Here's my final attempt at the solar corona. I promise not to do another repro... honest!
The attached version looks blotchy on my screen which I think is due to the compression that IIS does to uploaded images. The version here probably looks better: http://rickstevenson.smugmug.com/Ast...10b-med-X3.jpg
This image was created by stacking 171 subs captured with a Nikon D800E and 70-200mm lens @ 200mm on a fixed tripod at shutter speeds between 1/640 sec and 4/10 sec. Conditions were less than ideal with high cloud and the sun at a low altitude. Processing was done with PixInsight and Photoshop.
A few people have asked for details of the processing, so here's an overview of what I did. I'm not going to give a detailed recipe because I don't have time to write a long essay and also because different data will need different treatment. I'm happy to answer questions if anybody wants to try it...
Processing summary (all steps in PixInsight unless otherwise noted):
My subs were in batches of 9 bracketed images. I registered and summed each batch of bracketed images into a single image using the FFTRegistration script. This registered the images against the moon which is not ideal (the moon and the corona are moving relative to each other) but good enough for short sequences.
I registered the summed images using Photoshop and PixInsight. This step was a bit complicated as I wanted to register against the corona itself. I used the LarsonSekanina filter process to create copies of the first and last summed images with coronal detail. I used Photoshop to register these images (difference layer and manual pixel by pixel tweaking) and calculated how many pixels I had to move the last image in X and Y directions. Using this info and the timestamps in the EXIF data for my images, I calculated the transformation required for each of the intermediate summed images. Then I registered the original summed images in PixInsight using PixelMath to transform each image by the required delta X and Y. This effectively did what a tracking mount would have done automatically and effortlessly At the end of this process I summed all the images into a final stacked image.
I used the LarsonSekanina process to bring out the coronal details of the stacked image. Much tweaking of parameters!
Followed by lots of typical PixInsight magic: crop, slight stretch, luminance masked noise reduction, contrast tweaking with HDR Multiscale Transform and Local Histogram Equalization, more noise reduction, some careful sharpening with Multiscale Median Transform and a quick Curves Transformation to adjust brightness and contrast again.
Final steps in PixInsight were to use PixelMath to create a dark blue background and then add a scaled copy of the image to it.
I then exported the image to Photoshop and pasted a nice copy of the moon and prominences over the hole in the middle of the image.
Voila!
I hope that's helpful to somebody... I'd guess I probably spent 30 or 40 hours playing with this. Next eclipse it will be easy so long as I can remember what I did
Even if you're not interested in the processing I hope you enjoy the image and thanks for looking...
That is a spectacular image and thanks for the link to the hi-res version, it is heaps smoother and the tones look more natural compared to the compressed image. Well worth clicking on the smugmug link!
It has been quite illuminating to read about your processing marathon – truly these world class images require heroic efforts to present the data so fully and comprehensively.
Grand effort and a great overall result Rick! I enjoyed the process too
Thanks, Carlos!
Quote:
Originally Posted by iceman
Excellent result Rick, really good.
And thanks, Mike!
Quote:
Originally Posted by Dennis
Hi Rick
That is a spectacular image and thanks for the link to the hi-res version, it is heaps smoother and the tones look more natural compared to the compressed image. Well worth clicking on the smugmug link!
It has been quite illuminating to read about your processing marathon – truly these world class images require heroic efforts to present the data so fully and comprehensively.
Cheers
Dennis
Thank you very much for the kind words, Dennis! I know you put a lot of effort into your lovely time lapse too.
Quote:
Originally Posted by Paul Haese
Superb image Rick. I will need to go back over my image as a result of this splendid example.
Thanks, Paul. I got some inspiration from the other great corona images posted here including yours. It's a virtuous cycle
Quite possibly the best one I've seen yet, Rick! You might be setting new records for the "ratio of processing time to image capture time" in amateur astronomy!
Quite possibly the best one I've seen yet, Rick! You might be setting new records for the "ratio of processing time to image capture time" in amateur astronomy!
Thanks very much, Dave. I might be a bit slow but I get there in the end
Too bad I still haven't learnt how to use PixInsight!
What parameters on the Larson Sekanina filter did you end up with?
Phil
Thanks, Phil. It takes a bit of effort to learn PI but I reckon it is worth it.
That image was based on summing multiple LS applications. I generally used a Radial Increment of 1.5, Angular Increments between 0.8 and 2.5, and Amounts between 0.25 and 0.75. All other parameters at default values (apart from the X and Y center parameters, of course).
Thanks, H! It's quite amazing to see all that detail pop out of subs that show very little coronal structure at all. The Larson-Sekanina filter is magic. It was originally developed for showing structure in comets.
At the end of this process I summed all the images into a final stacked image.
