A dream is to do a complete mosaic of the Scuttling Chook. Here we have a five panel region spanning just over 1 degree, with a total of 65 hrs of exposure.
H-alpha 20 hrs, OIII 24 hrs, SII 21 hrs, all in 1 hr subs. Aspen CG16M on 20" PlaneWave. Except for embedding an sRGB profile, all processing using GoodLook 64. We've reduced the original 7657 pixel wide image to 4096 pixels, for ease of handling.
[link to original full sized image deleted. New link to revised version below]
The range of textures in the region is quite amazing. A grand overview would be that we are looking at a hollow cloud of gas with a convoluted oyster-shell surround, strong in H-alpha (green) and SII (red), the space within glowing with OIII (blue). There is a strong impression that the OIII is blasting out toward the right, through the open face of the oyster.
The top of the oyster shell (12 o'clock) is adorned with the face of a bull, delicately outlined in thin shock fronts.
The left hand side, especially toward 8 o'clock, shows the rough, outer shell of the oyster, with bright crags and dark crannies.
Dead in the centre, we have the Running Chicken's famous Bok-Bok-Bok globules.
Worth a close look: About 30% of the way toward the top left corner is the Chook's nemesis, the evil Fox-Cat (whose face is a quantum superposition of vulpeculine and feline), quietly smoking a post-prandial cigar. For those who have trouble, the head is on the left, the curled-up contented body on the right, and the cigar is poking out toward the left of the mouth. This evil but well-fed creature explains the lack of a visible chook in this image.
Update:
New version with careful attention to red rings around stars, and slightly more contrasty.
Ah yes, "The Cauldron" as I like to think of it- as it resembles the olympic flame.
Crikey that's sharp- guess you can't argue with a 20" light bucket!
Great work guys, so does your software do mosaics too, I'm intrigued?
Q: Did you try to remove the star halos - the red/magenta? Or is this something you prefer not to do?
Thanks, Chris.
You should have seen the magenta stars before! We automatically identified well over 65,000 stars, and greatly attenuated their magenta haloes. However, we try not to mess too much with the haloes of the very largest ones, like Lambda Centauri. Perhaps we could have been more zealous here, but we think it's not too distracting. To get rid of yet more magenta, all we'd need to do is make the boxes around the stars a bit bigger.
Quote:
Originally Posted by Andy01
Ah yes, "The Cauldron" as I like to think of it- as it resembles the olympic flame.
Crikey that's sharp- guess you can't argue with a 20" light bucket!
Great work guys, so does your software do mosaics too, I'm intrigued?
A very nicely baked chook, yum
Thanks muchly, Andy! Yes, the "Prometheus" module of GoodLook, which does registration, stacking, and artifact rejection, treats all images as highly overlapping mosaics. If it really is a mosaic, you don't have to do anything different or special. So long as all frames overlap the key frame by at least say 25%, all the registration, normalization, and blending happens automatically. Some of the subs were taken under full moon, others at new. The difference is handled automatically without having to do anything special.
Thanks for the explanation. So removing the magenta is done when the frames are combined? What about trying to reduce the star sizes on the appropriate channels before combining?
It's certainly not too distracting as it is - there's a lot of features to look at.
Gee thats very good Mike. You see lots of pics of the middle bit, but this large mosiac with the walls of outer nebula is far more interesting, with fantastic detail in those walls, especially the middle left and bottom right. Far more going on there than the just the boks in the middle. Excellent composition too.
Thanks for the explanation. So removing the magenta is done when the frames are combined? What about trying to reduce the star sizes on the appropriate channels before combining?
It's certainly not too distracting as it is - there's a lot of features to look at.
The magenta halo removal is done at a fairly late stage. The first steps are uncontroversial:
- Register
- Winsorize the stack to remove outliers (cosmic rays, satellites, hot pixels, etc)
- Combine to produce an RGB image.
- Crop off margins with insufficient data.
- Set zero point.
- Colour balance to make image on the whole colour neutral.
- Say 5-10 rounds of Richardson-Lucy deconvolution, with anti-panda protection.
- Preliminary nonlinear (arcsinh) stretch, to produce close to final image.
Only now do we mess around with haloes:
- Find the stars, by comparing with a template. In this image, we found well over 65,000 stars. There are usually quite a few close double stars missed by the automatic algorithm, which have to be found manually.
- In a copy, automatically remove the stars. To do this, GoodLook estimates what is "under" the stars (using 3rd order bipolynomial regression on a Winsorized sample) to produce a more-or-less starless image. We prefer to err on the side of leaving some distant halo rather than messing with genuine nebulosity, but the distance we go out into the halo is adjustable.
- Subtract "starless" from original, to produce "stars only".
- In the "stars only" image, discard the OIII and SII utterly, leaving only H-alpha, and remap it to white.
- Now we do wavelet sharpening on the "starless" image. This is a good time, because you don't have to worry about black rings around the stars.
- Add the stars back in. We could of course just leave the stars out, and have a pretty good "starless" final image, but firstly there are always faint haloes left, especially on the very brightest stars, and secondly, stars are people too.
- Final tweaks in brightness and contrast.
- Yippee!
Apart from the deconvolution, and the complete discarding of the SII and OIII stars, there is no further attempt at shrinking the stars or rounding the stars. That seems unjustified, and perhaps the way to artifact and perdition.
Hope that makes more sense now.
Best,
Mike
(and Trish, who is trying to stop Honey West, the lorikeet, from typing too).
Absolutely love it MnT. From the 3D Bok Globules to the yellow fish being electrocuted (also yellow) towards the upper left.
Thanks, Colin! Tickled.
Quote:
Originally Posted by Bassnut
Gee thats very good Mike. You see lots of pics of the middle bit, but this large mosiac with the walls of outer nebula is far more interesting, with fantastic detail in those walls, especially the middle left and bottom right. Far more going on there than the just the boks in the middle. Excellent composition too.
Cheers, Fred! We started with the middle, a year or so ago, but kept wondering what was just out of field. This image does have some sort of edge to it now.
Massive effort and beautiful image M&T. A great demo of your software too. Fun description.
Just loved scrolling around the big image, such crisp detail.
Just one minor point, the odd area has some hot or extremely saturated pixels, I'm sure you'll fix that. Unless you've discovered some new alien beacons or whatnot!
Massive effort and beautiful image M&T. A great demo of your software too. Fun description.
Just loved scrolling around the big image, such crisp detail.
Just one minor point, the odd area has some hot or extremely saturated pixels, I'm sure you'll fix that. Unless you've discovered some new alien beacons or whatnot!
Thanks, Simmo. Guilty as charged. No aliens. There are some peripheral regions with only two OIII subs contributing, and they are awash with blue hot pixels, cosmic ray hits, and even a satellite trail. So no outlier rejection in these regions. Another smaller patch has only two Ha subs. Just one more sub and they would go away. Will see what we can do.
An excellent image guys, clearly a bit of work in that. Great colour palette, sharpening has been handled very well (you are getting under my radar more and more ) and the contours and 3D features are coming through nicely. Winston Churchill's Cigar looks excellent and the stars look good - overall a rather dramatic portrait of this perennial favourite. I think the only minor buggy thing stopping this from being a complete show stopper is the red flaring/halos around a lot of the medium-fainter stars, even so it is really excellent anyway but love to see it again with the red reduced significantly.
Great work.....and even with the red, the amount of work, the programming and processing in that there image... makes it still worthy of one of these ....fix the red and you can have a whole row of em
An excellent image guys, clearly a bit of work in that. Great colour palette, sharpening has been handled very well (you are getting under my radar more and more ) and the contours and 3D features are coming through nicely. Winston Churchill's Cigar looks excellent and the stars look good - overall a rather dramatic portrait of this perennial favourite. I think the only minor buggy thing stopping this from being a complete show stopper is the red flaring/halos around a lot of the medium-fainter stars, even so it is really excellent anyway but love to see it again with the red reduced significantly.
Great work.....and even with the red, the amount of work, the programming and processing in that there image... makes it still worthy of one of these ....fix the red and you can have a whole row of em
Mike
Thanks muchly, gentle teacher. Reckon you are right about the rings. We did go through the motions as mentioned to Lazjen, but obviously not well enough. Just a plain mistake. A great excuse to get a few more hours on the panel with insufficient OIII, make a strong espresso, and reprocess without changing anything else.
Expecting next few days, including 50 mm rain. That will help the garden and wash the observatory dome.
The magenta halo removal is done at a fairly late stage. The first steps are uncontroversial:
...snip...
Apart from the deconvolution, and the complete discarding of the SII and OIII stars, there is no further attempt at shrinking the stars or rounding the stars. That seems unjustified, and perhaps the way to artifact and perdition.
Excellent. Thanks for the detailed explanation. Gives me something to think about when I finally get back to processing an image with more than one channel...
I was really happy to see this pop up here as I'm also working on my own rendition of this subtle area of the sky. Gives me an idea as to what to look forward to.
I'm completely in awe of not just the 1 hour sub-exposures you guys do, but, also the fact it's processed in your own home-grown software. Very impressive!
There are some peripheral regions with only two OIII subs contributing, and they are awash with blue hot pixels, cosmic ray hits, and even a satellite trail. So no outlier rejection in these regions. Another smaller patch has only two Ha subs. Just one more sub and they would go away. Will see what we can do.
) and the contours and 3D features are coming through nicely. Winston Churchill's Cigar looks excellent and the stars look good - overall a rather dramatic portrait of this perennial favourite. I think the only minor buggy thing stopping this from being a complete show stopper is the red flaring/halos around a lot of the medium-fainter stars, even so it is really excellent anyway but love to see it again with the red reduced significantly.
....fix the red and you can have a whole row of em 
next few days, including 50 mm rain. That will help the garden and wash the observatory dome.
