Skip to content

Capture option to wait for low guider drift.

Hy Murveit requested to merge murveit/kstars:guider-fix6 into master

Overview

  • Can optionally delay the start of capturing an image until the guider has low RMS error.
  • Also guider keeps better track of guide star.
  • Analyze can plot this better rms.

Details
This main emphasis of this MR is to improve the guiding issue as shown in this Analyze screenshot (note the spikes in the red guide RMS line, after each focus session): https://photos.app.goo.gl/rmajK4nUKtf5CXDn7

What's happening is that guiding can be suspended during focus, and without guiding (and with the focuser pumping in and out) the guide star can wander away from its target position during the couple minutes of an autofocus session. When capture restarts, it can take several guiding iterations to correct this guiding error, during which the image can be significantly offset from the desired position. Further, if the target is sufficiently offset, or if the guide star wandered out of the reticle, the guider might fail to even find the guide star, and can be restarted by the scheduler. A new guide star will be found, possibly with a significant positional offset from the origin image coordinates. This guiding fail happens shortly after the red and green focus sessions in the screenshot.

This MR corrects the above with several changes. You can see the improvement in this screenshot: https://photos.app.goo.gl/eZP2HFHxoUwo39qU7 Note that there are no guide RMS spikes after the focus sessions.

The main change is a new capture-tab checkbox which causes the capture process to wait for a small guider offset before starting up a new image capture. https://photos.app.goo.gl/aSNbpcfHD7cujeCc7 This might be, for example, 2 arc-seconds as in the screenshot. Thus, if after focus is completed and the guider restarts, should, for example, the guider offset be 15 arc-seconds, and requires 4 guider iterations to get back to a < 2a-s offset, capture would wait until those iterations are complete.

Other changes allow the SEP Multistar guider to continue to track the guide star, even when it has large offsets, and to allow the guider to return back to the target position once it is re-enabled. In general, I've relaxed a few of the conservative restraints on the guider, now that SEP MultiStar has been successful in the field, in order for it to be able to overcome the large guiding drifts described above.

In addition, Analyze now has a new graph. In the capture section there's a capture/guide RMS graph, which only takes into account guider drift when an image is being captured -- so it more properly reflects the "important RMS" that will be baked in to the captured image.

Changes

capture.cpp/h
sequencejob.cpp/h
ekos.h

  • capture can wait to begin capturing image until guider settles

analyze.cpp/h

  • Capture RMS graph
  • keyboard scroll with arrow keys move slower

analyze.ui

  • add new captureRMS graph controls

capture.ui

  • add new ui element for capture to wait until guider settles

guidestars.cpp

  • increase number of detected stars searched to be guidestar
  • don't honor tracking box in SEP multistar
  • refactor some logging

internalguider.cpp

  • don't stop sending tracking pulses with 3 bad rms iterations (just wait for guider to be aborted)
  • abort after 10 bad frames (up from 5)

kstars.kcfg

  • store variables for analyze graph and new capture "wait for guider to settle" control.

fitssepdetector.cpp
skybackground.h

  • make SkyBackground::SNR const

Merge request reports