Annotation script

[georg.viehoever]

This seems to be interesting not only for color calibration, but also for brightness calibration. Part of a mosaic problem that I recently investigated http://pixinsight.com/forum/index.php?topic=4105.msg28803#msg28803 was that the different panels had different calibrations, and that LinearFit/FrameAdaption only has limited success in doing a relative calibration due to small overlaps and some issues in the original data. I was wondering if the ImageSolver plus some catalog could be used for calibrating them all to the same brightness levels.

Georg

[vicent_peris]

At first I was thinking that I'd like to be able to label all G2V stars.  A simple way didn't occur to me at that time of the morning, but I did recall reading about another method of white balance using B-V magnitudes.  I discovered that the NOMAD1 catalog has the appropriate information.  Based on this PDF article on B-V colour calibration, I learned that G2 stars have a constant B-V magnitude of 0.65.  The article also talks about V-R magnitudes of 0.36 as well, although that's less relevant for some reason I don't understand yet.  He goes through a method using Aladin, and uses the NOMAD1 catalog to filter out just stars with B-V in an "acceptable" range of 0.6 to 0.7, and R-V of 0.2 to 0.6.

Hi,

I never trusted this method,. Let me put one example.

We have this M106 picture:



The color has been calibrated using the entire galaxy. In this particular picture, this calibration results in a slightly green color, but it doesn't matter for this example. As you say, in the PDF they have a tolerance for selecting stars of 0.2 mag in the B-V color index, and 0.4 in the R-V. If we transform these ranges to linear (2.5118^0.2 and 2.5118^0.4), we get that .2 magnitudes means an increase in flux of 20.2%, and the increase in flux for .4 magnitudes is 44.5%!

Let's transform the above image to really understand what means the specified tolerance in terms of the resulting color balance. If we multiply the red channel by 1.445 and the blue channel by 1.202, we get this image:



If we saturate a bit the colores... Then we have these images:





I'm not particularly happy with the idea of having this amount of error in my color calibration. Even if we have ten stars and make the average of them. On the other hand, the NOMAD color have nothing to do with your instrumental color. The resulting color using the same calibration stars but different cameras and filters will be also different. This method can be only made by people that don't know/understand what's absolute photometry.

And, at last, I don't like the idea of mixing astrophotography and religion. This method has been done to force the idea of calibrating the color with the sunlight more happy and feasible. To me, the Sun it's only a star, not a god...


Regards,
Vicent.

[sleshin]

I agree with Andres, if someone wants to have a color calibration tool in PI based on B-V or V-R color index data, then make it a separate script/tool. I hope this wonderful script can continue to be devoted to providing information, ie names, color indices, etc. Parenthetically, IMHO, the current color calibration tool in PI is terrific and for me, works great, much better or at least more consistently than when I've tried color calibration based on finding G2V stars in an image to determine the RGB ratio using ExCalibrator.

Andres, in this script I would hope you can configure it to show B-V and V-R ratios, or just B, V, and G mags, for as many stars in an image as possible. It has been my practice to use a near finished image and check the accuracy of the star color using these values. In this way, a star with a B-V index, for example, greater than 0.65 should be red and one less than 0.65 should be blue, and one with a value close to 0.65 should be white. So, if possible allow all values to be shown, ie an option to have the data unfiltered.

Steve

[vicent_peris]

Parenthetically, IMHO, the current color calibration tool in PI is terrific and for me, works great, much better or at least more consistently than when I've tried color calibration based on finding G2V stars in an image to determine the RGB ratio using ExCalibrator.

That's what I'm talking about. No one can say that a method with 40%+ error is a robust one.


Regards,
Vicent.

[Andres.Pozo]

New version of AnnotateImage:

• Multiple labels per object
• NOMAD-1 catalog with optional B-V filtering for white balance
• Fields B, V and B-V in Tycho-2 catalog

Juan, I think this script is ready for publishing it. At least, the solver script should be updated since the published version with the update server doesn't work.

[troypiggo]

Vicent, thankyou so much for your experienced and educational reply.  I understand what you're getting at about tolerances and variability.  Those ranges are only a recommendation by the author of that paper, so I guess the end user could tighten them up if desired.  You would have far more knowledge than I on the matter of accuracy of the overall method.

I've noticed you recommending use whole galaxies as the white balance reference several times.  But I rarely shoot galaxies to date, mostly nebulae.  Do you recommend the G2V method as more accurate?  If there aren't any G2V stars in the field of view, what other candidates do you recommend as acceptably close enough?

I would like to point out that my intention was never to make this annotation script a colour calibration tool.  Merely to use it to highlight/annotate potential colour calibration stars.  This could be extended to whatever method the user wants.  If someone can find a suitable way of doing similar for G2V stars, that'd be awesome.

I agree the ColorCalibration tool is working wonders for me.  But I have a bit of a hard time finding stars to white balance on for cluster or nebulae shots.  I thought this would be a good way of assisting using the available features, that's all.

[NKV]

Is it possible to use Annotation script without Internet ?

[Andres.Pozo]

Is it possible to use Annotation script without Internet ?

Yes, but in this case you can only use grid layer, text layer, NGC/IC catalog and custom catalog.

[NKV]

custom catalog.

I have downloaded Tycho-2. Is it possible convert the Tycho-2 to custom catalog?

[Herbert_W]

Hi Andres!

I have several 'Custom Catalog'.
Could it be possible, to create a user defined name instead of 'User defined catalog'?


Kind regards.
Herbert, Austria

[Andres.Pozo]

custom catalog.

I have downloaded Tycho-2. Is it possible convert the Tycho-2 to custom catalog?

Tycho-2 has 2.5 million objects. I don't know if the Javascript engine can load so many objects and how much time it would need. It would be an interesting test that you could do.

In any case, IMO the best way for creating custom catalogs is filtering the original catalog using Vizier so that it contains only the area of interest. Then using Excel is easy to convert it to the format of the custom catalogs.

[NKV]

Tycho-2 has 2.5 million objects. I don't know if the Javascript engine can load so many objects and how much time it would need. It would be an interesting test that you could do.

MaximDL+PinPoint do it without problem, so IMO PixInsight should do it too.

In any case, IMO the best way for creating custom catalogs is filtering the original catalog using Vizier so that it contains only the area of interest.

Sometimes, I go to place in mountain where no electricity, no cellular, no Internet and of course no Vizier. And when I take some photo of new area and I want to know what is it in my image or etc. I should use other software for inspecting my image. 
Maybe possible implement Vizier to PixInsight It seems to me stars-generator - this is a good candidate.

[Andres.Pozo]

MaximDL+PinPoint do it without problem, so IMO PixInsight should do it too.

But they don't do it in Javascript using voluntary time.

In any case, a way for reading text files by lines would help a lot. Currently, the script has to load the entire file into memory and then split it by lines. If I could read the file by lines, I could process them independently without loading all into memory at the same time.

Also, I doubt that Excel can load 2.5 million rows, so it would be necessary a script for converting the format.

[Sean]

Andres,

This is a wonderful addition to PI - thank you so much for your hard work! Somehow I missed this new script until now - perhaps Juan can publicize it a bit more in the Releases forum.

I'm giving a short PI tutorial at a conference next week, and will use the Annotate Script to illustrate the power of user additions to PixInsight.

Thanks again,

Sean

[Luigi]

I just 'got around' to trying it last evening as well.

Bravo, Andres! If you're ever in Pennsylvania I'm buying