Antares Rho Ophiuchus reprocessed

[rcheshire (Rowland)]

It's not great, but I'm re presenting this after lots of work ironing out as many of the bugs as possible. Processed in Pixinsight, including calibration, registration and integration.

The difference with this result is just a matter of doing all the major fixing while the image was linear - different star size between channels. Stars in R channel were morphed up to match B and G channels - later morphed down. Besides removing gradients and colour calibration the background was enhanced, which improved image response to colour saturation while reducing noise propagation throughout processing. Finally, rescaling the L* channel gave the image a contrast lift.

The aim was to tease out as much detail as possible, consistent with saturation without blotting out finer structures.

http://synergous.com/photos/picture.php?/170/category/4

If you got this far please feel free to provide constructive criticism.

If anyone is interested in the PI work flow I'll be happy to post.

Once again: Canon 1000D 200mm prime lens, f/3.2, iso800, Astronomik CLS-CCD filter, 38 exposures 15 - 5 minutes, bias, darks and flats.

[multiweb (Marc)]

Originally Posted by rcheshire

If anyone is interested in the PI work flow I'll be happy to post.

Cool shot - looks a little on the blue side to me. There's a lot of dust around Antares normally yellow that's rendered blue in your shot.

Interested in that channel stars morphing so yes please post your PI steps. All here to learn new cool tricks.

[rcheshire (Rowland)]

Thanks Marc. Now that you mention it I can see that it is on the blue side - colour blind? I've reprocessed it.
http://synergous.com/photos/picture.php?/170/category/4

R channel star sizes were smaller - I suspect focus. Fixed using the ChannelExtraction, PixelMath and the MorphologicalTransformation (MT) tools. There might be another way, but this worked well.

To the integrated linear image;

ScreenTransferFunction (STF)
AutomaticBackgroundExtraction (ABE)
Extract L* for the mask
Extract R and subtract from the integrated image
Apply L* mask to R (inverted)
MT - Midpoint: (MIN+MAX)/2 operator and amount 0.25 - this increased star size to same as B and G channel.
Add modified R to integrated image.

The other way of doing it is to morph down the other channels, but after trying different methods and masks I found it was easier to morph down after stretching the image. It just needs good masking techniques and experimentation, which ever way you do it.

If you want to know the full details let me know.

[Paul Haese]

Agree with Marc there Rowland. Try upping your green channel a little or dropping your blue too in combination. The area immediately around Antares should appear as a yellowy dust area. Well done other than that.

[gregbradley]

Nice image Rowland. Overall very nice.

Yes the colours around Antares should be more yellow but still good.

A bit of coma showing in the stars when you zoom into the corners. Did you stop down your lens a couple of F stops or did you take the shot wide open? Most commercial lenses require knocking them down a couple of stops. I am going through the same thing imaging with a CCD and some Pentax lenses.

Also there was a thread here recently I think from Marc about aperture rings. As I recall it mentioned that they can stop the multi spikes on the bright stars that come from the Iris mechanism in the lenses.

I'd like to find out more about that if that works as again I would like to prevent that with my own lenses.

A great shot of a fabulous area of the sky.

Greg.

[irwjager (Ivo)]

Wow! Really nice going Rowland - that's an awesome shot!
My 1 little nitpicks; To the far left, I can see a fuzzy vertical blue band, running from top to bottom. It's very faint, but it's there and it looks artificial in origin. Any idea what could have caused that?

Nitpick #2 is that I'm not so sure that the ABE did it's job that well - there seems to be a quite a bit of brownish background level, which seems to be swamping the yellow/orange part in the complex around Antares somewhat (as Greg mentioned). If you get rid of it properly (try PI's DBE?) I reckon it'll come out quite nicely (quick repro attached). I know I'm nitpicking, but Rho Oph is all about those 3 contrasting colors. You know, they need to pop!

This is all very minor stuff - you should be proud of this one. I know I would be!

Cheers,

[rcheshire (Rowland)]

Thanks Paul, Greg and Ivo.

I think the green noise reduction is a bit to aggressive too early in processing wiping out the yellow.

The 200mm lens performs well at f5.6 3.2 and 2.8 is coma city. I've not tried in between. But aperture rings are the only solution to the iris spikes. They haven't been in vogue for years - where can they be sourced.

Ivo. StarTools rules supreme again. Looks much better than the original - nice going. I was going to use the SPFfunction to fix all the stars and haven't got around to it yet. I see what you mean about the line in the background - I think it's to do with flats. I've tried DBE and ABE same effect. Looks like a manual DBE after all.

I'm not a very good advertisement for Pixinsight. It's definitely a user problem.

[gregbradley]

Did you use HLVG for the green reduction?

It is a great tool but yes you have to like any other tool check out its effect before committing to it permanently. Sometimes it can damage
features. It is also supposed to be applied to a luminosity layer rather than a normal layer. Did you do that?

The repro looks better.

Greg.

[rcheshire (Rowland)]

Hi Greg. Pixinsight uses SCNR - which is I think selective colour noise reduction. My application was a bit extreme. It is applied to the RGB and usually does as advertised. I would suggest that my application is the cause. For the repro, instead of applying it to the image near completion, it was applied shortly after colour calibration, in the linear stage. This seems to have made a difference.