1 Gigapixel image Milky Way [Finished]

[Atmos]

This is my first, albeit brief process of my 10x10 panel mosaic. I loaded it into Photoshop (a program of which I have NO experience with) and since come to realise that I need to clean my computer out a bit. Apparently editing a 16.5gb file uses a lot of HD space that I don't currently have :P

So here is a screen shot of my 5 minute edit :) Processing wise, I need to figure out a way of getting it into Pix Insight :P Only problem is that most file formats don't allow 16gb files! Currently have it has a .psd and cannot save it (not enough HD space [36gb available]) so it is a work in progress.

It is a 47109x28477 pixel image at 2.6"/pixel with a OSC (Nikon D700).

Finished Version (http://www.astrobin.com/full/268010/0/?nc=AtmosFearIC)

[RickS]

You'll probably break a lot of software with a file like that, Colin :lol:

[Atmos]

PS handles it fine, I just don't have any experience with it!
Managed to save a 311mp TIFF for PixInsight though and had some fun with that. Have done some serious down sampling though. Gone from 1.34 gigapixels to 20mp :P

I do still have plans of ATTEMPTING to export a full res file to PI (if possible) but not sure if it'll happen any time soon.

For anyone that wants a high res (http://www.astrobin.com/267364/), go view. Having been down sampled SO MUCH, it really has lost its immersiveness :/

[RobF]

Very nicely done Colin. Serious FOV there! :thumbsup:

[spiezzy]

that is Fantastic Colin very deep image and love the colour in the second one I can see a couple satellites in it which adds to it showed my wife and she couldn't believe the was so much colour in the milky way thanks for sharing
cheers Pete :thanx::eyepop:

[Atmos]

Thanks Rob. The FOV is similar to what you'd get with a 50mm lens as a rough comparison.



Thanks Pete. I didn't really want the satellites :P Working with single 60s exposures there isn't any rejection that I can do. Down on the lower area there is some redder and bluer banding, this is a consequence of shorter exposures. My D700 does not like NOT being sky limited, only 3.9e- read noise but it is a very noisy read out.

[IanP]

Nicely done, Colin :thumbsup: :thumbsup: :thumbsup:

PixInsight has a process called CloneStamp, carefully used, can help you to almost remove traces (satellites) ...
:hi:

[Regulus]

Oh my! That does look nice.
Amazing job Colin, and I really hope you can sort the PC out, because I really, really, want to see your end product.

Trevor

[gregbradley]

Photoshop is not the best mosaic program. I can see where PS has blended in the panels rather poorly. It seems to blend panels in with a gradient tool or something and its a bit not so good. A dedicated program like PT Gui Pro or Microsoft ICE (free) will leave PS for dead.

Also make sure each panel is processed the same way especially curves and levels. I would do the bulk of the final processing after the stitch and not before.

Photoshop cc 2015.5 has an auto background fill in when the stitched image is not square. That may be fine for daytime shots but for astro its probably a no go.

Try a free trial of PT Gui Pro (its faster than other pgms as well). I would also make the files smaller to make it easier to process. A neatly stitched image is better than a large but not neatly stitched pano.

Greg.

[Atmos]

I'll have to give PTGUI a looksy, I think I may have an older version (6 years at least) laying around somewhere.

A part of the issue I'm having is that there is some nasty noise left over by the camera. At ISO1600 the camera has 3.9e- read noise which under dark skies at F/5 means I need 120s to get above the read noise.

When I FIRST started in AP with my 10" LX200 I was doing 6s exposures and for a long time I thought I was actually dealing with electrical interference either from the power lines out the front of the house or even the WiFi signal. A large sine wave passing through the image. This is still noticeable at 30s at F/5 but at 60s it is just a discoloured banding. At 120s it is swamped by sky glow so it isn't noticeable.

The attached picture is the lower left corner (my first frame for the night). Areas where there is a lot going on it isn't noticeable, darker areas however have this banding which at 60s is a good proportion of the frames. Most of the stitching is pretty good, there are a handful of BAD spots but the bigger problem seems to be the noise in the frames themselves :/ Have tried the BackgroundExtraction in PI and it helps on frames like the one below but will likely cause more issues on the higher signal areas.

[SimmoW]

Looking ace! Yes id be interested how processing such a large file after mosaicing would go, performance wise

[Atmos]

What I've done above has largely been processed before mosaicing but I have just started toying with stitching and then processing. Doing it this way certainly seems to allow for a more intense process.

The problem that I am running into however is that creating such a large mosaic in PI is a struggle unless I downsize each image by 50% to begin with, as I have done in the attached image. What I have just noticed is that it is causing a lot of blue haloing around the blue stars in a couple of the star clusters. I know this isn't caused by the telescope as it isn't non downsized images. Certainly allows for a more dynamic stretch though.

[multiweb]

That's a huge undertaking Colin. :thumbsup:

[plantnerd]

Nice work collecting all that data Colin it is a most magnificent region for an ultra high definition mosaic. It seems like you have some banding issues to deal with. I have some banding in my images also but I have found turning the camera vertical or on the diagonal and reshooting over the same area tends to solve this as the different directions of banding cancels out the other banding.

It certainly is a challenge for your software and hardware I have had ICE deliver an error message saying I require 2TB of free space to save the image when all the source images are not even 0.5TB.

[Atmos]

They don't looks as neat as your FSQ mosaics Marc :P



I have found the banding issues are a consequence of exposures being too short as I haven't noticed them at 120s. No possibility to reshoot this year though I don't believe. Have you found them with longer exposures? I haven't done enough testing to be sure yet.

Your undertaking is a bit more ambitious than mine! I managed all of mine in just under 2 hours and the mount did all the tracking and pointing. Luckily what I have is not going to be as demanding as your size right now. Takes about 10 minutes or so for my laptop to output a full res 1.34gigapixel PSD file. All of mine files only take up a couple of GB.

[Atmos]

I think I am about as finished as I am going to get for a while, been working on it every day for over a week and my eyes don't want to see it anymore :P

I have dropped the first two panels as I just couldn't get them to match, may have still been that fraction too early when I started or it may also just be that the subs aren't long enough that far from the bulge. So now it is a 10x8 panel mosaic. Or a nice and neat Gigapixel image.

"High Res" on Astrobin (http://www.astrobin.com/full/268010/0/?nc=AtmosFearIC) being about 14mp or a little more than what my D700 does with an approximate FOV of a 50mm lens.

After a break I'll probably go and remove the satellite trails in PS and then get this printed :)

Any suggestions for the mandatory reprocess in a few months time?

[Placidus]

Colin that is a work of art, a thing of beauty, and a technical masterpiece.

The mosaic is seamless and beautiful.

It'll take ages to explore. We quickly found pipe, Lagoon/Bigfoot, Trifid, Lobster, Cat's Paw, three asteroids. Would you consider an annotated version as your retirement project?

The pipe nebula at that resolution is quite wonderful - too used to seeing it as a general blur.

Not enough room on this line for all the :thumbsup: symbols.

[multiweb]

The finished product looks terrific Colin. Well done. :thumbsup::thumbsup::thumbsup:

[rcheshire]

That is superb Colin. What a spectacular view!

[Atmos]

Thanks Mike and Trish, I am really happy with how far it has come. Taken near a dozen attempts to figure out a workable work flow.

I was actually hoping that Astrobin may be able to do its astrometric solve and annotate it for me :P I'll give it an attempt tonight in PixInsight. Actually not familiar with the Pipe Nebula, have to look that one up :)



Thanks Marc, it has been some of your previous FSQ mosaics that gave me a bit of inspiration :P



Thanks, it is a bit better than what I have been able to achieve with a 24mm lens!

[multiweb]

Mate, the FSQ is easy. It's flat and easy to stitch with very minimal overlap. What you did with the amount of panels and the surface covered is nothing short of outstanding. I did a couple of very wide milkyway fields in the past and I can appreciate the distortion and work going into this as well as manipulating large files and having gradients to contend with.

Kudos to you. :thumbsup:

[Atmos]

I'm actually really happy with the correction of the Sky Rover, I put 10 frames into CCDInspector and came back with a result of 8.8% curvature with 0.1-0.2mm tilt (not noticeable in the corners) with a full frame DSLR. It is about the same level of correction I got with my previous Sky Rover + 0.8x Flattener with a KAF-8300 sensor.

[Octane]

Well done, mate. That's a huge undertaking and you've done well.

I'm yet to start building my observatory (hole for the pier has started to be dug out) and this is inspiring me to put my mint gear back into production.

H

[Atmos]

I'd so love to have even a semi permanent setup :) Looking forward to seeing yours when it is complete.

To those with a colour calibrated monitor, do I need to run another pass of green removal? It's showing up better on my laptop than my phone which shows it as needing more green cast removal. Should probably invest in a spider and a non-laptop screen :P

[strongmanmike]

Wow Col....the latest iteration is bloody great with excellent rich colour! Rather Cassewellesque even.. ;)

Only minor blemish is a faint green stripe across the centre and a cyan stripe bottom right corner....and maybe juuust a touch more magenta and blue might look good..?

Well done dude :thumbsup:

Mike

[atalas]

Very nice Colin:thumbsup:

[Atmos]

I am a bit staggered by the difference that a different processing method does! I was quite happy with the first rendition until I started processing it this last time. As I was processing each frame, especially towards the top, I was like "So that's what it was SUPPOSED to look like!"

When I was going through them I didn't want to overdo the green cast too much so I dialed it back, looks like it was a bit too much. I personally struggle to see the green issue until it's gone and the change in contrast kicks in; I have green colour issues :P I'll go through every frame and do some more green removal :P

Did some quick playing with the magenta and blue, going to have to think about how to do that without effecting the overall colour balance. Quick play with the curves made M8 & M20 look better but ruined the overall balance.

Not sure why but it looks like Rows 3-7 (I dropped the bottom two) through Column 7-10 have a bit of a blue cast in the blacks, getting worse as it goes down (earlier shots). Will test some blue background reduction when I get around to it :)



Thanks Louie :)

[Atmos]

Had an attempt at plate solving, has some issues. Due to all of the correction over the large mosaic, it really struggled and the accuracy wasn't the best. It was close but just not quite there all the way across the frame. No matter. The first shot shows that I REALLY shouldn't have annotated the Tycho catalog :P

The second shows what I had originally assumed to just being a distorted star near the edge of the frame or something.

[eskimo20]

Colin
That image reminds me of a comment my wife made, just this morning, when we discussing Life, the Universe and Everything. It went something like – “Are there really billions of stars in our galaxy?”


As a short-attention-span video astronomer, I salute you.

Robert

[alan meehan]

Well done Colin what a huge effort the finished version is just superb
AL

[rustigsmed]

looks fantastic Col - i would be stoked with that!

only minor gripe is some banding across the middle (red on top then blue/green underneath) then some perhaps on the bottom section. only mentioning it because it looks possible to take care of it with some processing?

still amazing shot.
Russ

edit: sorry didn't see others mention the same banding

[RobF]

Great work Colin! Had a lot of fun exploring the hi res :)

[graham.hobart]

That is so great mate!!!:thumbsup:

[Atmos]

Thanks Robert, it is when you look at the FULL res version that it becomes apparent just how many stars there are. The noise in the background of most of the frame isn't so much noise as it is poor SNR stars!



Thanks Alan, I may start working on addressing some of the issues next week. Need a bit of time away from it :)



I have been thinking about how to remove that banding, I have made some attempts but haven't had much success. It is embedded into the background of every frame and theoretically should be almost identical in every frame so I should be able to subtract it out. That is the cause of the blue stripes (the red bits don't have that banding, or maybe the red is the banding... not sure), the green is me not doing a heavy enough green subtraction during the processing. That one is easy fixed.



I'll look at putting up a HIGH res version. Gonna be big.



Thanks Graham :)

[dylan_odonnell]

Looks great! The dust details have come through amazingly after the star reduction. There is some horizontal red and green across the frame.. is that just an artefact from the mosaicking?

But yeh, stunning shot. Probably Malin worthy!

d

[Shiraz]

Colin, that is a wonderful image - great to just cruise around in. Thanks for posting it.

[Atmos]

Thanks Dylan. The stars have an average FWHM of ~8" (~3 pixels) which is tiny when compared so the fact that even with a Nikon D810 (36mp sensor) will likely have FWHM closer to 35-40" with a 85mm lens. If I wanted to be REALLY adventurous I could actually do some deconvolution but I am probably too lazy to do that on every frame :P

The green is something that I attribute to using a OSC and not doing enough green reduction during processing (I was trying not to do too much). There is a red/blue banding in the image that is a consequence of exposure times being too short. It is in the background of nearly every frame and I haven't quite figured out how to remove it. It is readout noise generated by the camera as it is the same in every frame (2 blue bands and 3 red bands). In higher signal areas it isn't as visible as it is swamped by signal.



You're welcome Ray, once I actually finish this I'll have to put up a 50% res or something for those dedicated pixel peepers :P

[Atmos]

For anyone interested in a near full res.

https://easyzoom.com/image/95587

I've dropped the resolution by half but it doesn't really effect image quality at all.
This isn't the version I got printed, that one was processed again afterwards which made it pop more BUT really damaged the higher resolution views.

[strongmanmike]

Really nice bit of work there Col...wonder how many life forms or civilisations are in that shot...?

Mike

[Atmos]

Probably not a whole lot as the closer you get towards the centre the more violent it is. Life forms don't like supernovae. Burn baby burn.