The reconstruction result after rotating the detector in place is incorrect

[zeminLi66]

My projection method is to set COR and offDetector. The origin is first rotated 90 ° along the Y-axis, and then the origin and detector are rotated in place at the same angular velocity, as shown in the following figure。

The code is as follows

The projection image is like this

The result of the reconstruction is as follows

But if I set rotDetector all zero，geo.rotDetector = np.hstack(zeros_array, zeros_array, zeros_array))；There will be no problem with the reconstruction result.Where did I make a mistake？

[AnderBiguri]

The geoemtry you describe requires more parameter changes (as discussed in other posts) than the ones you are doing. I suggest not changing geo.COR as it will make everything more confusing and it's a redundant parameter. You need a combination of moving and rotating both the object and detector, each of them. So you will need offdetector, DSD , offorigin, a 3 valued angles and rotdetector, each set per projection.

The exact way to add them, I don't know. Work out the geometry to make it happen. It requires understanding the geometric parameters well.

[zeminLi66]

If I add an offOrigin, the object will default to rotating around the Z-axis. How can I make object rotate in place without adding a COR？Can adjusting angles (z1, y1, z2) be achieved?

[zeminLi66]

I set offOrigin to move the object along the y-axis, but if COR is not set, the object will still rotate along the z-axis (0,0,1). It seems that the rotation axis will not move with the object's movement, and only by setting COR can the rotation axis be moved.

[zeminLi66]

Manifesting as first translating and then rotating around the z-axis (0,0,1), rather than first rotating and then translating the distance off Origin

[AnderBiguri]

If you have an arbitrary system (forget about CT) in where you need to position 3 object in relationship to each other, you mathematically only need to set 2 positions in space, and set their rotations w.r.t. each other. In TIGRE, you can do that for the detector and the object (per projection) with the parameters I told you.

[zeminLi66]

I sincerely appreciate your invaluable feedback. I've fully grasped your point. Thank you very much, I will try it out tomorrow.