Ultra Deep M51 Cooperation Project

[Yuriy Toropin]

Hi,

First, some introduction:

I'm participating in M51 cooperative project of 9 Russian amateur astronomers.
Let me name them here to keep correct copyright:
Andrey Kuznetsov, Boris Satovskyy, Dmitry Ivanov, Igor Checkalin, Ivan Ionov, Nikita Shamorgin, Oleg Milantiev, Vlad Onoprienko, and myself (Yuriy Toropin).
The progress of the project is covered in this thread on popular Russian astroforum.
Sorry, Russian only there, but if you will scroll through you will see some preliminary results.

Recently "master sums" of L,R,G,B channels have been cooked, magnitude of the faintest object identified so far on the L sum is ~24.4^m.
Right now we're doing posprocessing to get the final image to show and share.

For this project we've consolidated our materials that we have for M51 galaxy, obtained mainly the last or previous seasons with scopes ranging from my TEC-140ED APO to R-C 500 f/8, with wide variety of cameras.
I still have to cound total exposure for shots used but it will be something in 30-40 hours range or close to it.
The goal was to see how deep can we go in magnitude even in less than perfect seeing in Russia that is usually in FWHM~2-3.5" range, results is above and below you can see crop of the "L master sum".



I performed all prepocessing of supplied sums from individual contributors, cleaning them if required, aligning and integrating in master sums for each channel. PixInsight really shines during this work and I think that soon we will see some new PI adepts.

Nonetheless, there was one sad exception that finally cost me 3-4 hours of additional work.
When I was trying to align frames from different scopes, everything went well at the center of images even if scales varies in several times, but unfortunately stars don't match at the extreme corners of the frame with StarAlignment... For me it was total surprise. I was 100% trusting this process in PI. Investigation show that StarAlignment fails if frames have pincushion distorsion of "different signs". Frames from APO and R-C were easily matched while there was problem with Newtons with Paracorrs, see image below and check especially for the low-right corner of it.



Juan or gurus,
can you please tell me if something was wrong with StarAlignment settings or could be improved via them?
Otherwise, after initial alignment with StarAlignment, I had to postprocess those misaligned files manually with DynamicAlignment and it was REALLY time consuming...
If needed, I will upload ZIP archive of those frames to ftp for investigation and tests.

And the final words -
if you have good materials for M51 (scale from 1"/px, total exposure from 3...4 hours at least, round stars on sums) then
you are more than welcome to join this M51 cooperative project!
Feel free to respond here or contact me via email at yuriy.toropin(at)gmail.com.
With another 30 hours of contribution we will brake 25^m barrier!

[Carlos Milovic]

Hi Yuriy

In those cases with distortions, you should use DynamicAligment. It has a much more flexible algorithm, and will align anything

[Juan Conejero]

Hi Yuriy,

This is terrific work, and an absolutely wonderful image. And indeed  I am glad to know you've used PixInsight

On StarAlignment, try the following to improve the adaptability (the elasticity, we could say) of the registration model:

(i) Interpolation > Registration model: Use 2-D surface spline interpolation, instead of projective transformation.

(ii) Star Matching: Increase the maximize overlapping RANSAC optimization. In this way the star matching algorithm will try to maximize the overlapping region defined by all star pair matches detected.

The most important optimization here is (i). There are other possibilities, but all of them will degrade the robustness of the image registration procedure. (i) will improve your registration, and perhaps will fix the problem, at least to some extent. However, bear in mind that (for now) StarAlignment has not been designed to deal with differential large-scale distortion.

cost me 3-4 hours of additional work

Maybe that's because you defined too much alignment stars. After initial registration with StarAlignment, 10 - 12 alignment stars should suffice with DynamicAlignment: one at each corner of the image, one at the middle of each side, and 2 - 4 around the center. Note that you can store a DynamicAlignment process icon on your workspace and reuse it for each pair of images, provided they are very similar —and they should be, since you've aligned them.

Anyway, I am definitely interested in having a look at this material. I think it can be an excellent data set to improve StarAlignment in many aspects. If you want to upload them, you can use your FTP account.

Thank you for sharing these results with us; it is a privilege to have works of such high level on PixInsight Forum.

[Yuriy Toropin]

Hi Carlos,
you're right, after switch to DynamicAlignment (tip was provided by NKV on mentioned Russian forum) the problem was solved, but, according to Juan's mail, I overused it with two many alignment points selected, and it was really the case and led to a lot of time spent on this.

Juan,
first, thanks a lot for kind word, this project is really in progress, probably, more people will join and we will get "the deepest amateur M51" sooner or later with help of PI
Also, I'm uploading archive with 4 frames to ftp right now, with 2 of them StarAlignment failed at the extreme corners.

Thanks a lot for the tips, will try them!

For those of you interested we've issued "public" version of the best "total sums" of M51 we have for now.
There are 5 zipped files, available at this link.
They are aligned and integrated L, R, G, B channels of the total sums, together with subset L_BestResolution - sum of L channels of individual contributions with the best resolution (in terms of FWHM). "Public" version is saved in 16bit uint format to save the traffic, also it's not the full frame of the sum available but 1800x1800 pixels 100% scale crop around M51.

Feel free to download and play with the data, it's ready for channels combination and postprocessing.
Looks like we have to establish some prize for the person who will identify the faintest object on those sums or will generate the most beautiful rendition from this data!

Conditions under which it can be used can be found in provided Disclaimer.

Check the data, may be you will also want to join the project!!!

Regards,
   Yuriy

[Yuriy Toropin]

Recent news -
our project went to its international phase after Harry page and Steve Mazlin from SSRO contacted by another SSRO fellow imager Jack Harvey kindly shared their materials on M51.
Below is a screenshot of (strongly) stretched recent "grand sum of L channel" together with statistics on S/N ratio for individual contributions and for sums, previous best with S/N=12.79 and the last one that includes data from Harry and Steve at S/N=17.15 (~34% improvement in S/N). S/N is measured in FitStacker program, green boxes refer to areas where noise were measured (with filtering stars out) and red boxes are used for Signal (= Intensity - Background) evaluation.

After this contributions limiting magnitude on the sum should go up to ~24.6^m.

Also, Steve's SSRO input is not only valuable due to high S/N but also due to high resolution of his data!

New sums and aligned subsets will be available for participants soon!

Looking forward for new contributions!
All the best,
   Yuriy

[Yuriy Toropin]

Ultra deep amateur M51 project
is progressing, with recent updates from
   Pedro Re, 445 min in L channel, [link] to contributed shot,
   Russel Croman, who has kindly contibuted LRGB = 480:210:75:180 = 945 minutes from his great M51 [link].
the M51 grand sum has total exposure of ~120 hours!!! (including those data from Harry and Steve mentioned in previous message and initial "Russian" sum).

The latest and greatest "grand sums" were compiled, a little bit cropped, downsampled by factor of 2, 16-bit sums are available for public at this link
http://files.mail.ru/8R5DQ4

Hope, you will like to work with it!
There are 3 zipped files available (see screenshot below), sum for L, sum for R,G,B (combined in color RGB file, with background calibrated and color balance set based on bckground galaxies, so it could be a little bit blue), and special sum of L combined from contributions with the best FWHM.

Disclaimer is added to every ZIP archive.

Download from this file sharing service should be pretty straightforward, see screenshot, switch to English via button in right top corner if for some reason it was opened in Russian and you can't read it (and you don't, just click on Download buttons )

[Yuriy Toropin]

The question is - how can we identify the magnitute of the faintest object visible on the latest "grand sum"?

The issue is - even NED catalog in Aladin is not enough here, all objects available in it is easy to identify on the sum, and there are still fainter unidentified objects.
Predicted magnitude for it should be around ~24.8^m on the sum, how to prove it? Some other catalogs on Aladin?

Anyway, below is the latest 120-hours "grand sum" in 1:4 scale with fragments in 1:1 in positive and negative editions presented below.
The fragment's location is marked on the sum, the most noticable objects on frangment is IC4277 (at the right-top corner) and IC4278 (near middle of the bottom side), scale on fragments is ~0,455"/px.

Can anybody, please, help with identification or with clue on how to identify objects weaker than 24^m on the sum?

   
[Click on preview to open]

[Harry page]

Hi

I wonder what the faintest a amateur image has gone ? , I have had problems my self when going fainter than mag 22 so I am very interested in this as well


Harry

[Yuriy Toropin]

Meantime, 25^m object has been identified on the latest international "grand sum", see picture below.
Definitely, this is not the faintest object presented on the sum, unfortunately NED catalog in Aladin is not complete for such faint objects.

Literally, just next to that 25^m object there are some even fainter ones, yet unidentified for now.

[georg.viehoever]

Yuriy,

Meantime, 25^m object has been identified on the latest international "grand sum", see picture below.

here is a 26.1^m galaxy (MESSIER 051:[DMF2003] 52) that appears to be visible. Found by solving the image with the Blind Solver script, and following a hint by Vincent Peris to this object .

I am not sure, however, how useful this information is. The magnitude depends on the filters that have been used. The SDSS catalog that is also accessible via Aladin contains more details about this, but unfortunately it has few objects that are fainter than 24 mag.

Georg

[georg.viehoever]

See more on the magnitude topic in http://pixinsight.com/forum/index.php?topic=1943.msg12496#msg12496 and following messages.