M31 Mosaic revisited

[tomato]

I had given up on trying to add the colour channels to my 12 panel M31 project, but after some tips from @Laurin Dave, I had another go. I cheated a little by using the colour obtained from my RASA8/268c image which just seemed to be there right out of the camera.

Anyway, the galaxy now has some colour, the background is still blotchy and I have some great colour artefacts around some of the stars, but I am still astonished how the software (APP in this example) can combine data from totally different optical set ups.

Thanks again Dave, you have rekindled my enthusiasm  for mosaics.

 

[ollypenrice]

I'm a great fan of the mosaic and have just done an Astronomy Now article, I think for the next edition, on one approach to constructing them from individual panels.  The project in question began with APP but wandered off into Registar and Photoshop as well. (My fault!) So, having done a few big ones of my own, and having participated in two absolute giants, I thought I knew how to do it. Then, last week, I was given a 12 panel M31 dataset to try and construct. Hmmm... So far I am not succeeding. What's difficult is that only a small number of panels have any background sky to use as a calibration reference. I suspect that software aiming to do it all in one click has exactly the same problem, which is why it fails. So, I'm sorry to say that I don't have the answer on how to build a seamless and even M31 galaxy mosaic. I think I would be able to do it if I had a smaller number of panels, most of them containing at least a bit of background sky.

Difficult! But that's why we like it.

Olly

Edited January 12, 2021 by ollypenrice

[tomato]

24 minutes ago, ollypenrice said:

I'm a great fan of the mosaic and have just done an Astronomy Now article, I think for the next edition, on one approach to constructing them from individual panels.  The project in question began with APP but wandered off into Registar and Photoshop as well. (My fault!) So, having done a few big ones of my own, and having participated in two absolute giants, I thought I knew how to do it. Then, last week, I was given a 12 panel M31 dataset to try and construct. Hmmm... So far I am not succeeding. What's difficult is that only a small number of panels have any background sky to use as a calibration reference. I suspect that software aiming to do it all in one click has exactly the same problem, which is why it fails. So, I'm sorry to say that I don't have the answer on how to build a seamless and even M31 galaxy mosaic. I think I would be able to do it if I had a smaller number of panels, most of them containing at least a bit of background sky.

Difficult! But that's why we like it.

Olly

Agreed, it’s kind of encouraging that you are not finding it easy!😊

One of the major issues I encountered was capturing the data over so many sessions (18 in total), which meant I had a whole range of sky conditions with, dare I say it, the moon in attendance on a couple of occasions.

But you can’t beat the detail captured on a large galaxy like M31 with the mosaic approach so I will persevere!

[Sunshine]

WOW, that is one big image of andromeda, beautiful detail! 

[TerryMcK]

Great work Steve.

[gorann]

Steve, how does the resolution of your Esprit 150 images compare to your RASA data? I would have expected more resolution with the Esprit but when I compare it to my first light with only 2.6 hours with the RASA (https://www.astrobin.com/0kff02/C/) I do not see a difference in resolution. Maybe aperture wins over focal length here? Makes me worried since I also have an Esprit 150. Could of course be related to different sky conditions / Bortle between our sites. In the crop below, your Esprit image is to the left.

[Laurin Dave]

Well done Steve that's looking good and I'm pleased that your enthusiasm has been re-kindled.    For those with Pixinsight there is a new Photometric Mosaic Script (available in the latest release) which has been developed by fellow Basingstoke Astronomical Society member John Murphy.  On the examples that have so far been tested it has given excellent results and is worth investigating although by necessity it is most definitely not one click.  

Dave

[tomato]

20 hours ago, gorann said:

Steve, how does the resolution of your Esprit 150 images compare to your RASA data? I would have expected more resolution with the Esprit but when I compare it to my first light with only 2.6 hours with the RASA (https://www.astrobin.com/0kff02/C/) I do not see a difference in resolution. Maybe aperture wins over focal length here? Makes me worried since I also have an Esprit 150. Could of course be related to different sky conditions / Bortle between our sites. In the crop below, your Esprit image is to the left.

The L is the 12 panel mosaic taken with the Esprit 150/G2-8300 combo, the RGB channels are from my RASA8/268c image, which is shown below.

 

[Laurin Dave]

Steve kindly sent me his data so that I could try out the new Pixinsight Photometric Mosaic Script that I mentioned in an earlier post.  I already had an M31 taken with my Esprit100/SX46 which captures nearly all of Steve's fov in one frame.  After registering everything together using Pixinsight's Platesolving and Mosaic by Co-ordinates scripts  I firstly used the script and my M31 data to remove the gradients from Steve's higher res data, I then used the script again to combine the High res data as a mosaic and finally again to pad out the high res data with a wider star field.  The resulting L and RGB images were then processed in Pixinsight and Photoshop....

Here's the result....   not perfect with a number of joins and other artefacts visible if you pixel peep but otherwise speaks for itself and the power of the the new script..  Great detail from the Esprit150 too..  

Dave

Edited 19Jan21....  reprocess to remove the magenta halo which in my rush to post last night  failed to notice...   also added a version with Ha data 

With added Ha jewels..

 

 

 

 

 

 

Edited January 19, 2021 by Laurin Dave
improved to remove magenta halo..

[geeklee]

10 minutes ago, Laurin Dave said:

Here's the result....   not perfect with a number of joins and other artefacts visible if you pixel peep but otherwise speaks for itself and the power of the the new script..  Great detail from the Esprit150 too..  

Absolutely fantastic result.

A Sunday evening talk on the methods described in that post would be interesting too 😃

[gorann]

30 minutes ago, Laurin Dave said:

Steve kindly sent me his data so that I could try out the new Pixinsight Photometric Mosaic Script that I mentioned in an earlier post.  I already had an M31 taken with my Esprit100/SX46 which captures nearly all of Steve's fov in one frame.  After registering everything together using Pixinsight's Platesolving and Mosaic by Co-ordinates scripts  I firstly used the script and my M31 data to remove the gradients from Steve's higher res data, I then used the script again to combine the High res data as a mosaic and finally again to pad out the high res data with a wider star field.  The resulting L and RGB images were then processed in Pixinsight and Photoshop....

Here's the result....   not perfect with a number of joins and other artefacts visible if you pixel peep but otherwise speaks for itself and the power of the the new script..  Great detail from the Esprit150 too..  

Dave

 

 

 

 

Oh, that is very pleasing to they eye - great image!

Edited January 18, 2021 by gorann

[tomato]

I was pretty sure the data was badly compromised given the great variation in sky conditions I encountered during the capture sessions but Dave and the script has produced a result far beyond my expectations.

+1 for the Photometric Mosaic script!

[Laurin Dave]

On 12/01/2021 at 20:03, ollypenrice said:

I'm a great fan of the mosaic and have just done an Astronomy Now article, I think for the next edition, on one approach to constructing them from individual panels.  The project in question began with APP but wandered off into Registar and Photoshop as well. (My fault!) So, having done a few big ones of my own, and having participated in two absolute giants, I thought I knew how to do it. Then, last week, I was given a 12 panel M31 dataset to try and construct. Hmmm... So far I am not succeeding. What's difficult is that only a small number of panels have any background sky to use as a calibration reference. I suspect that software aiming to do it all in one click has exactly the same problem, which is why it fails. So, I'm sorry to say that I don't have the answer on how to build a seamless and even M31 galaxy mosaic. I think I would be able to do it if I had a smaller number of panels, most of them containing at least a bit of background sky.

Difficult! But that's why we like it.

Olly

Hi Olly 

Based on what it managed to do on Steve's data the Pixinsight Photometric Mosaic script might be worth a try on this,   the script could use your wide field M31 RGB and Lum data to remove complex gradients from the high res panels..  

HTH

Dave

[MartinB]

Wonderful image!  It is strange that one of the brightest and, potentially, most photogenic of deep sky targets is also the hardest to capture and process. 

[tomato]

Following @Laurin Dave's first class results with the Photometric Mosaic script in PI using one of his widefield images as the reference I was determined to have a another go using all my own data. Interestingly, a widefield shot taken with the RASA8/QHY 268c did not work, too much gradient in the image, (see PI screen shot below), but I had more success with  another taken under better conditions with the same scope and an Atik Horizon OSC.

The colours still aren't quite right, and there is a background cross hatching effect in areas where the high res data doesn't overlap with the widefield image,  but the script has certainly given me the best results from the data by a country mile.

I'm done with this mosaic for now, I'm looking forward to imaging some tiny galaxies that fit into the FOV.

Thanks again Dave and John.

 

 

 

 

[tomato]

Folks must be getting tired of me resurrecting this thread, especially as I promised to call it a day on the previous post, but...

The colour was some way off so I have run the data through from scratch again and have hopefully come up with something a bit better. 

It could do with  some Ha data, but that will have to wait until Autumn rolls around.