Hi,

I'm currently working on a project, where I want to use photogrammetry data to match my 3d scene to a photo.
I managed to import both the model and the cameras via alembic (fbx messed up camera rotations in a weird way).

But the model wasn't correctly oriented and scaled, when I experted the alembic file from Metashape, so in between, I used Cloudcompare to orient the model to the correct horizontal plane and Rhino to set the model's scale and orientation around the vertical axis. I added the resulting matrix4 xform from Rhino to the one from Cloudcompare, to get a single matrix4 xform, that I could use to set correct positions for the cameras inside Houdini.

The problem is, while Rhino and CloudCompare seem to interpret the xform matrix in the same way, Houdini doesn't. Since I didn't know of a way to apply this xform to a null node parent of the cameras directly, I used an attribute wrangle node and the cracktransform function to pick the matrix apart into detail attributes for translation, rotation and scale, but the values I get are all wrong.

The matrix is a txt file with values separated only by spaces and ‘start new line’(like the ones when you press Enter) characters, so I had to add commas when copying them to Houdini.

Anyone with any experience in these matters?

Cheers,
Ivan

If you want a Houdini way to extract the matrix, you can try using the “Extract Transform” OBJ node, as long as the topology is the same between the objects.

Thanks. While it's a kind of a workaround, this seems to work most of the time.
Took me a while to find out, that I had to simplify the models by reducing it to just around 100 points as well as remove all Object level transforms for the Extract Transform node to work correctly, but it seems to work for the moment now.