Taurus to Cygnus 15 Panel Mosaic

[msmythers]

Hi,

I thought I'd share a very poor quality image that I think is very good for what it is. A 15 panel mosaic that covers close to 150 degrees of the sky. While the images and the capture of the images are mostly poor the sum of them is very different. I used images from 4 different camera lenses. A 20, 45, 90 and 135mm lens on a Sony Nex-5 mirrorless camera. The images were taken over the last year and a half or so. I live on the east coast of Florida with heavy light pollution, very poor skies due to high temps and humidity. None of the images were taken with exposures longer then 30 seconds. They all varied in total exposure time. The panel with the Pleiades and the California nebula was the longest at 1.4 hours of total exposure time and completely overhead.

Because of the wide variation in image quality and saturation levels I did this mosaic with processed images. Completely non-linear images. I matched(as close as possible) brightness levels with Statistics and HistogramTransformation. I had to resample every image because of the final mosiac size and my computer locking up. 10937x4645 is the processed size and one half of what the actual size could have been. I ran all panels through the wonderful Image Solver script and was able with much trial and error get each one to solve. I used Mosaics by Coordinates(can't do this without it) and then GradientMergeMosaic to bring them together. 3 excellent pieces of code worth the price of admission. This was the best work I've ever done with PixInsight and the most fun.

Again while not a great image, not a great registration in areas but it is something very few people have done from there front yard.

http://www.astrobin.com/107815/  Image Only

http://www.astrobin.com/107816/0/ Image annotated


Mike


 

[mads0100]

Dude, that's awesome. Thank you for sharing it!  How did you combine it all together?

[msmythers]

Hi Chris,

The process is easy but really computer intensive. The images that MosaicByCoordinates created were then used in GradientMergeMosaic. Those 2 steps are the easy part. The hard part is image solving non-linear extremely wide images that are very low resolution. Even that was quick once I found the right settings in the ImageSolver script.

Now I have a living image that can grow with more panels.


Mike

[cs_pixinsight]

This is awesome.  Great job with such minimal capture time on the individual frames.  I've been working on a similar project for about 2.5 years now.  I have captured the Milky way from Lupus in the summer to just south of Canis Major in the winter using a 50mm lens on Canon cameras.  Now I just have to find a workflow that will put it all together with good blending and equalization between frames.

I'd love to get much more details on how you solved your issues with the ImageSolver script for these very wide angle images.  In fact, knowing your workflow from beginning to end would help tremendously as I'm sure I'll have to solve all of the same issues you just experienced.

Again, terrific image.
Craig

[msmythers]

Craig,

Sounds like you have a good amount of data for a very nice mosaic.

Any wide field image with such a low resolution lens, focus will be critical. For both a good image and for the image solving. When I find focus with my camera lens for the night I tape the focus ring to the lens body. I use manual lenses. This prevent any chance of a mishap during the nights session. I also use a dew heater and large sun shade. I hate surprises. 

Since my image captures are so different in quality it's hard to give you a good detailed workflow. Each image required very different parameters. The best suggestion I can give you is the normal start of DynamicCrop,ABE or DBE. Do a color calibration how ever you want. Then its noise reduction and stretching. I like the new MLT and then MaskStretch. From there it is really image dependent with wide images. That is if you have variable sky conditions. I used ExponentialTransformation and the Background Enhance script in many of my images.

If you can build your mosaic from images of similar qualities then start with crop/ABE-DBE and ColorCal. Then build your mosaic with the linear images. This will allow GradientMergeMosaic to do all your blending. At this point you can then process the whole mosaic for a better quality image. I could not do this so I manually used Statistics and HST to get as even of a background appearance as possible across all panels.

In ImageSolver with processed, wide field, non-linear images I start with a magnitude limit of 8 and if needed raise it. I use Bright Stars and Polygons. Make sure you have your pixel size and focal distance set. I always check that they are correct. If you resampled your image then you need to change one of these settings to follow the resampling. You might also need to change the Star sensitivity setting. Once you have your image solved check it with the AnnotateImage script and the Constellation Lines. If the registration is off try ImageSolving again with Triangles. I've had a few very wide images where this makes a difference.

I missed an important bit of information When solving wide images. Try to pick an object close to the center of your field of view for the object search. Or if you know the coordinates of the center of your image use that.

With patience I haven't had many images that ImageSolver can't handle as long as the resolution isn't too low. The other option is to use CatalogStarGenerator and ManualImageSolver.

I hope this helps.


Mike

[cs_pixinsight]

Thanks for the info on ImageSolver Mike.  I'll try your settings as a starting point and go from there.  If all else fails I'll just match the frames a close as possible and use standard panorama making tools to do the job.

Craig