View Full Version here: : Imaging mosaics
08-06-2012, 07:57 PM
Do you create image mosaics of the larger deep sky objects? If you have an imaging setup that delivers only a small field of view the chances are that you do, or that you have at least tried to. If you do, how have you got on with positioning your camera so that each image in the mosaic precisely lines up with its adjacent images and has sufficient, but not too much, overlap? My experience is that that can be a real challenge and I usually end up with a ragged outline that needs to be severely trimmed back and I often unwittingly leave gaps that need to be filled in later. Being unable to find a tool that could help me guide my camera to exactly the right spot to image each frame I decided to create my own. I have now finished developing an application that works in conjunction with a sky atlas. The application allows you to design and then use a mosaic of imaging frames suited to your imaging camera and optical train. The mosaic floats over the sky atlas display and provides an exact frame of reference for positioning your imaging camera for each image.
Currently the application works only on Windows platforms and only with a sky atlas that allows you to display the sky in equatorial mode i.e. with lines of declination running horizontally across the screen and lines of right ascension running vertically.
I am looking for a few persons who are willing to trial the application. They are under no obligation to provide feedback although obviously I will welcome receiving it. In particular I would like to have their views as to the applicaton's general fitness for purpose, to learn of any detected bugs and to get any suggestions for improvement that they may care to make. And if they do provide feedback I will make sure that they receive upgrades that correct detected bugs and implement accepted suggestions for improvement.
You will get some idea of the application by looking at the attached screenshots.
If you would like to participate please send an email to firstname.lastname@example.org stating your name and a return email address. I will then arrange for the application files to be sent to you.
09-06-2012, 10:46 AM
I create a mosaic using the Mosaic tool in TheSky6 with a 5-10% overlap between subframes.
I then use Maxim DL's telescope control along with the builtin PinPoint LE for precise positioning and camera rotation.
Have successfully managed to create mosaics imaging over several nights a year apart, even! Refer to my Rho Ophiuchus thread in the deep space sub-forum.
09-06-2012, 04:21 PM
Thanks for the update Octane. It's good to learn that TheSky at least provides a mosaic feature.
09-06-2012, 05:04 PM
Would this work with very wide fields, e.g. 10x15 degree FOV with my DSLR and 135 mm lens?
I'd personally also be interested in being able to plan the exact time of day (possibly over several nights) when each pane of the mosaic should be captured so that imaging is done at the highest altitude possible (i.e. near/on the meridian).
Ideally, I'd like to be able to look at a sky chart, draw/lasso the area that I want to image, provide the camera/lens FOV and a range of dates/times of when imaging would be done, the duration to spend imaging each panel, and have the software dump out a list of panels with RA/Dec coordinates and imaging times.
I don't know if there's any free software that can do this already, so I'll probably be doing it by hand...
09-06-2012, 08:24 PM
Mosaics are quite difficult.
Overlap by 20%.
Its best to image the same time and position in the sky each night rather than several panels in a night at different angles in the sky. They will look different.
Stitching is tricky and blends difficult. There are several approaches. Pix Insight had the edge at one stage but its not a particularly user friendly program but they do have video tutorials which is a big plus.
Photoshop CS4 and above has a pretty decent stitching program. Below that its pretty ordinary.
I am planning to use PTGui Pro to stitch my next mosaic as it does a great job on terrestial images. Hopefully it will on astro images.
Tony Hallas instructional video has a section on doing and processing mosaics. Its somewhat helpful. Nevertheless a tricky proposition in both acquisition and processing.
Also label everything very well. I did a 16 panel mosaic and spent hours trying to work out where one or two non descript panel fitted in to the jigsaw puzzle!
Try to do the same exposures etc on each.
Prepare to spend a lot of time in processing and learning.
I found taking a 10 minute luminance and then loading the luminances of the other panels into Photoshop CS4 before committing to a night of imaging a panel worked well. You can see if the panels overlap on all sides well enough and don't leave gaps that will defeat the mosaic.
09-06-2012, 08:38 PM
Dave, so far as building a mosaic goes there is no essential difference between mosaic frames measuring only a few arcminutes across and those measuring 10 to 15 degrees. It is simply a matter of scale.
Attached is a 4 x 4 matrix superimposed on the Orion region and centered on the Flame Nebula. It uses an imaging frame that measures 10 degrees in declination. Because I don't know the shape of your DSLR's imaging frame I have assumed one that is 1.38 times as wide as it is high. An overlap between frames of 15% has been used. You can probably see that each frame is labelled with the coordinates at its center. For example, the coordinates for frame 1 are RA 04:31:12.8 and Dec. 10:54:0. Assuming that the camera frame shape is correct, so long as you aim your camera at those coordinates you will image the area bounded by frame 1. Your camera's live view should display the field stars contained within the frame's borders.
As to the timing of the image, your sky atlas program may be able to tell you that. For example, the one that I use indicates that star HD 28608 is very close to the center of frame 1 and at 10 o'clock at night that star is best imaged next in mid-December at the dark of the Moon when it will reach an altitude of nearly 45 degrees. The sky atlas also provides guidance as to exposure time.
09-06-2012, 08:52 PM
Thanks for all the tips, Greg. I read Rogelio Bernal Andreo's AIC 2010 presentation on making widefield mosaics using PixInsight in earnest... I like a good challenge!
The biggest challenge I've found with mosaics of very wide fields is that the field for each panel isn't flat due to the 3D-to-2D projection of the celestial sphere... usually requires panels to be either re-projected, or inaccurately registered.
Perhaps the ideal solution for perfectly registered ultra-wide field mosaics would be to follow Nick Risinger's approach (http://skysurvey.org/) of plate solving each panel and re-projecting it to the final mosaic projection. There's a lot more work involved though...
09-06-2012, 09:06 PM
Yep, it's easy to work out the best timings for an individual panel by hand. However, let's say I'm planning a larger mosaic:
* 4x6 panels - each 10x15 degrees,
* 25-30 mins of data per panel,
* shot over two nights in Winter between 9 pm and 3 am, and
* keeping all imaging as close to the meridian wherever possible (i.e. +/- 7.5 degrees).
It would be a pain to work out which panels to take, when, and where...
Anyway, just a thought experiment! I'll stop derailing your thread now :)
10-06-2012, 07:42 AM
That's why I am thinking that a program like PTGui Pro may be the go as it does all that automatically.
I am about to process a 4 panel mosaic and use it to do the stitching so I can let you know.
As I say Photoshop CS4 and above does a reasonable job but I think its still not as good as purpose built software.
My workflow would be like this:
Plan out the mosaic and allow plenty of overlap (20% on all sides).
Take each panel at similar times over several nights (hard with weather moon) Or take 2 in one night (it will mean more gradients to handle).
Take same exposures, same binning etc etc.
Do a modest mosaic - 2 to 4 panels. 24 panels would be for an expert and doomed to fail. I would take a 10 minute luminance and save it for each panel. I would do a quick and dirty Photoshop CS4 stitching of the 4 luminance panels to make sure they fit before committing to an imaging run. I also open up the luminance file of the last panel and compare on the same screen to the focus image that I am about to take to make sure they are aligned and will work. Label them well so you know which one goes where - important! (not with 2 panels but do 12 and you can easily get lost).
Process each exactly the same way. Don't do different steps to each.
Don't worry about the final colour, noise control etc etc leave that until after it is stitched.
Stitch the panels using PTGui Pro or CS4.
See how it looks. It may need gradient handling on each panel, some curves to balance each one, perhaps try normalising the backgrounds of each.
Once successfully stitched then do your final colour tweaking, sharpening, noise control etc etc.
The hardest part of mosaics I have found is the imbalance in brightness levels between panels so when blended one will stand out. Also colour hues can shift significantly between panels so the Photoshop selective colour tool (especially tweaking "neutrals" s vital.
10-06-2012, 05:58 PM
I find stitching wide angle mosaics needs a program that can handle the lens distortion. For example Kolor's AutoPano Pro or Giga which you can tell the f-ratio and focal length so it can calculate distortion matricies.
The usual astro programs can fail totally at lengths like 18mm. For example, Nebulosity turns a series if frames into a mixture of stars and varying length trails depending on how far they are from the ones chosen to to the alignment.
With longer focal lengths - say beyond 200mm - the distortion is minimal and just about any stitching program will work.
vBulletin® v3.8.7, Copyright ©2000-2013, vBulletin Solutions, Inc.