Some question About CL demo (demo 05 Laminography with TV )

[AmazingYi]

Thanks for your excellent work !

I have some small questions about demo 5 Laminography with TV .

First , about the dataset

The lego_30 _tilt dataset has a config that shows the source to obj is 392 ... on above . But the demo shows that SOD is 295 . Is it
a wrong in the org dateset confg ?

And I also has a question about how to know the COR of this dataset .

In the demo the COR is 6 x pixel_size . which means it is caculated by images of detector . So I was curious that how to caculate COR because of that the rotated plane was tilted 30 ° any mark point int this plane on the detector should be a Elliptical trajectory ,rather than a Circular trajectory . So I'm little confused that How to know the COR.

[gfardell]

Hi @AmazingYi, the paper that goes with the dataset is here, we will make sure it's added to the notebook introduction. I think it'll answer all your questions!

The Nikon system this was collected with couldn't natively perform a CT scan with the axis tilted. As you can see form the diagram the sample is mounted on a foam pillar, this means the SOD reported is that of the system, then you can calculate the more accurate position along that axis as you know the tilt and the height of the tower. A side note is that an wrong SOD won't give you a bad reconstruction, it'll just scale the size of your object.

I think in this case the centre of rotation was found manually by adjusting the offset until the reconstruction looked good. You can see an example of doing this with CIL at the end of this jupyter notebook how to adjust and reconstruct to find it yourself. It'll still work for Laminography.

We have an alignment tool to fit tilt and centre of rotation, currently it only works on parallel CT data, but hopefully we'll have the cone beam version tested in v26.1.