PlateSolve2 incorrect initialization

Last night I installed v2.4.2.5 to test PlateSolve2. I had previously installed the stand-alone version of PlateSolve2, so I was familiar with it. However, running it inside of SGP, PlateSolve2 failed to solve frame and focus subs 100% of the time. Doing some testing, I discovered that it was being invoked with an incorrect plate size. Instead of showing 25’ x 19’ (the correct size of my QSI683’s FOV) it was showing 34’ x 34’.

Setting up PlateSolve2:

Do frame and focus and attempt to solve:

Solve fails even though the hint was highly accurate:

I kept the PlateSolve2 window open and manually edited the FOV value to the correct value of 25 x 19 and then clicked the ‘Plate Match’ button. It solved almost instantly:

I don’t know where the 34’ x 34’ entry is coming from since the setup specifies 25’ x 19’ but the mismatch with the actual FOV of the camera is causing PlateSolve2 to fail.

BTW – as a test, I also performed the “second solve” routine without changing the 34x34 and it failed the second solve attempt, as well.

cm

This is working for most folks so your issue is likely data related (possibly regional if you are not in the US). It is also most certainly related to image size. Image size is not a setting… it is a property that belongs to each solve. SGPro extracts the image, height and scale from the image. It seems like the images you are using might have something unusual stored here. Can’t help much unless you make the logs and image available via dropbox or similar.

Ken:

I am located in Austin and all the images being tested were either frame and focus images or images taken as part of a “centering” task. The failure of PlateSolve2 was solely related to those types of frames.

If I load an image previously captured by SGP as part of a sequence and saved to disk, PlateSolve2 solves those frames reliably when I reload them and do a solve. According to the PlateSolve2 Help file, it reads info from the FITS header and uses that if it can. This seems to imply the FITS headers being built as part of a target sequence and saved to disk are formatted differently or contain different info than the FITS headers being created for Frame/Focus images or Centering images.

If SGP is calculating the plate size as the displayed size of 34’ x 34’, then there is problem since my system has a pixel scale of 0.451" / arcsecond per pixel and the chip is 3326 x 2504 (KAF-8300). When PlateSolve2 successfully solves an image from a target sequence, PlateSolve2 shows the expected 25’ x 19’ plate size as well as the 0.451 plate scale:

In the FITS header of an image taken as part of a sequence and saved to disk, it appears that PlateSlove2 can retrieve the image size (3326 x 2504) and the pixel scale (0.451) and then calculate the correct image size – ignoring the 34’ x 34’ setting SGP is generating. However, when it is a Frame and Focus image, PlateSolve2 is apparently being forced to use the 34 x 34 value supplied and that is what is causing the failure.

All these images are being taken by a QSI683wsg camera so there is nothing unusual about the way they are formatted, stored, etc.

I will be back at my observatory Saturday night and I will rerun the tests and capture the SGP log file. Since these are Frame and Focus images, you want me to right click and do a “Save As Mono FITS…” and send those files along with the SGP log?

As an aside, I was using Astrometry.net as the blind failover and Astrometry.net was solving these frame and focus images in 40 seconds or less.

I did capture a snippet from the SGP log file right after one of the fails:
[8/13/2015 10:54:58 PM] [DEBUG] [Main Thread] Tried to display solved stars, but list is null
[8/13/2015 10:55:18 PM] [DEBUG] [Main Thread] PopulateDataModel: Transferring view to the data model…
[8/13/2015 10:55:18 PM] [DEBUG] [Main Thread] Performing serialize…
[8/13/2015 10:55:27 PM] [DEBUG] [Main Thread] Tried to display solved stars, but list is null
[8/13/2015 10:56:50 PM] [DEBUG] [Image Plate Solve Thread] SafeParseDouble: Input string was not in a correct format.
[8/13/2015 10:56:50 PM] [DEBUG] [Image Plate Solve Thread] SafeParseDouble: Input string was not in a correct format.
[8/13/2015 10:56:50 PM] [DEBUG] [Image Plate Solve Thread] SafeParseDouble: Input string was not in a correct format.
[8/13/2015 10:56:50 PM] [DEBUG] [Image Plate Solve Thread] SafeParseDouble: Input string was not in a correct format.
[8/13/2015 10:56:50 PM] [DEBUG] [Image Plate Solve Thread] SafeParseDouble: Input string was not in a correct format.
[8/13/2015 10:56:50 PM] [DEBUG] [Image Plate Solve Thread] SafeParseDouble: Input string was not in a correct format.
[8/13/2015 10:56:50 PM] [DEBUG] [Image Plate Solve Thread] SafeParseDouble: Input string was not in a correct format.
[8/13/2015 10:56:50 PM] [DEBUG] [Image Plate Solve Thread] SafeParseDouble: Input string was not in a correct format.
[8/13/2015 10:56:50 PM] [DEBUG] [Image Plate Solve Thread] SafeParseDouble: Input string was not in a correct format.
[8/13/2015 10:56:50 PM] [DEBUG] [Image Plate Solve Thread] SafeParseDouble: Input string was not in a correct format.
[8/13/2015 10:56:50 PM] [DEBUG] [Image Plate Solve Thread] SafeParseDouble: Input string was not in a correct format.
[8/13/2015 10:56:50 PM] [DEBUG] [Image Plate Solve Thread] SafeParseDouble: Input string was not in a correct format.
[8/13/2015 10:56:50 PM] [DEBUG] [Image Plate Solve Thread] SafeParseDouble: Input string was not in a correct format.
[8/13/2015 10:56:50 PM] [DEBUG] [Image Plate Solve Thread] SafeParseDouble: Input string was not in a correct format.
[8/13/2015 10:56:50 PM] [DEBUG] [Image Plate Solve Thread] SafeParseDouble: Input string was not in a correct format.
[8/13/2015 10:56:50 PM] [DEBUG] [Image Plate Solve Thread] SafeParseDouble: Input string was not in a correct format.
[8/13/2015 10:56:50 PM] [DEBUG] [Image Plate Solve Thread] SafeParseDouble: Input string was not in a correct format.
[8/13/2015 10:56:50 PM] [DEBUG] [Image Plate Solve Thread] SafeParseDouble: Input string was not in a correct format.
[8/13/2015 10:56:50 PM] [DEBUG] [Image Plate Solve Thread] SafeParseDouble: Input string was not in a correct format.
[8/13/2015 10:56:50 PM] [DEBUG] [Image Plate Solve Thread] FitsFileHeaderData: Angle - 124.349756022391
[8/13/2015 10:56:50 PM] [DEBUG] [Image Plate Solve Thread] FitsFileHeaderData: Scale - 0.451656374008052
[8/13/2015 10:56:50 PM] [DEBUG] [Image Plate Solve Thread] FitsFileHeaderData: RA - 20.7631240788335
[8/13/2015 10:56:50 PM] [DEBUG] [Image Plate Solve Thread] FitsFileHeaderData: DEC - 30.7434163666526
[8/13/2015 10:59:03 PM] [DEBUG] [Image Plate Solve Thread] Image Plate Solve: Failed to solve image…
[8/13/2015 10:59:03 PM] [DEBUG] [Image Plate Solve Thread] Solve failed, no blind failover option, quitting…

cm

In your sequence, what is your 1x1 scale set to?

Ken:

Just arrived at observatory. I wanted to verify plate scale setting in the .sgf file I was using:

I will run the plate solve and capture the image file and log.

Charlie

After getting setup and camera focused, I started the PlateSolve2 test. I took a Frame and Focus image and invoked a plate solve. This screen shot is a composite showing the hint info (very accurate) and the PlateSolve2 dialog box for the failure to solve the frame:

I immediately edited the 34’ x 34’ entry to 25’ x 19’ and clicked the Plate Match button and it solved immediately.

Next, I used the “Save as Mono FITS…” option to save the Frame and Focus image to disk. Then I closed the Frame and Focus image in SGP and reloaded it from disk. Curious as to the FITS header info in that image, I opened the FITS header dialog box:

It shows a very basic set of info. I then performed a plate solve on the image. This resulted in a processing error in PlateSovle2:

I have the FrameAndFocus.fit file and the SGP log file that I will upload to Dropbox later today.

I believe the solution is to provide PlateSolve2 with the proper frame size (in my case, 25 x 19) when it is invoked, either in the referenced FITS header or in some kind of command line parameter.

cm

Here are the links to my drop box files:

The SGP log was copied as soon as the last testing was completed, so everything is at the bottom. The focusing for the frame and focus image was finished at 9:09:50PM.

cm

Sorry Chas,

I don’t think we can be of much help at this point. I downloaded your FITS file, entered the hints, right clicked, chose to solve and everything seems fine:

And please don’t confuse the lack of FITS data in the image you saved off as part of the problem you are experiencing… that data is not the same as data passed into the solver. As you can see FF frames do actually pass all hints in automatically. This is not your rig, but it shows what I am talking about:

I do have both catalogs installed… not sure what your setup looks like. You might try the latests beta… probably won’t make a difference, but it has regional fixes that change how PS2 is called from the command line.

Ken:

I’m using 2.4.2.6. The solution seems to be to fix the plate scale parameter – PlateSolve2 is being initialized with a plate size of 34 x 34 instead of the correct value of 25 x 19 (for my system). When I override the 34 x 34 entry with 25 x 19, PlateSovle2 solves almost immediately. The PlateSolve2 configuration dialog box has the proper values but these are not being passed to PlateSolve2 or they are being overridden.

Alternately, Control Panel has everything needed to calculate the correct plate size – pixel scale (0.451) and chip size (3326 x 2504) so where is the value 34 x 34 coming from? That is the code that needs to be reviewed.

cm

I have reviewed this code and there is nothing immediately apparent. Try
solving at 1x1 to see if somehow, dimensions are incorrect at other binning
levels.

Also, there is still the fact that I can solve your image without issue (from SGPro)… Are you able to do that?

The next version of SGPro will contain the actual set of args passed to the PS2 application.

Ken:

Being able to see what parms are being passed to PS2 will certainly help us identify what is going on.

I reloaded “FrameAndFocus.fit” (the test frame we have been using) and then attempted a plate solve. Since the FITS header for that image doesn’t have RA and DEC entries, the plate solve dialog box came up empty:

I used Astrometry.net to get coordinates:

I entered those coordinates and the other info into the Plate Solve dialog box and clicked “Solve”:

PS2 scrolled through the 999 regions and failed to solve the image even though it had exactly correct coordinates and plate scale for the image. I had checked “Keep Open” to allow me to then edit the plate size field to match my system of 25 x 19:

and clicked “Plate Match” to perform a second solve:

It solved instantly. If I don’t change the 34 x 34 entry and just click Plate Match a second time, it fails again.

So, let’s wait for the next beta showing the PS2 command line parms and I can test again. Maybe we can figure out where the 34 x 34 entry is coming from.

cm

Actually… I found the issue. It was one of precision. At higher focal lengths the precision of arguments was not high enough. All the numbers were correct, just too imprecise.

All of our testing was done at a much lower FL…

Ken:

Great news! Yes, I am imaging at 2440mm fl – 12" Meade LX850-ACF (f/8) optics.

cm

Hi

I have found at fL of 500mm and my 35mm chip ( sx h35 ) I can not solve a image with platesolve2 , even though at longer length all is fine ??

Regards

Harry

Not sure. At this point we should probably table all PS2 issue until the next beta is released… bug fixes and way more logging are included.

Hi

I still struggle to get platesolve 2 to work well out of the box , I did manage to get it to solve my problem images by upping the number of stars to 100 in the settings of platesolve 2 , even though this made it slow ???

who decides the default values in platesolve2 ???

regards

Harry

The author of PlateSolve2 I suppose.

One question I have is, can the ps2 solve window close sooner? Less than 10sec default?

No. That is a decision made by PS2 that we have no control over.

Hello,
I am still evaluating SGP (having no chance of a clear sky here) in the lab, so I was happy to see, that SGP uses Planewaves Platesolve engine as APT tool does, which I am evaluating too at the moment. I used 2 different image from my ccd (via open image). The astrometry.net blind solver within SGP has not trouble with this images. I wasted one or two hours with no success with platesolve2 than I saw this posting. I guess I ran into the same behavour.
Even I had equipment profile the scale value states 0,00, so I decided to populate it manually with my pixel scale of 3,38. Than I give a starting parameter RA/Dec for a faster solve. This does not work at all. The parameters in platesolve are wrong as chasmiller46 describes. If I stop platesolving and overwrite it with fancy values which are nearer to the reality (2x3deg) platesolve makes it in seconds. It can be a localisation issue of the input box…
What should I do? I am using the new SGP 2.4.2.10