Between Sagittarius, Ophiuchus and Scorpius
Click here for a larger version
The mosaic above is made out of 52 frames. If we take out the 12 frames from the already mosaic I had done of the Rho Op area, that leaves the 40 frames I captured and processed this month of June alone. Each frame is 3x5 minutes of L and 3x3 minutes each RGB, all bin 2x2.
I captured the data in 9 different nights: 6 outside of the DARC Observatory, 2 during a camping weekend in Plettstone (Bear Valley), and one additional frame catured at the back of the Lick Observatory. All during a period of around 12 days. Um, yes... I do have a DAY job too!!
As I mentioned earlier, the total round trip driving time to these dark sites added up to 1,840 miles driven. Total exposure time for the 40 frames is around 40 hours (48 if we count the Rho Op area) with a time in the field exceeding over 70 hours. The original image, over 18,000 pixels wide, can be provided upon request. It's just too big in size I'd rather not to pay the bandwidth toll (yes I pay for my bandwidth).
People familiar with this area will soon realize the image is upside down, and that's my favored composition. It gives me a stronger feeling as if the nebulosity in the Rho Op area is "escaping" from the Milky Way, although for those who must see it north up, I have a north-is-up 4000x2170 version here (2.1mb).
PROCESSING
A few comments about the processing, in case anyone's interested in the insanity involved in building a 52 frames mosaic and try to get it processed in barely 4 days.
Making the 52 processed luminance frames to match seamlessly wasn't easy. I had to give up three different methodologies, the first one being an effort of over 10 hours that I finally discarded. Finally, with the lightness data more or less registered seamlessly, doing the same with the color made the previous work on the luminance data look so easy! Throughout the whole process I used PixInsight, Registar and Photoshop, each of them with their advantages and disadvantages.
Mosaic'ing all red frames, then all green frames then all blue frames over the lightness mosaic didn't work well. You need a 100% perfect result on each channel in order for them to match later, Registar generated way off R, G and B registered frames. PixInsight was much more accurate (also much slower to produce) but still not perfect across the 40 frames (and the moment one channel frame is off, it;s all useless), so I registered and aligned each master R-G-B frame separately, gradient correction (remember, each frame is 5.5x3.5 degrees, which is going to generate much more severe gradients than on smaller FOVs), color balance, etc... 40 times. All this was done with PixInsight which has proven to me its registration process is almost flawless and better than anything else I've tried - including a trial version of CCDStack v2. Making the processed 40 RGB frames to register with the lightness data wasn't hard, but still a work of patience, having to do one by one, etc.
Some "purists" may criticize this methodology, as it requires the RGB data to be registered twice: first for each frame, then over to the (lightness) mosaic. All I can say is... Let me know how it goes when you process your own 52 frames mosaic using more "rational" approaches. And I'm not being entirely sarcastic here - I would really like to know how it goes! I did what I could make it work, and out of the different approaches I tried, this is the only one that worked across the entire 40 frames, and it's because of that I can now say that this mosaic has a nearly perfectly seamless lightness data - no funky stars around the seam areas, and each tiny little star tat shows color it's because the RGB channel data gave that color to that tiny star and such data matched 100% accurately over the lightness data. No other methodology I tried gave me that over the entire field. Maybe I goofeed off, I don't know...
Anyway... But of course, once I had the RGB mosaic nicely aligned, despite it matched perfectly with the lightness data, the color from frame to frame was showing quite some differences. How did I process the color data to generate an almost seamless transition? Basically by doing individual adjustments one frame at a time, repeat, repeat, etc. As if it wasn't enough having processed 40 frames individually, I just couldn't see having the color to match seamlessly.. Because most gradients had already been taken care of, color information was uniform, so all adjustments were mainly applied over each entire frame without the need to create individual masks.
In the end, other than 3-4 frames for which the original data was pretty bad and there wasn't much I could do to save them, I managed to get things to match more or less seamlessly.
With all that done, and the L and RGB composites done, the final processing on both was very simple. I know I could improve it if I work on it a bit more, but I'm heading for a trip and I wanted to post what I've got so far, which, with all its defects and so on, it's no small effort. I hope you like it!
Get a poster, t-shirt, mug, mousepad... with this image!
Comments
Yeah I like It! humongous work !looking forward to seeing another 52 panels added! hehe
Cheers
Louie
Como no puedo leer el pie de foto por el idioma, pues solo hago las comparaciones de lo que veo en una foto con la otra, sé que fue tomada en la misma área de la anterior, asi es?
Me podrías aclarar un poquito, por favor.
Gracias, la foto es muy linda.
Gratulation.
But you work very hard!
Regards
Robert
I'll be checking back for more...