Large mosaic in APP

Can anyone help or advice in above topic ? 

Hi Marcin @rosma,

Thank you very much for your post, and please accept my apologies for not helping you sooner.

Wow, that 14 panel mosaic with the desert foreground is fantastic, great work !

Your description indicates that you are already very familiar with how to do this with APP. Indeed you need the calibrated projective model with a different projection than rectilinear. In addition, you then need to start modifying the Center of Projection or COP to get the most pleasing result like you were already trying.

Now, the 45 panels ar 40mm will indeed create a huge field of view in degrees. which might make it a bit difficult to get the correct COP. DId you try to get a good milkyway shape an projection by only using the panels with the milkyway data first? Maybe that will help you solve this. If you can get that correct, then simple add the other panels while keeping your reference frame fixed and the COP fixed. I would think that should get you the result wanted?

Let me know if you need assistance. Data like yours is also very helpful for us to test this and improve APP on this front as well. So please, if you want us to test this, please upload the data here and I will use it for more improvements:

https://upload.astropixelprocessor.com/

username: uploadData

password: uploadTestData

Please make a folder with your name and issue like Marcin-40panelMosaic-COP

and let me know once uploaded, thanks !

Mabula

Hi Mabula, Thank you very much for your response. No problem with response time :), I fully understand the amount of questions you need to answer every day. I am very happy I received an advice from you. I will centrainly have a try this week and definitely I will upload all the panels to the link you shared. Speak to you soon. REgards. Marcin 

Hi Mabula,

I am writing as I need your help in this topic. Yesterday following your advice I have tried to put this together one more time with panels which consists only milky way.
This end up with much better result however this was not perfect. Below you may see it in attached photo. Milky way shape seems to be correct now however projection of the milky way is not put accross entire photo width but only half of the photo width. In this exercise I have limitted it to 26 panels instead of 46. Anyhow I have tried to take out 1 or 2 additional panels hoping this will help but it did not. As I did not receive a proper result I decided anyway to have a try to second attempt according to your advice I mean to put all the rest of the panels together with below attached photo used as a fixed frame with fixed COP. This time I also encountered an error which is attached to this mail. I have no idea why is this happaning. Below couple of questions which may help me in further process:

1. Why the milky way projection now in not put accros entire photo width but only half of the photo width ?
2. Can I somehow identify a panel which is the reason of having the projection only for the half of the photo width? to take it out and doing it again
3. Why the stars above the milky way seems to be completely incorrect ? I mean they seems to be swollen but only above the milky way.
4. What do you exactly mean by fixed frame and fixed COP? ( "DId you try to get a good milkyway shape an projection by only using the panels with the milkyway data first? Maybe that will help you solve this. If you can get that correct, then simple add the other panels while keeping your reference frame fixed and the COP fixed" ). What I have done in the process I have simply uploaded below photo as a single frame and than I have uploaded all the rest of the panels. as a registration model I have still used "calibrated projective" and as a projection method as still used "mercatorprojection". What I understood from you by fixed COP, I have changed the paramiter of MOV COP vertical and this time I fixed this to 0,00 degrees. Is this correct or I should do this in another way ?
5. How I should fix reference frame?
6. Which format I should have a reference frame ? when I am adding to this all the rest of the panels? Shall I use a stretched file or without stretch ? I have use a fit format without stretch. but maybe this also has a meaning here as rest of the panels where in tiff format and maybe this is the reason of the error which I encountered while trying to put all this together. Error message you may see also below.

I would be very grateful for your answers on above as otherwise will be difficult to go forward.
Thanks a lot in advance.
Waiting to hearing from you soon 🙂

Hi Marcin @rosma,

First of all, that error is an indication that APP was not able to register all the frames at that try.

Please run it again with our latest beta25, that will give a better error message which should help me solve that bug 😉 much faster if it happens again.

You can download 2.0.0-beta25 here: https://www.astropixelprocessor.com/downloads/

1) Why the milky way projection now in not put accros entire photo width but only half of the photo width ?

How large is the field of view here? Is it more than 180 degrees in width? I suspect so?

2) Can I somehow identify a panel which is the reason of having the projection only for the half of the photo width? to take it out and doing it again

I don't think that is affecting this. A single panel will not cause this, it is likely the width of your field of view in actual degrees I would think.

3) Why the stars above the milky way seems to be completely incorrect ? I mean they seems to be swollen but only above the milky way. 

This will be caused by the chosen/set Center of Projection, you must have set it very close to the horizon. then the stars at the top of the field of view are maybe almost 180 degrees away ? the stars look like this because of the projection .

4) What do you exactly mean by fixed frame and fixed COP? and 5)How I should fix reference frame? 

You need to know which frame whould be the center of your field of view. By looking at this image, I would choose a panel that is in the center of this weird image. Set that frame as reference and run the mosaic registration. When it finishes, Have a look at the reference frame  with the registered image viewer mode. you can set this with the dropdown box above the image viewer, or load it registered with the right mouse click menu on the frame list.

Then also have a look at some of the panels with the milky way visible. Then tune the Center of Projection with this reference frame in such a way that you get what you aim to achieve. and not move the Center of projection to the horizon again, but probably to the east or west from the reference frame. Then remember the reference frame and the set COP, keep those fixed and run the entire mosaic first with a small scale factor in 6) integrate, then it will complete much faster and will give you a preview of how it will look 😉

6)  Which format I should have a reference frame ?

It is always better to run any integration/registration on NOT previously stretched data. The linear data will be much better. If the data is stretched, the star centroid calculations will be less precise and image normalization, MBB and LNC will work not as expected. All these techniques assume linear data. Once data is stretched, statistics and calculations on the data make no sense, really 😉

Let me know if this helps 😉

Mabula

Hi Mabula,

I am getting back with feedback after multiple test which I have performed last days.
Unfortunately result is still not satisfactory therefore I have decided to upload all the panels into your serwer and I would be very grateful if you could have a try to put them all together and let me know what is the end result. You may find them in the folder name: Marcin44panelsmosaic. There are 44 panels in total. Each consists of 2 frames. Each panel was registered and integrated already in APP as well. Final output saved as FITS without any stretch.

Below summary of what I have done so far with some print screens.
In short I have managed to test both mercator and equirectangular projection with multiple settings with modification of COP verticals and horizontals. The best result I have manage to receive in equirectangular projection with all the panels together at once ( not a milky way first and then additional panels ) . This you may see also below, however stars above the milky way was still a bit incorrect so I was not able to save this project and use this in the final photo. I think I am very close to desired effect here but I can not make it better. 
I have tried also multiple approaches like: first only panels with milky way shape with fixed reference panel and than the rest. This did not work properly and I am afraid I have done something wrong maybe. Anyhow no approach with expected result. Below you may find my workflow:

1. First i am registering all the panels as I am not able to set a frame as reference without registration. I understood from your previous mail this should be possible but I did not manage to do this therefore first I registered all the panels.
2. Than I am setting one panel as reference and doing one more time registration and integration. It is not possible to set a panel as reference without registration all the panels. Unless I am doing it wrong.
3. All of this I am doing as integration scale for 0.1 to speed up the result as much as possible. Anyhow even with integration scale 0.1 it takes around 2 to 3 hours to see the result.

I have done this multiple time like I mentioned in description both for mercator and equirectangular projection, modifying COP vertical and horizontal for + and - to find an expected result. However none of these test ( I have done around 12 or maybe 14 ) gave me a desired result. the closest was equirectangular projection with vertical COP modification as -40 with all panels at once. See below attached photo however still above the milky way in the area of Rho Ophiuhi I could find incorrect stars like with tracking.

I am still not 100% sure what you exactly mean by this statement " Set that frame as reference and run the mosaic registration. When it finishes, Have a look at the reference frame with the registered image viewer mode. you can set this with the dropdown box above the image viewer, or load it registered with the right mouse click menu on the frame list.Then also have a look at some of the panels with the milky way visible." I can only than set the reference frame after fist registration . Than I need to integrate entire milky way in order to see projection. I understand where I can choose from image viewer "registered" but I am not sure what is added value here when I still need to integrate all panels in order to the change in the COP setting. Anyhow all the below attached pictures are from equirectangular projection. 

I would be very grateful if you could have a try with my files and what is more important for me to understand how you did it. Maybe would be possible to describe workflow in a couple of steps including print screens from tool settings. I am not sure how many people are doing so big panoramas with 40mm lens but this would certainly help all of them. LIke mentioned previous panorama with 14 panels was piece of cake in contrast to this one for which I have spent so far increadible amount of time without success yet 😀

Thank you very much in advance for your help

This is example of the fact that there is no possibility to set frame as reference before registration

Place where I can choose "registered" from the image viewer

This one was the closest to expected result but some stars in RhoOphiuhi seems to be incorrect like mentioned above

Below couple of example of settings I have used 

Hi Marcin @rosma,

Thank you very much for your upload, please give me a couple of days before I can give more feedback after testing your data 😉

This is example of the fact that there is no possibility to set frame as reference before registration

Did you try to set the reference frame before registration after 3) Analyse stars? That really should work. So first analyse the stars which is required to get a reference in the first place. Then set the reference manually and then proceed to 4) Register . I will test, but this should work 😉

Mabula

Hi Mabula, did you have a chance to test the data I have shared with you ? 

Hi Marcin,

No not yet, will try to do that this week 😉 I have it on my todo list for sure !

Mabula

@mabula-admin Hi Mabula, any update ?

Hi Marcin @rosma,

Please accept my apologies, not yet. Will try to make it up to you soon.

Mabula

Hi Mabula, it's been a month and a half since I have uploaded data for your server. Is there any change you can have a look at this ? or shall I forget about this ? Thanks a lot for your response. 

Marcin

Hi Mabula, any chance to finalize this topic ? 

Hi Mabula, any chance to finalize this topic ?