Closed tritemio closed 7 years ago
If you use a 1D array of I64, you can easily account for a 2D detector matrix by using the lowest word for X, highest for Y. If you need X, Y and Z, the same principle can apply (e.g. 20 bits per dimension). The only open question with this approach is how to describe the encoding used.
No need to encode it in a int64. We can just use a 2-column array, first column for X, second for Y. But you are talking about detector ID, that is already in the specs (it can be specified better).
I was thinking a general mapping (a single 2D array for each detector) mapping IDs to a position. In this way we don't necessarily need to use 2-tuple detector IDs for 2D arrays.
Addressed in #33 with the optional field /setup/detectors/module
.
When using SPAD (or other) arrays would be useful to specify the number of array modules used in the measurement. Currently we only save num_pixels which is the total number of pixels (see specs).
Also, it would be useful to save the "position" on the chip for each pixel. For 1-D SPAD array this can be a 1-D numerical array containing detector ID for each position. For 2-D arrays, it can be a 2-D numerical array.