Skip to main content

Modifications to VPhot

3 posts / 0 new
Last post
MZK
MZK's picture
Modifications to VPhot

All VPhot Users:

Over the past two weeks and especially over the last two days, Geir, George and I have made some changes to VPhot that we hope will significantly improve the queue processing speed. I am not going to go into excruciating detail but give you a summary of what we changed and mention how the changes may impact your images and photometry.

Change 1: Two queue processing operations that support the VSS (Variable Star Search) function in the VPhot webapp (where you conduct your photometry) were deactivated (commented out in the code). These operations added significant delays to queue processing time with perhaps only limited use by members. In other words, it appears that very few members use the VSS tool?

Impact 1: The VSS function no longer works. We hope to move the two queue processing operations into the VPhot webapp in the future but cannot tell you when. I would appreciate knowing who uses the VSS function, not just once or twice but regularly. 

Change 2. One of these two operations also wrote the FWHM value to the database. This parameter is used in the Time Series (which most members use) function. The FWHM values continues to be calculated and saved for images that are plate-solved by VPhot but not for images that are pre-plate-solved and thus do not need additional plate-solving in VPhot. The FWHM keyword is provided in the Fits keyword provided by some plate-solving software (e.g., Maxim). If you make sure it is added this can be used to easily solve this issue. 

Impact 2:   For all those who pre-plate solve their images (I do and have been pushing hard for others to do the same), you can continue to use Time Series but you need to THINK about the procedure rather than just press the button. Since the FWHM of each image is not available, the use of the "automatic" aperture selection skips to the "fixed" aperture selection while the time series runs. Therefore, the value in the fixed aperture box needs to be set carefully. The default of 5 pixels works but that may not be the best choice for your images! So you should determine what the best aperture (radius) in pixels is for your system. I use one value for virtually all my images. My system parameters are stable (e.g., I bin 2x2 on my scope) and never (rarely) changed. I suspect the same is true for your system. You can confirm the FWHM after you run the Time Series and look at the FWHM graph shown on the time series results page. This info can be used to determine what fixed aperture value works best for your system. If needed you can change the fixed value on the time series execute page and run the series again to get the best photometry. You can leave the default at 5 pixels and it will work for most images but I have noticed that the photometry of fainter targets are impacted in a small but measurable way. So be careful!

Conclusions: We have made these changes to significantly improve queue processing speed. I feel that these changes do that. George is creating some new metrics to prove whether I am correct (or not). It does impact some users. In fact, one user has just reported the loss of VSS functionality while I'm writing this post!

We are able to revert to the original code at any time but need to be convinced that this is worth it as it relates to the improvement in processing speed. We also are planning to move these functions into the webapp.

I have now told you what has been done. Now I'll have to deal with the fallout from this modification. I know everyone will not be happy. I'm strong but please understand what we are trying to do in terms of throughput! I'm listening.  

Ken

 

 

 

MZK
MZK's picture
Repair to Time Series Auto-Aperture

All VPhot Users:

Issue number 2 noted above has been fixed. Geir spent a couple of nights writing code to calculate the FWHM from comp stars in each image. The auto aperture function in the time series is back to its normal functionality. Thanks, Geir.

Queue processing speed appears to be significantly improved. Most of you have been plate-solving your images before uploading, Thank you. These images take about 4-10 seconds to complete processing. The duration depends on the size of your image. Images that approach 30MB take the longest. The processing includes writing images to a temp directory, changing the fits header, and sending the image to your personal image list.

Queue processing for non plate solved images may take about the same time unless Pinpoint has problems. Typical problems may be related to short images with few stars or deep images with many crowded stars. So the problems cover a wide range of variability. VPhot tries to plate solve using the UCAC4, GSC and Tycho2 catalogs, in that order. Sometimes, an image may take 20-30 s to solve but sometimes it may fail all three catalogs (~120 s)? If it still fails, you might try to solve one of your images from your own computer with astrometry.net. If you then upload this solved image to VPhot, it can be used to solve other images in the webapp using Update WCS.

I hope most of you are having success with your images and are happy with the speed improvement. I watch the daily log of images and occasionally send an email to those whose images have problems.

It is nice to see an empty queue when all your images have flown through the queue in a few seconds. Keep uploading them to VPhot! The VPhot team is here to help.

Ken

 

 

 

spp
spp's picture
Update WCS

Ken,

Many thanks to the VPhot team for its dedicated work in addressing these issues and keeping this great tool working for the rest of us.   

I was unable to locate documentation (in Help or the VPhot manual) for the Update WCS tool.  Perhaps a short "cheat sheet" would be helpful.

Phil

Log in to post comments
AAVSO 49 Bay State Rd. Cambridge, MA 02138 aavso@aavso.org 617-354-0484