I'm attaching a sample still of what I'm trying to accomplish. Please take a look:
The idea is to create a 3D layer (red grid) that will have a text displayed behind a person's back in a perspective view against two points on a person's shoulders. I know that I can track those points with Mocha, but what I can't figure out is how to apply that data to the movement of the text layer in a 3D world.
Any suggestions will be appreciated.
I know that I can track those points with Mocha, but what I can't figure out is how to apply that data to the movement of the text layer in a 3D world.
Well, there is no 3D data. It's all just 2D. Even if you choose to export transform keys instead of corner pins and apply it to a 3D layer, it's just pseudo 3D. mocha has no concept of depth.
OK, thanks. But what about 3D trackers, like Boujou for instance. Can it do what I want? (I hope they allow to download a trial version though)
In theory yes, but not with such tight shots. You would have to use a wider frame and crop it in post. There is not enough background visible for a stable 3D solution in your example. And of course having distinct features/ markers in the BG helps.