Photoscan XML structure explained

[baltena]

Dear,

The help and the forum is very limited concerning information about the file structure of the Photoscan XML structure. Nevertheless, I want to use such files to import internal and external parameters from other software solutions, and let it proces by Photoscan.

Most information is self explainable in the XML, nonetheless some essentials are missing such as:

In the camera tab a 4x4 matrix is given, which is composed of a 3x3 rotational matrix, a 3x1 translational matrix and 1x3 zero column followed by a 1 entity. However, from which coordinate system is this originating?

In the ground truth tab, coordinates are given, though these seem to be initial values and no estimated.

Furthermore, a tab is given with "transform" being a 4*4 matrix again, what kind of purpose does this entity have?

Kind regards, Bas

[James]

Maybe some useful info here: http://www.agisoft.ru/forum/index.php?topic=2351.0

The individual 4x4 camera matrices exist in arbitrary coordinate space, only giving relative orientations of cameras to each other.

The transform matrix is introduced when you establish your own coordinates and so it describes the transform between the arbitrary and real world systems.

When you export to bundler format then this transform is applied to all cameras so i found that the bundler format was better for me, hence the link above where i worked out how the two formats relate to each other.

[baltena]

Thanks, that's the info I needed! Unbelievable, how just a couple of lines have the potential to increase the usability of the help-function considerably. ...if only