruction that covers the object on the right.

The boundaries between the object on the left and right are shared with the tracking of the object.

Is there a way to not fully independently track all 3 objects, given that they share tracking of the borders? That is the splines nodes are used for the object and for the obstructing objects? I’d hope to not have to carefully place the tracking nodes redundantly, since some are shared.

I can drag the nodes to the other end of the boundary, and keep the shared nodes ON the boundary there, but that is very cumbersome. and trying to track after doing that a few times, it does not go well.

Thanks.

#2

If I understand your description, it might be a good shot for “join/attach layer” tool. This allows you to join the borders of different shapes, while the general motion can still be linked to their own surface tracks.

This is an old video but shows how the tool works: https://vimeo.com/54257297
Hope this helps.

#3

I don’t think that addresses the issue. Maybe to be more clear, say I have a square on the left, 1 on the right, and a tube inserted between them. They all move. Because the squares overlap the tube, I need to track them, importantly the border between the squares and the tube.

After that tracking of the squares, I need to track the tube. The left and right borders of the tube have already been tracked when I tracked the 2 squares. I would like to not have to track those 2 end borders again, as adjustments are always needed.

So for now those borders are tracked multiple times (when tracking the square and when tracking the tube). Can I use the rotoscoping nodes used when tracking the squares at the boundaries between the squares and tube when I rotoscope the tube?

I know when tracking the tube I am getting tracking data, and not creating, say, a matte, and so I don’t need to be as exact as when tracking the squares, but if i can use what I did with the squares it would help.

Hope that’s not more confusing. :wink:

Thanks.

#4

You can track the right square, the left square, and the tube over the top. They will all have different surface tools so they will have different tracking data, but it will not be tracked to the tube. The layers would look like:

  1. Tube
  2. One square
  3. Other square

If you want to get the tracking data you can save the corner pins and combine the tracking data for the corner pins from the left side of the square and the left side of the tube, and the right side of the other square and the right side of the tube, but you’ll have to manually copy and paste those keyframes or pickwhip them, depending on your host app.

Ross is right too that you can connect the pole roto to the touching edges of the squares, but that will just give you moving matte data, not transform tracking data like corner pins. Though you can convert that point data to a track with a script if you have to, but that is starting to get complicated.

What’s the final goal? Maybe we can give you a better suggestion?

Cheers,
Mary

#5

Thanks. Will try to understand all you said.

I’m replacing the tube with another same-sized tube. So I use the original tube’s tracking data for the new tube. And I need to creates mattes of the squares to lay over the top at the end.