Hey there, I am new to Mocha and I a bit of a dilemma… I need to place a fake “projection” on a very textured building facade (old church) and the camera is panning up and down form an extreme angle.
I was able to use the Corner Pin method to keep the image in perspective BUT it looked super fake as the “projection” wasn’t warped to the building facade. As such I have been trying to learn POWERMESH but as it is so new I am only seeing the most basic examples where a logo or design is put onto an object that never leaves the frame or changes perspective. Any help / assistance on how to do this would be greatly appreciated!
Here is a link to the source footage (not my choosing nor did I film it) SOURCE FOOTAGE
Hey Mary!
Thanks for your reply…
So I am using this with After Effects.
and I did try a PowerMesh workflow, however I couldn’t’ get it to work the way that I had hoped and it seemed to me that due to the extreme moving perspective change, PowerMesh would also maybe need to have a corner pin type application involved as well.
Firstly I am very new to Mocha, but on top of that due to PowerMesh’s newness there seems to be very few tutorials dealing with more complex camera movements… basically all I am seeing is pretty straight forward limited movement use cases where the subject is in frame at all times and there are not extreme perspective cases…
I am not familiar with true 3d texture projection… is that included with Mocha or is that another Boris app?
No, you’d have to do 3D texture projection in a piece of 3D software. PowerMesh is still a 2D solution at the end of the day, so while it can do objects that are curved, and objects that are organic, trying to warp around an extreme 3D perspective shift in z space might require an actual 3D model of the object you’re trying to replace. And then you can use the camera to do a texture projection onto that model. Maya, Cinema 4D, and things like Nuke, Fusion, or Blender can all do 3D projections.
You could also try this with an alembic workflow with Fusion or Nuke, pushing and pulling points in z space once you got them out of PowerMesh, but I haven’t experimented with that technique yet so it’s purely theoretical at this point. It’s a big question as to whether or not that would actually match the camera motion properly because of the z-space changes.
Barring that, you could try splitting this up into pieces to try to get a more accurate warp. So instead of warping the whole wall and the arch together you could try using power mesh for the wall, another PowerMesh for the bottom of the arch, and another PowerMeshsh for the top of the arch.
Basically at this point if you need a really 3D looking move you’re either going to have to use real 3D or you’re going to have to fake it as best you can with multiple layers.
Thanks for this! I don’t think I am ready for diving into 3d Modeling yet but the splitting the area up into different fields to fake the projection on those noticeable levels may work!