I have a still image of a view looking up at a complicated building (no ground in shot). There are lots of square features so I know where the right angles are but I have absolutely no information on any distances/dimensions at all and no info on the camera (focal length etc…).
I can get close by eyeballing it in 3D Max, tweaking cam position and focal length and proxy geometry but I’m sure there is a way to do this more automatically in Matchmover (which now ships with Max).
In Matchmover I created 4 track points on my image in 2D mode, 1 on a vertical corner as the origin, 1 directly above as Z, and 2 either side of the origin using the mortar lines to keep them level (all are as far away from each other as possible to increase track accuracy). I then make a new coordinate system and snap the origin, then Z and finally the X,Y nodes to the corresponding track points.
The last node I snap whichever one it happens to be, wont snap like the others. It just pings back.
When I “solve for camera” it complains that it needs at least 2 images with 7 points in common or 4 points with known 3D coordinates.
There is visible lens distortion and I think this needs correcting somehow but have no idea how, other than the distortion parameter for the camera but this feels like stabbing in the dark as I have no way to know what distortion to apply.
Thanks in advance for any advice!
3D animator and VFX specialist