blocks on cylindrical geometry still unexpected geometry

[KrisThielemans]

Looks like we don't correctly centre the "bucket" in transaxial direction when there are gaps between blocks. Below are examples wtih 2 blocks/bucket and 1 block/bucket. with the blocks_spacing 20% larger than the block (generated with an improved plot_scanner_LORs.py, now even better on #1391). Of course, a gap between trans-blocks if there's only 1 block per trans-bucket makes no sense. We could use this to "shift" blocks, but I can't see why anyone would want to have such a PET scanner, so I don't think we should do this.

I think we should just centre the bucket.

in axial direction, it looks fine

@danieldeidda @markus-jehl @robbietuk

[KrisThielemans]

When using the commented code for start_x as opposed to the csi stuff at https://github.com/UCL/STIR/blob/0da0f080dbc76c93ef7f500e7d626d84c7a8f67b/src/buildblock/GeometryBlocksOnCylindrical.cxx#L94-L100 is what I expect.

@danieldeidda you made this change in 42bf1e98e1cfb3c7fae131fb8590b439ff07068c. Can you remember why?

[robbietuk]

Yes I agree. The second images are correct. There should a test to ensure this is the case...

[KrisThielemans]

@danieldeidda could you have a look at this? @markus-jehl will make a PR

[danieldeidda]

When using the commented code for start_x as opposed to the csi stuff at

https://github.com/UCL/STIR/blob/0da0f080dbc76c93ef7f500e7d626d84c7a8f67b/src/buildblock/GeometryBlocksOnCylindrical.cxx#L94-L100

is what I expect. @danieldeidda you made this change in 42bf1e9. Can you remember why?

looking back at my drawings I think this was when we had the configuration with first detector at 0 degrees but the calculation of the starting point doesn't seem to make sense for this configuration. Possibly I had used this because of the failing tests, I can't remember very well though.