Batch measure tilt behavior
Stacking program, astrometric (plate) solver, fits file viewer
Brought to you by:
han59
Menu
▾
▴
v06.29 Batch measure tilt runs through the images. 'Completed, all files processed.' and no error (that I noticed?) but no TILT written to header. I've had this on a couple different sets and I'm thinking maybe it has to do with the number of stars.
On a recent set of T CrB images with this issue I opened one in ASTAP and went Image Inspection on it. Triangle seems to work as expected where I see the triangle with vertices and center values and the bottom has:
Median HFD = x.x (x.x")Tilt[HFD]=x.xx (xx% ssss) Stars=xx Off-axis aberration[HFD]=x.xx
Switching to Octagon there's no octagon, just boxed, numbered stars and no Tilt at the bottom, viz.
Median HFD = x.x (x.x") Stars=xx Off-axis aberration[HFD]=x.xx
In both cases Stars=27. On an earlier set dealing with a cropped ROI there were fewer stars and Triangle had no Tilt results as well. No keyword with either Triangle or Octagon as last setting.
Since I'm posting this I'm wondering if I can't request a batch process to write the median HFD to the header.
Thanks,
Jesse
Hi Jess,
I tested the batch tilt measurement and it worked here. You get in the header a new keyword like this:
TILT = 0.57 / Delta HFD between worst and best corner. SNR>30
Can you attached to your reply one image where it does not work?
About your second question, I could add the median hfd to the measure tilt option and write it to the header. As a separate keyword , something like HFD =?
cs, Han
Last edit: han.k 3 days ago
I attached one of the T CrB images. It displays a TILT value for triangle but not octagon and batch doesn't write anything to the header. I think it might be related to too few Stars
Thanks,
HFD = is fine. I did a quick Google search and couldn't find any one else writing something similar. I occasionally use PinPoint and they write a FWHM after a plate solve. I often plot the FWHM (HFD) to see how it varies by itself over time or to see if it correlates with focus changes if that's enabled. Thanks,
The number of stars is too small. There are nine areas and I assume one area mid & right is too empty. This is quickly tested with the F4 button.
I have no problem to add HFD keyword, but is it useful? You could add the images to the tab lights. Hit the button analyse (quick unselected) and you get all HFD values. Then you could select all (ctrl+A) and copy & paste all info including HFD to a spreadsheet. If you have an other tool which could read keyword HFD, then I will add it.
cs, Han
Thanks for getting back. Since there wasn't an explicit too-few-stars-message I figured I'd make sure. I'll try the F4 thing as well. I almost didn't bother to post but then I was wondering with the T CrB image if triangle was able to measure a TILT then why wasn't that value being written to the header.
I also wanted to ask about HFD. It's convenient having the metadata in one place in the FITS the header but it's not like I can't read a spreadsheet as well.
Thanks again,
Jesse
Last edit: Jesse Goldbaum 3 days ago
I have modified the Window development version.
1) Added the keyword HFD.
2) Added a message not enough stars for tilt
https://www.hnsky.org/astap_setup.exe
Other operating system will follow soon. I'm making some other changes.
cs, Han
You'll have to forgive my limited ASTAP skills if I'm missing something here. I ran a non-batch plate solve using 07.13 and didn't see an HFD keyword in the header (attached). That being said, I noticed after an Analyse that a Median FWHM was reported by ASTAP. Since MaxIm/PinPoint writes a FWHM keyword then that works better than HFD for me (I wouldn't have to edit my code :)
Hi Jesse,
In made a coding error. The HFD report was not always reported during analyse. I have corrected that and it should work for your star poor images. For both F4 analyse and batch tilt measurement.
I have also added the median FWHM.
https://www.hnsky.org/astap_setup.exe
It is currently not executed for astrometric solving! I have only the mean HFD value after solving. Not the median since it will take more processing time. I can look tomorrow if that could be added after solving.
cs
What I tried to say, I can look if the mean HFD can be reported after solving.
Sorry to be a pain but I'm still not seeing any HFD or FWHM in the FITS header after a solve (attached). I think all I'm really suggesting is to add either the HFD or FWHM keyword to the Batch solve images -> e.g. Option "Add Lim_MAGN, SQM, CENTALT, AIRMASS, FWHM".
I guess I prefer FWHM since my software already checks for that though HFD is trivial to add. I'm agnostic about median vs mean as I'm comparing relative changes over time. That being said, if outliers are a regular thing throwing off the mean then maybe a median will be more consistent.
Thanks,
I've added the unsolved version of the test image I'm using if that'll help. (I can't figure how to add more than one attachment/post). -Jesse
Hi Jesse,
HFD and FWHM are currently only added by batch mode to add tilt value and the the tilt measurement activated by F4. It is NOT added after solving.
To add it after solving, I have already in the code the mean/average HFD values after solving but it is not stored in the header.
I need to finish some complicated changes in the code first and then I will add the mean HFD after solving. The median HFD is not calculated after solving since it would be waste of processing power. The mean is much easier to calculate.
I have added it to the solve function. The mean hfd is maybe 20% higher then the median hfd . Therefore I have used the new keyword HFD_MEAN. So you can compare it with HFD from the analyse function. Tell me if this works for you.
https://www.hnsky.org/astap_setup.exe
cs, Han
Thanks for your prompt attention on this and my confusion regarding solve vs tilt. I think it should be fine. I'm on the way out for a few hours so wont have a chance to test as soon as I'd like. Thanks again,
In short I can use the current version. Thanks,
In long my requests are somewhat simpler.
For batch plate solving I'd like to see a FWHM keyword. If that's too much then an HFD keyword is fine as the conversion is simple (assuming a Gaussian profile). Don't need both. The keyword comment section would include the method used for the estimate (e.g. average, median, mode... ;) and the number of samples used (aka number of stars). If for some reason ASTAP can't calculate a value then write a bogus value (e.g. -999) and indicate the reason in the comment (e.g. /Error! Too few stars).
For batch measure tilt a similar idea. If ASTAP can't measure a valid tilt then as above write a bogus value and note why comment field. Optionally, the comment for TILT might include the geometry used, e.g. /Delta HFD between worst and best octagonal corner. SNR > 30. I'm not clear which geometry is currently used going batch. E.g. do you start octagon and drop to triangle if not enough stars or do you go with triangle regardless, or do you go with whatever was last selected in the gui?
Thanks,
Jesse
P.S. Yet optionally more...
FWHM matches PinPoint and what my code currently looks for. I'm interested in this value as a quality metric for the image from both a relative and absolute standpoint. As a relative value the method used for the estimate is less important than from an absolute standpoint. If the mean is typically higher than the median (a right skew) then the median would be preferred, but it's up to you. Since PinPoint writes the mean FWHM then that makes comparison easier. I'd want to compare them as if they don't correlate than there's question whether FWHM is even a reasonable/meaningful quality metric in the first place.
Finally, the bogus value at least tells me that ASTAP did what I asked. If I ask it to write a TILT and there's no TILT then... was it something I ate ;)