Jump to content

  •  

CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.

Photo

New plate solver program

astrophotography
  • Please log in to reply
48 replies to this topic

#1 han59

han59

    Mariner 2

  • *****
  • topic starter
  • Posts: 283
  • Joined: 09 Mar 2015
  • Loc: the Netherlands

Posted 30 April 2018 - 12:30 PM

For those interested or have a use for local plate solver the following:

 

The free ASTAP stacking program has now it’s own plate solver. It can process images in FITS format but also JPG, PNG, TIF file and give the image centre position and orientation.

It is command line compatible with PlateSolve2 and produces an similar APM file. It can also update the FITS headers or convert JPG,PNG, TIF to the FITS format.

Other outputs options are WCS or INI files containing the solution.

 

In addition it comes with a flexible viewer and stacking menu for FITS files.

 

With an good estimated position, the solution is found within seconds. With an offset of 20 degrees in minutes.

 

Available for MS-Windows and Linux. Still in development and at version ß0.9.44.  It is already integrated in CCDciel (CCD capture software) as external plate solving program.

 

Feedback is welcome.

Han

Link: http://www.hnsky.org/astap.htm

 

Ubuntu/Debian users can install ASTAP from Astronomy Linux Packages repository:

https://astronomylin...t/en:news:start

 

For plate solving a star database of 420mbytes needs to be downloaded and extracted to the program directory. It has in principle only three settings, 1) start magnitude and 2) limiting magnitude and 3) search radius in degrees. The limiting magnitude is a value for the faintest stars in an image of 10 seconds exposure. For FITS this value is automatic adapted to the exposure time. For other formats this has to be manually entered or specified in the command line. There is no need to define the pixel scale or the image astronomical size.

 

It uses four star pyramids for alignment as shown below:

astap_pyramids.jpg


Edited by han59, 30 April 2018 - 04:37 PM.

  • Ptarmigan, Rickster, Oleg Astro and 11 others like this

#2 artem2

artem2

    Ranger 4

  • ****-
  • Posts: 324
  • Joined: 08 Nov 2013
  • Loc: Österreich (no kangaroos in Austria)

Posted 30 April 2018 - 01:06 PM

waytogo.gif  Great Software Han, thank you Martin



#3 ChrisMoses

ChrisMoses

    Viking 1

  • -----
  • Posts: 988
  • Joined: 22 Oct 2014
  • Loc: Fort Wayne, IN, USA

Posted 03 May 2018 - 11:23 AM

Very cool.  Is it open source?  I would love to see the plate solving routine.



#4 han59

han59

    Mariner 2

  • *****
  • topic starter
  • Posts: 283
  • Joined: 09 Mar 2015
  • Loc: the Netherlands

Posted 04 May 2018 - 02:08 PM

It is open source and written in Object Pascal ( Lazarus& FPC). I have documented some principles at the webpage.

 

It is a brute force method but it works very fast. It tries to find identical (smallest) four star pyramids using the distances between the four stars in ratio.

 

http://www.hnsky.org...ernal_alignment

 

To program it and make it work as a developer you have to master things like:

 

  • Star detection, calculation background, noise level, standard devation, weighted center of gravity of a star image, histograms.
  • Having quick access to a star dabase up to magnitude 16+
  • Finding, handeling and comparing pyramids,
  • LSQ fitting in two dimensions.
  • Spherical Geometry

It is an own development. Working a long time on the HNSKY planetarium and some work on CCdciel helped a lot.

 

To get a flavor of the problems, the following document for astrometry.net describes it pretty well:

http://www.ppenteado...etry_201110.pdf

 

And it doesn't use quads smile.gif

 

Later, I have written down the method: http://www.hnsky.org...ate_solving.htm


Edited by han59, 05 May 2018 - 10:11 AM.

  • Oleg Astro, artem2 and ChrisMoses like this

#5 han59

han59

    Mariner 2

  • *****
  • topic starter
  • Posts: 283
  • Joined: 09 Mar 2015
  • Loc: the Netherlands

Posted 23 May 2018 - 09:30 AM

FYI,

 

Latest ASTAP version can add deep sky annotations to the solved image:

 

Han

 

astap_annotations.jpg

 

Sh2-232,Sh2-235_20180320_204438 x 21 stacked annotated compressed.jpg

 


  • Rickster, psandelle, Oleg Astro and 2 others like this

#6 daw316

daw316

    Vostok 1

  • -----
  • Posts: 127
  • Joined: 10 May 2012

Posted 24 May 2018 - 07:13 AM

I have been using the Linux version of ASTAP.  It takes a little learning, but is quite nice.  It is a good interface to a local astrometry.net solver as well as having its own method.  Han was very helpful in getting me past some start-up issues.  Upgrades are easy, so I will certainly try the annotations.  Definitely worth a try!


  • Oleg Astro and ChrisMoses like this

#7 Jairo

Jairo

    Explorer 1

  • -----
  • Posts: 66
  • Joined: 06 Nov 2012
  • Loc: Brazil, BA

Posted 02 August 2018 - 11:53 AM

han59, I'm interested in detecting stellar parallax with amateur equipment. By any chance, could this plate solver provide the centroids of the detected stars with the full precision? It would be exactly what I need right now.



#8 han59

han59

    Mariner 2

  • *****
  • topic starter
  • Posts: 283
  • Joined: 09 Mar 2015
  • Loc: the Netherlands

Posted 04 August 2018 - 04:34 AM

Jairo,

Interesting idea, but the stellar parallax is pretty low. I looked it up and Proximity Centauri has the largest parallax 0.7687 arcsec.

Assuming the seeing for most amateurs is a few arc seconds, you will still be possible measure is having the average centeroid relative to the nearby stars. The absolute position in ra,dec will not work since the errors will be to large for any solver. So the x,y positions of stars are required. The program calculates the x,y positions but they are not stored or exported.

It will be a pretty tough challenge. Since the astrometry.net plate solving program (local/online) standard exports the x,y values as a file would first it using astrometry.net.

http://astrometry.net/doc/readme.html
• augment-xylist: creates “augmented xylists” from images, which include star positions and hints and instructions for solving.
• image2xy: source extractor.
• plotxy: plots circles, crosses, etc over images.
• xylsinfo: print stats about a list of x,y positions (xylist).
….


p.s. the ASTAP viewer has been improved. It has now an experimental thumbnail viewer for FITS files.

Han

ASTAP v0.9.90 has now a FITS thumbnail viewer (ctrl-T). This could be useful to browse your FITS files. By clicking on the thumbnail it will be opened in the viewer. With a right mouse button click some options are available as changing directory, copy, move, rename or rename to *.bak.

The thumbnail size is depending on the form size. Make it larger, the thumbnails will follow. Thumbnails are organized in 3*X. So the thumbnails are pretty big by purpose. The images are fully loaded in memory so it will consume some memory and time. So don't try get thumbnails of 400 images.


astap_thumnails.thumb.jpg
  • artem2 likes this

#9 han59

han59

    Mariner 2

  • *****
  • topic starter
  • Posts: 283
  • Joined: 09 Mar 2015
  • Loc: the Netherlands

Posted 14 August 2018 - 05:49 AM

The development continuous. The latest ASTAP version 0.9.92 has a blink routine (for FITS files) to detect minor planets or nova or just make a video of a comet moving:

 

Youtube demonstration:

 

https://youtu.be/7cheLI-3ggQ

 

At the moment the blink sequence can't be exported, but you could capture it by a screen capture program.

 

So ASTAP has become more a software suite. The main functionalities are:

 

- FITS viewer and manipulator including annotation of deep sky objects and stars.

- Stacking program, blinker

- Plate solver with two types of command line interfaces, remote accessible.

 

Feedback is welcome

 

Han

 

The new blink tab:

astap_blink_routine.png.797886b93052f886


Edited by han59, 14 August 2018 - 06:05 AM.

  • bsavoie likes this

#10 daw316

daw316

    Vostok 1

  • -----
  • Posts: 127
  • Joined: 10 May 2012

Posted 14 August 2018 - 07:49 AM

ASTAP just keeps getting better.  Thanks!



#11 han59

han59

    Mariner 2

  • *****
  • topic starter
  • Posts: 283
  • Joined: 09 Mar 2015
  • Loc: the Netherlands

Posted 21 September 2018 - 08:27 AM

The plate solver in ASTAP version 0.9.114 just released can do blind plate solving. Om my system is achieves a full sky scan in about 380 seconds. This means that blind solving should in most cases have a result in half of the time is 190 seconds of better depending on the offset. Not bad for a brute force method, single processor program.

 

For this the search spiral was adapted for the sky sphere. Secondly the dept of the star database reading is adapted to match the number of stars in the image.

 

Note that the internal plate solver works best with raw unstretched and sharp images of sufficient resolution where stars can be very faint. Heavily stretched, very long exposures or photo shopped images are problematic.  It requires minimum of about 30 stars. Images containing of a few hundred stars are ideal. For star rich images the detection limit is adapted to limit the number of stars detected. This will only work for unstretched images. In the latest version also binning prior to solving is available for large images above 3000 pixels wide. This will also improve signal to noise ratio. For very problematic images Astrometry.net will do a better job.

 

 

The plate solver work fine with CCDCIEL and can work as a PlateSolve2 substitute in APT.

 

The development is ongoing. Please report any problem.

 

Clear skies, Han

 

astap_performance.png


Edited by han59, 21 September 2018 - 08:38 AM.

  • artem2, RazvanUnderStars and arspi like this

#12 555aaa

555aaa

    Vendor (Xerxes Scientific)

  • *****
  • Vendors
  • Posts: 1441
  • Joined: 09 Aug 2016
  • Loc: Lynnwood, WA, USA

Posted 23 September 2018 - 10:40 AM

Jario, astrometrica can do what you want. It has both centroiding and polynomial fitting for field distortion. It is possible to get plate solve residuals as low as 50 milli arc seconds with amateur equipment. But you need to use an astrometrically precise catalog such as Gaia.
  • bsavoie likes this

#13 han59

han59

    Mariner 2

  • *****
  • topic starter
  • Posts: 283
  • Joined: 09 Mar 2015
  • Loc: the Netherlands

Posted 24 September 2018 - 01:50 AM

Astrometrica's objective is to measure accurately.  The objective of the ASTAP solver is to fix the "lost in space problem" if your telescope pointing way off or stack/stitch images to a mosaic. So different objectives with different criteria.

 

ASTAP uses a local star dabase based on Gaia but does not pretent to reach milli arc seconds accuracy and uses just a linear fitting in two dimensions.



#14 catalogman

catalogman

    Viking 1

  • *****
  • Posts: 998
  • Joined: 25 Nov 2014

Posted 18 November 2018 - 03:18 PM

For a wide field (~20 deg), I tried a local Astrometry.net (ansvr) in ASTAP using the Tycho-2 catalogue .fits index at

 

http://data.astrometry.net/4100

 

Unfortunately, ASTAP doesn't have smaller catalogues such as Tycho-2 for wide fields, so unless you use the local Astrometry.net method you're stuck doing a slow search through one of ASTAP's much larger catalogues. (The smaller catalogue from an HNSKY install doesn't work in ASTAP.)

 

In fact, none of the .zip and .rar star catalogues load into ASTAP. The only way to load a star catalogue into ASTAP is to run one of the .exe for HNSKY's large star catalogues and install it into the ASTAP folder. When I try to solve the same wide field above using the AstroImageJ/ansvr setup without ASTAP, even when I use the WCS tab the "solution" that AIJ gives doesn't add any position info to the .fits file unless the image is from a digital mount that has recorded the central position. The HN_FITS tool has the right idea to fix this (locate the center and create the transformation matrices from two known stars), except that the mouse doesn't work in the tool (I tried several OS's).  So if ASTAP could make full use of the WCS tools and could include a working version of the HN_FITS tool then it would be the free plate-solving app that we have been looking for.

 

In short, my recommendations for improving ASTAP are:

 

- fix the reading of star catalogues (from .zip, .rar, etc.)
- include smaller star catalogues for faster solving of wide fields (HIP, Tycho-2, SAO, BD/CD, etc.)
- fix and include the HN_FITS tool (to provide a very good hint for ASTAP)
- implement WCS tools (for blind solve of images not from a digital mount)

 

--catalogman


Edited by catalogman, 18 November 2018 - 03:18 PM.


#15 catalogman

catalogman

    Viking 1

  • *****
  • Posts: 998
  • Joined: 25 Nov 2014

Posted 21 November 2018 - 11:29 AM

Other possible recommendations for improving ASTAP:

 

The solved .fits files in ASTAP cannot be read in another plate solver such as Astrometry.net, and vice versa.

When a .fits file solved on Astrometry.net is opened in ASTAP, for instance, all I see is a white screen (the

DSO annotations are visible but the image is not).

 

Also, there should be a way of changing the number of DSO's displayed. (Maybe include different levels

of the database on the pages of the DSO spreadsheet. For instance, if the observer displays Page 3 on

the spreadsheet then annotations from that page are displayed on the image.)

 

--catalogman 



#16 han59

han59

    Mariner 2

  • *****
  • topic starter
  • Posts: 283
  • Joined: 09 Mar 2015
  • Loc: the Netherlands

Posted 22 November 2018 - 05:37 AM

For a wide field (~20 deg), I tried a local Astrometry.net (ansvr) in ASTAP using the Tycho-2 catalogue .fits index at

 


In short, my recommendations for improving ASTAP are:

 

- fix the reading of star catalogues (from .zip, .rar, etc.)
- include smaller star catalogues for faster solving of wide fields (HIP, Tycho-2, SAO, BD/CD, etc.)
- fix and include the HN_FITS tool (to provide a very good hint for ASTAP)
- implement WCS tools (for blind solve of images not from a digital mount)

 

--catalogman

 

 

Other possible recommendations for improving ASTAP:

    The solved .fits files in ASTAP cannot be read in another plate solver such as Astrometry.net, and vice versa.
    When a .fits file solved on Astrometry.net is opened in ASTAP, for instance, all I see is a white screen (the

    DSO annotations are visible but the image is not).

ASTAP has a dedicated Gaia dr2 (the latest) star database up to magnitude 17 (or 16 or 18 mag versions are also provided). This database is sorted from bright to faint so can easily provide the brightest stars of a 20 degrees section of the sky. It is also ultra compact stored. So an other database is not required. ASTAP can't read any other star database format as indicated. 

 

The solving of large fields will be problematic by optical distortions, not due to the used database. ASTAP assumes a perfect telescope and projection. Also due to the optical laws, stars in the outer regions image will be oval/comet like. ASTAP has successfully solved fields up to 4 degrees. You can play a little with the tolerance (<=0.007) and numbers of stars (50-1000) used but it will most likely fail for fields above 4 degrees.

 

I considered using only the center of the image for solving wide fields. As long it contains 50 stars, the center solution could be used to extrapolate the solution for the whole image. I haven't worked out this idea since wide fields are normally not considered "lost in space" but probably it is still nice to have an astrometric solution to identify object. 

 

I could try to add this large field solving. Could you provide some 20 degrees image samples to work on?

 

The HNS_FITS tool is no longer developed. To solve JPEG/TIFF/PNG  images, just enter the approximate image center in the ASTAP viewer. The field size of the image can be set in the ∑ menu, tab alignment. The JPEG, TIFF, PNG image will solve fine as long the image is not heavily photo-shopped and not too much saturated. To identify bright stars from faint star, they should have a lower intensity in the image. ASTAP is most happy with raw unsaturated images, sharp in focus. As long star intensity is maintained it will solve either a noisy single 10 seconds exposure or 100 images stacked of 300 seconds each. So as long the the telescope was focused, good tracking was maintained, star intensity preserved and the image scale <=4 degrees it will work fine. For image larger then 3000 pixels wide use the binning option.

 

>>  The solved .fits files in ASTAP cannot be read in another plate solver such as Astrometry.net, and vice versa.

I haven't found a solution for this but you could bypass this by adding the option "--no-verify" to the Astrometry.net  call. This option is given in the ASTAP menu for calling Astrometry.net as an external solver.

 

>>   When a .fits file solved on Astrometry.net is opened in ASTAP, for instance, all I see is a white screen (the  DSO annotations are visible but the image is not).

Maybe the viewer data range is in manual. Set in the viewer the data range to low or medium or bright.

>>  Also, there should be a way of changing the number of DSO's displayed. (Maybe include different levels of the database on the pages of the DSO spreadsheet. For instance, if the observer displays Page 3 on

    the spreadsheet then annotations from that page are displayed on the image.)

At the moment is is fixed at 30.000 objects in a simple .CSV file. For a 20 degrees image this will be overcrowded with deepsky objects. Yes probably it is better to limit automatic on DSO size once a 20 degrees image can be solved. Again give me some 20 degrees sample images and I will have a look. Feedback helps to develop this program further.


  • artem2 likes this

#17 han59

han59

    Mariner 2

  • *****
  • topic starter
  • Posts: 283
  • Joined: 09 Mar 2015
  • Loc: the Netherlands

Posted 22 November 2018 - 06:18 AM

In addition,

 

Astrometry.net and ASTAP have a different approach in solving images. Astrometry.net can solve a pure black and white images without any grey levels. It requires only the x, y positions of the stars. By using the provided pre-computed index files, it can do blind solving.

 

ASTAP uses the star intensity as a third dimension. So it measures the x,y star position and star intensity. The measured star intensity allows to limit number of database stars to match and by doing so it can  do "blind solving" using the brute force method. Without the star intensity, ASTAP is lost.

 

I assume ignoring the star intensity by Astrometry.net is a design choice which makes sense if you try to solve classic glass plates from ages ago. If you restrict to electronic sensor images only it makes sense to include the star intensity.

 

Both programs use 4 star patterns for recognition where image scale and rotation are irrelevant. The documentation of Astrometry.net calls them quads. For ASTAP I call them triangular pyramids (or tetrahedrons). Using 3 star patterns give too many false detections.  The way the 4 stars patterns are measured is totally different.

 

As indicated, feedback is most welcome. smile.gif

 

Update: Astrometry.net also uses a criteria to identify the brightest/best stars. E.g. you can limit the number of stars by the parameter "--depth". The Astrometry.net algorithm is just more robuster for image saturation.

 

Han


Edited by han59, 23 November 2018 - 02:53 AM.

  • artem2 likes this

#18 Der_Pit

Der_Pit

    Viking 1

  • -----
  • Posts: 857
  • Joined: 07 Jul 2018
  • Loc: La Palma

Posted 22 November 2018 - 07:17 AM

ASTAP uses the star intensity as a third dimension. So it measures the x,y star position and star intensity. The measured star intensity allows to limit number of database stars to match and by doing so it can  do "blind solving" using the brute force method. Without the star intensity, ASTAP is lost.

Just wondering - does that limit it to linear images?  I'd assume stretching will largely change the relative intensity ratios....



#19 han59

han59

    Mariner 2

  • *****
  • topic starter
  • Posts: 283
  • Joined: 09 Mar 2015
  • Loc: the Netherlands

Posted 22 November 2018 - 11:52 AM

Just wondering - does that limit it to linear images?  I'd assume stretching will largely change the relative intensity ratios....

No really. ASTAP tries to extract the stars with the highest SNR value from the image up to the maximum number stars set. This is default set at 500 but it can be typical set between 30 and 1000 stars. As long there is a difference in SNR values between the bright and fainter stars it will work. If measures for each star the SNR values, so the total flux above the noise level. So a non-linear stretch will not do harm as long most stars are not fully saturated. Saturation if really disturbing to the SNR calculation and therefore the identification process.

 

Example:

 

Maximum number of stars is set at 500 stars.

 

Situation 1) 1x1.5 degrees image of 10 seconds exposure containing 100 stars above noise level..  100 stars will be extracted from image. For each search area the 100 brightest stars will be extracted from the database (area  corrected)

Situation 2) 1x1.5 degrees image of 500 seconds exposure containing 5000 stars above noise level. 500 stars with the highest SNR value will be extracted from image. For each search area the 500 brightest stars will be extracted from the database (area  corrected)

 

Situation 1) is not critical. It will probably work for black and white images.

Situation 2) is critical. It really has to find the 500 brightest stars with the highest SNR value. If not there is unlikely a match between the 500 stars extracted from the image and the 500 brightest stars extracted each time from the database.

 

In principle ASTAP will use the same approach as a human-being to identify a piece of sky. Identify the brightest stars and try to match them with constellations figures. If your standing in a medium dark location identifying the constellations is easy. But standing at the superb dark location with the Milky-Way prominent in the sky, the identifying the constellations could be more challenging due to the overwhelming amount of stars visible. 

 

An short exposed image containing only 100 stars will be blind-solved quicker and the whole sky will be searched in minutes. Setting the max star limit to 100 stars rather then 500 will speed up the blind solving for long exposures significantly, but with a higher risk of failure.

 

Images of 10 or 20 seconds exposure taken with my ASI1600 MM cool camera in bin2x2 and 100 mm APO telescope are solved by ASTAP in no time using CCDCIEL as imaging program. An initial telescope offset of 30 degrees is no problem and solved in maybe 1 minute maximum.

 

Han


  • artem2 likes this

#20 Der_Pit

Der_Pit

    Viking 1

  • -----
  • Posts: 857
  • Joined: 07 Jul 2018
  • Loc: La Palma

Posted 23 November 2018 - 05:15 AM

Thanks for the clarification Han - I had interpreted it wrong, thinking you (try to) match star intensities one-by-one.  



#21 han59

han59

    Mariner 2

  • *****
  • topic starter
  • Posts: 283
  • Joined: 09 Mar 2015
  • Loc: the Netherlands

Posted 23 November 2018 - 09:08 AM

While writing above explanation, I decided to visualize the process in an unreleased ASTAP version for testing and learning.  Below a stacked image of M17, exposure 26x50 seconds made with my ASI1600. The red boxes indicated the 300 brightest stars detected from the image. There are much more stars available but they are ignored. Same as a visual observer will use the brightest stars for orientation only.

 

This image was easily solved. The software requires only 3 pyramids for a match and typical about 30 stars minimum.

 

The yellow boxes indicate the stars read from the Gaia star database maintaining the same star density as the image stars. The Gaia stars are only read for a square area in the middle of the image, not the whole image. They red and yellow boxes mostly match but not for all. Stars in M17 cloud are not selected due to the high background signal of the cloud giving them a lower SNR value. Secondly some stars are skipped by the software since they are close doubles. The Gaia star database contains only the Gaia Bm (blue) magnitude which should be a reasonable match with the spectral range of the ASI1600 camera (including IR/UV filter) but not perfect.

 

The better the image stars and database stars match, the more reliable the solving will be, so this is an important topic. An image taken with a blue filter is in principle more difficult to solve using a star database with visual magnitude only .

 

Han

 

M17_50s_20180709_000507  0xFD  0xF  0xD  0x0R  0x0G  0x0B  0x0RGB  26x50L  _stacked used g18.jpg

 

 

 


  • artem2 and bsavoie like this

#22 catalogman

catalogman

    Viking 1

  • *****
  • Posts: 998
  • Joined: 25 Nov 2014

Posted 23 November 2018 - 11:28 PM

 

I could try to add this large field solving. Could you provide some 20 degrees image samples to work on? <snip>

 

Again give me some 20 degrees sample images and I will have a look. Feedback helps to develop this program further.

I'll send a test image to your direct address shortly. (Watch your spam folder!) ASTAP and the Astrometry.net site can't solve it; Unimap and AstroImageJ/ansvr can solve it. By a tweaking of the data range controls, ASATP does display a badly rendered image of the solved .fits file.

 

 

ASTAP has a dedicated Gaia dr2 (the latest) star database up to magnitude 17 (or 16 or 18 mag versions are also provided). This database is sorted from bright to faint so can easily provide the brightest stars of a 20 degrees section of the sky. It is also ultra compact stored. So an other database is not required. ASTAP can't read any other star database format as indicated. 

ASTAP can't solve my image on its own, and the star catalogues may be why. Gaia is actually a poor choice (the distribution of m>17 stars doesn't match

the real sky due to overcrowding, many bright objects are missing due to oversaturation). To give ASTAP a better hint, I wanted to try the better tuc_star_mag15 database instead. But only the larger Gaia .exe databases load into ASTAP.

 

The .zip databases don't load, either. In other words, ASTAP can't read its own databases. 

 

 

 

The HNS_FITS tool is no longer developed. <snip>

I tried it anyway as another way of giving ASTAP a very good hint. But the tool ignores the mouse and no calibrating stars can be clicked on. IMHO, the tool would be a very nice addition to ASTAP if it wasn't broken.

 

 

 

At the moment is is fixed at 30.000 objects in a simple .CSV file. For a 20 degrees image this will be overcrowded with deepsky objects. Yes probably it is better to limit automatic on DSO size once a 20 degrees image can be solved. <snip>

The DSO database is probably okay as-is if users can create custom catalogues by processing a copy of the spreadsheet for different fields and using that copy in ASTAP.

 

...

 

Thanks for listening. We do need a stand-alone plate solving program that won't be affected by the hard drive failure of a website!

 

--catalogman



#23 han59

han59

    Mariner 2

  • *****
  • topic starter
  • Posts: 283
  • Joined: 09 Mar 2015
  • Loc: the Netherlands

Posted 25 November 2018 - 04:07 PM

For a wide field (~20 deg), I tried a local Astrometry.net (ansvr) in ASTAP using the Tycho-2 catalogue .fits index at

ASTAP version 0.9.138 just released can now solve wide fields. It will use the center of the image. This will only work if you have a sharp 10 or 20 megapixels image. Two test files taken from Astrometry.net upload will demonstrate this. You could download the "new-image.fit" from the links below to test it and e.g.modify the start search position in the viewer.

 

9.62 x 6.41 deg:

http://nova.astromet...21680#annotated

 

17.3 x 11.9 deg

http://nova.astromet...21647#annotated

 

 

Use the following new settings:

new_astap_setttings.png

 

Image stars outside the center (3 degrees) will be ignored. It will only use the stars from the centre.

 

I had no luck with the image from Catalog man. It looks like a scanned image and jpg quality was not good enough.

 

Han

 

 

 

 

 

 

 

 


  • bsavoie likes this

#24 catalogman

catalogman

    Viking 1

  • *****
  • Posts: 998
  • Joined: 25 Nov 2014

Posted 30 November 2018 - 01:01 AM

<snip>

 

I had no luck with the image from Catalog man. It looks like a scanned image and jpg quality was not good enough.

 

Han

 

Sorry to hear that my image doesn't solve in ASTAP. 

 

The ansvr loaded with the larger 2MASS catalogue doesn't solve my image, either.

 

BUT when the ansvr is loaded with the smaller Tycho-2 catalogue, I get a solution in a few seconds.

 

This makes sense because bright stars with large images are not greatly affected by distortion. The software tries to make a fit to only a few precise points and ignores the fainter points affected by distortion.

 

Therefore, I'm again making the case to include smaller search catalogues as a way of improving the performance of ASTAP. (Allowing the user to set the search magnitude limit wouldn't work with your Gaia catalogues because those lists do not include many bright objects due to CCD saturation.)

 

--catalogman



#25 han59

han59

    Mariner 2

  • *****
  • topic starter
  • Posts: 283
  • Joined: 09 Mar 2015
  • Loc: the Netherlands

Posted 30 November 2018 - 11:42 AM

If Astrometry.net fails using 2mass index files  indicates it is a problematic image The 2mass catalog is pretty complete on the bright side. Since Astrometry.net fails to solve using the 2mass indexes, I assume the near infrared magnitude J is too different from the image, so other star quads are taken. In the near infrared the star patterns look different then in visual range. The Tycho contains two magnitudes and probably the chosen magnitude matches more the image.  I assume that is the reason why Astrometry.net solves the image using the Tycho indexes.

 

The Gaia dr2 catalog compilation (G17) I'm using contains the Gaia Bm blue magnitude which matches the the visual range. About 456 Tycho stars were added after doing a Tycho2 database search up to magnitude 8.  For the added stars, the Bm magnitude was calculated using the Gaia documentation and epoch was also corrected.

 

What I'm trying to say is that the G17 star database is superior to any Tycho or 2Mass catalog and complete. It is always interesting to investigate why an image doesn't solve, but I want to restrict that to raw original images in the luminance channel. Scanned images, jpeg compressed introduce artifacts and saturated stars which make solving much harder. The problematic image you provided looks like a deep sky survey image since it contains some hand writing. ASTAP can solve DSS2 red images, but I never tested it above 4 degrees.

 

I also tried some days ago the G16 database which contains the calculated visual magnitude and it performed once a little better, but not enough to solve the problematical image. That could be a coincidence. In any case it would be ideal if the star catalog band-pass matches the image band-pass. Then the same star patterns (quads/pyramids) are selected.

 

 

Han




CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.


Recent Topics





Also tagged with one or more of these keywords: astrophotography



Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics