Phil Hart
03-10-2010, 10:59 PM
this image is probably the greatest saga of my seventeen year astrophotography career. it's certainly the longest project in terms of number of nights and hours of exposure and processing time.
it started as an already hopelessly optimistic four part mosaic but three clear nights at SPSP in May and 20 hours exposure later i was off to a good start. but when i started processing, i realised i'd squeezed the overlap between the four frames down too much (and didn't shoot accurately enough) and had severe gradients across all LRGB frames. it was beyond my skill to process effectively (although marc pulled some rabbits out of hats with PixInsight to show the potential of my Luminance data which certainly helped me persevere :thumbsup:).
rather than giving up which would have been the wise course of action, i then tried to scrape together another four frames across the seams - turning a four frame mosaic into an eight frame very innefficient mess.
and then the weather turned to s^&t. :mad2: it took three mainly cloudy nights at heathcote in june, then another promising night that didn't deliver in july to scrape together another 10 hours but i was still missing one part. on the way to parkes a week later i camped in the barmah state forest and ran the ccd/laptop off battery power for the first time to get another 3.5 hours and complete a bare minimum set of LRGB frames to fix the mosaic.
when i finally found time to start trying to compile it all, the gradients were still horrendous and several processing attempts led to dead ends. on several occasions i thought that all the data would come to nothing more than an ugly mess and even more often thought that the 34 hours exposure time would have been better spent doing something (anything!) else.
however, the improved photomerge facility in photoshop cs5 is quite impressive. initially i tried a lot of fancy gradient masks to manually stitch this without getting close to a decent overall result. but with a little bit of stretching so it could 'see' enough in the images, photomerge was able to automatically form very decent panormas for each of the sets of L/R/G/B frames (but trying to stitch already combined RGB frames was still a lost cause).
so there's a few quirky colours going on as a result of photoshop trying to correct all the gradients but the end result is pleasing enough, although definitely not worth the amount that has been put into it!
here's the first published version, but i expect to be reprocessing this for years to come: Rho Ophiuchi Mosaic Saga (http://philhart.com/files/RhoOphTake3_rz.jpg)
Takahasi Epsilon 160 (530mm f3.3) & QHY9 CCD.
Seven nights, three locations, two states and 33.5 hours total exposure.
there's a lot of light scatter from antares that i can't do much more about but happy to hear comments on where i should direct efforts in the next processing attempt! does it look too red/too green or too blue to your eyes? (i've got questions for another day about G2V star calibration giving non-linear results).
so what i have learnt..
learn to crawl before trying to run a marathon with a mono CCD
don't skimp on overlap on mosaics (20% rather than 10%)
increase minimum altitude for capture from 30 to 40 degrees if at all possible
shoot RGB frames as high as possible
single row panoramas (eg 2*1 or 3*1) are much easier to blend than multi-row panoramas (eg 2*2 or greater)
and i try not to think about the fact that if i had a KAF-16803 sensor (and a scope to match) i could probably have waltzed this in with a single 5 hour frame!!
at least now i feel i can make an appearance at IISAC and not have to explain what happened to the four nights of imaging i did at SPSP back in May!
cheers
Phil
it started as an already hopelessly optimistic four part mosaic but three clear nights at SPSP in May and 20 hours exposure later i was off to a good start. but when i started processing, i realised i'd squeezed the overlap between the four frames down too much (and didn't shoot accurately enough) and had severe gradients across all LRGB frames. it was beyond my skill to process effectively (although marc pulled some rabbits out of hats with PixInsight to show the potential of my Luminance data which certainly helped me persevere :thumbsup:).
rather than giving up which would have been the wise course of action, i then tried to scrape together another four frames across the seams - turning a four frame mosaic into an eight frame very innefficient mess.
and then the weather turned to s^&t. :mad2: it took three mainly cloudy nights at heathcote in june, then another promising night that didn't deliver in july to scrape together another 10 hours but i was still missing one part. on the way to parkes a week later i camped in the barmah state forest and ran the ccd/laptop off battery power for the first time to get another 3.5 hours and complete a bare minimum set of LRGB frames to fix the mosaic.
when i finally found time to start trying to compile it all, the gradients were still horrendous and several processing attempts led to dead ends. on several occasions i thought that all the data would come to nothing more than an ugly mess and even more often thought that the 34 hours exposure time would have been better spent doing something (anything!) else.
however, the improved photomerge facility in photoshop cs5 is quite impressive. initially i tried a lot of fancy gradient masks to manually stitch this without getting close to a decent overall result. but with a little bit of stretching so it could 'see' enough in the images, photomerge was able to automatically form very decent panormas for each of the sets of L/R/G/B frames (but trying to stitch already combined RGB frames was still a lost cause).
so there's a few quirky colours going on as a result of photoshop trying to correct all the gradients but the end result is pleasing enough, although definitely not worth the amount that has been put into it!
here's the first published version, but i expect to be reprocessing this for years to come: Rho Ophiuchi Mosaic Saga (http://philhart.com/files/RhoOphTake3_rz.jpg)
Takahasi Epsilon 160 (530mm f3.3) & QHY9 CCD.
Seven nights, three locations, two states and 33.5 hours total exposure.
there's a lot of light scatter from antares that i can't do much more about but happy to hear comments on where i should direct efforts in the next processing attempt! does it look too red/too green or too blue to your eyes? (i've got questions for another day about G2V star calibration giving non-linear results).
so what i have learnt..
learn to crawl before trying to run a marathon with a mono CCD
don't skimp on overlap on mosaics (20% rather than 10%)
increase minimum altitude for capture from 30 to 40 degrees if at all possible
shoot RGB frames as high as possible
single row panoramas (eg 2*1 or 3*1) are much easier to blend than multi-row panoramas (eg 2*2 or greater)
and i try not to think about the fact that if i had a KAF-16803 sensor (and a scope to match) i could probably have waltzed this in with a single 5 hour frame!!
at least now i feel i can make an appearance at IISAC and not have to explain what happened to the four nights of imaging i did at SPSP back in May!
cheers
Phil