[INSTRM-1037] Store AG geometry in pfs_insdata repo Created: 10/Jul/20  Updated: 17/Dec/21  Resolved: 14/Sep/21

Status: Done
Project: Instrument control development
Component/s: pfs_instdata
Affects Version/s: None
Fix Version/s: None

Type: Task Priority: Major
Reporter: yuki.moritani Assignee: Yoshida, Hiroshige
Resolution: Done Votes: 0
Labels: EngRun
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified

Issue Links:
Relates
relates to INFRA-207 Create repo(s) for varying instrument... Open
relates to FIBERALLOC-29 Interface shuffle to appropriate PFS ... Open
Story Points: 2

 Description   

AG cameras position on PFI is needed for "shuffle" to find guide stars.  Following the discussions at ICS/PFI+MCS telecon on 3 and 10 July 2020, 4 corner of each sensor is proper.

 

Hassan, Naoyuki and I talked about the file architecture of this information a while ago, looking the current pfi.yaml and Cobra stuffs. To have commonality with Cobra, we'd like to propose:

  • List available AG cameras in pfi.yaml (basically all 6 cameras)
  • Store 4 corners in another .yaml file (1 file for all cameras, or 6 files per camera)

 

 Comments   
Comment by Martin Reinecke [ 02/Apr/21 ]

Since shuffle is required to output pixel coordinates for guide star candidates, the list of four corners is no longer sufficient. To determine this information, we also need to know the "origin" corner, the "+X" and "+Y" corner, and the camera resolution.

Comment by Martin Reinecke [ 02/Apr/21 ]

Alternatively, a per-camera matrix transforming between PFI and the respective pixel coordinate system shold also work well. Then the only additional required information would be the number of pixels in x and y direction.

Comment by yuki.moritani [ 05/Apr/21 ]

Yoshida, Hiroshige how do you treat PFI[mm]-AG[pixel] conversion for agActor (and other AG software)? What information do you use?

Comment by Yoshida, Hiroshige [ 05/Apr/21 ]

Currently, the AG subsystem implements naive transformation methods internally, using:

  • reference position of the detectors (center: x = y = 511.5 pix)
  • PFI coordinates of the reference position (r = 241.292 mm, PA = 0, 60, 120, ... deg)
  • rotation angle of the detectors about the reference position (theta = 0, 60, 120, ... deg)
  • detector pixel pitch (p = 13 um)

In general, ignoring distortion, I think we only need, for each AGC:

  • reference position of the detector (x, y) pix
  • PFI coordinates of the reference position (x, y) mm
  • rotation angle of the detector in PFI coordinates (theta,) deg or rad
  • detector pitch (x, y) um

In stead (or in addition), we could have transformation matrices for each AGC for each direction.

 

Comment by yuki.moritani [ 05/May/21 ]

Yoshida-san and I talked about this topic. We think it will be nice to have a module to query the PFI position for a AG cameras, and vice versa. In this way, MPA can get PFI coordinates of 4 corners of individual AG cameras, and NAOJ can can get PFI coordinates of observed centroids.. Yoshida-san already has something to do so, so I think his routine should be a good start.

Comment by hassan [ 23/Jul/21 ]

Martin Reinecke mentioned during the ICS/PFI telecon today that he needs this in order to provide shuffle-related deliveries.

Comment by yuki.moritani [ 14/Sep/21 ]

The function was implemented, and Martin is now using this function, so I close this ticket.

What we may need to modify in the future is to treat the fact that AG pixel coordinates includes overscan region.

Generated at Sat Feb 10 16:31:14 JST 2024 using Jira 8.3.4#803005-sha1:1f96e09b3c60279a408a2ae47be3c745f571388b.