M83 repro - first image with PixInsight

[pluto (Hugh)]

Hi all,

Having become a bit frustrated with the limitations of my current processing workflow I've decided to give PixInsight a try.
I'm really liking it so far, it's much more logical and straightforward than PS, and it shares some workflow similarity with some of the image manipulation tools I use in my day job.

Anyway as a first image I decided to reprocess my M83 from IISAC 2014 with some Ha data taken from Sydney. IISAC was the first time I had done LRGB with my current setup and I had mistakenly labeled my filters in the wrong order in CCDSoft. I had them labeled L-R-G-B when, in fact, the filters were inserted R-G-B-L . So I ended up getting quite a lot of R data, thinking it was L, and not a great deal of L data. Also I didn't get flats so I had to go pretty hard with multiple passes of DBE.

I'm pretty happy with what I've got from this data using PI especially when compared to the previous process I did just after IISAC.
I'm not terribly happy with the star colours but I'll wait until I have an opportunity to process an LRGB image with adequate data before I really worry about that.
Also I may have gone a bit hard with the deconvolution as I found it a little hard to get the balance right, however practice makes perfect.
Next up will be processing some NB data of the Eagle I've got and see how I go in PI with that.

Bigger one here: http://hughsblog.wordpress.com/2014/...agb-reprocess/

I've also attached here my old processed image for comparison.

Any opinions/critiques/comments/tips would be welcomed.

Thanks for looking

Ha = 13x20min
L = 5 x 10min
R = 10 x 10min
G = 4 x 10min
B = 6 x 10min
Takahashi TSA-120 / SW EQ6 / SBIG STT-8300M / FW8G + AO-8T / PixInsight

[Rod771 (Rod)]

Welcome aboard Hugh!

I found Pixinsight quite easy to learn, though I must say I was starting from scratch without previously being used to PS. There's lots of support on the Pixinsight forum. Watching "Harry's tutorials" is a great way to get going.

The repo is nice! Look at all that red channel data . Did you try the LHE tool using a luminance masked to protect the background? Its a great way to enhance contrast after using the HRDMT tool.

[pluto (Hugh)]

Thanks Rod!
Yeah I watched Harry's tuts, they were a good start.
I'll check out LHE, thanks for the tip.

I didn't have much luck with HDRMT, actually it was weird, I did a Lum mask and a small preview and chucked on a default HDRMT and it looked really good, so I closed the preview and chucked the same HDRMT onto my main image and it was totally different with much less effect. I played with the HDRMT tool for another half hour and was never able to replicate what I had seen that first time. I'm sure I changed something but I'm buggered if I know what!

More trial-and-error needed

EDIT: just had a super quick play with LHE and I like it! I'll definitely be exploring that a bit further when I have time!
Here's the result of a bit of LHE and some curves to bring down the star bloat a bit (perhaps too much?)

[LewisM]

P I and me don't see eye to eye. MaxIM and PS5 for me I am afraid, with stacking and preprocessing in CCDStack.

Looks good Hugh. Teach me one day

[RickS (Rick)]

Originally Posted by pluto

EDIT: just had a super quick play with LHE and I like it! I'll definitely be exploring that a bit further when I have time!
Here's the result of a bit of LHE and some curves to bring down the star bloat a bit (perhaps too much?)

For both HDRMT and LHE it's good to use a luminance or lightness mask to control the strength. HDRMT has an option to do the masking but I like building my own mask to have more control. For these processes I also remove stars from the mask to prevent the bloat that you noticed.

Cheers,
Rick.

[pluto (Hugh)]

Originally Posted by LewisM

Looks good Hugh. Teach me one day

Don't look at me! I don't know what I'm doing!!
I can't recommend Harry's video tuts enough though, they're a great resource for starting PI from scratch.

Originally Posted by RickS

For these processes I also remove stars from the mask to prevent the bloat that you noticed.

Do you just use PixelMath to subtract the star mask from the Lightness mask or is there a process that does something similar?

[RickS (Rick)]

Originally Posted by pluto

Do you just use PixelMath to subtract the star mask from the Lightness mask or is there a process that does something similar?

I use the Swiss Army Knife that is PixelMath... a simple approach would be something like lum_mask*~star_mask.

Cheers,
Rick.

[strongmanmike (Michael)]

Ah repros, love'em (and hybrids )

I can see some improvements in the new version for sure but as is often the case with such exercises it looks like a mix of the two might be perfect

I do like the detail in the latest version (seems your decon is nicely handled in the arms, with no sign of the dreaded bright worms ) but I don't mind the colour balance and although a bit strong, the sharpness in the original.

I'm thinking use a blend of the two for the colour component and then the latest version (with just the stars from the original blended in 50/50) for the luminance just an idea

Great shot

Mike

[pluto (Hugh)]

Originally Posted by RickS

I use the Swiss Army Knife that is PixelMath... a simple approach would be something like lum_mask*~star_mask.

Thanks Rick!

Originally Posted by strongmanmike

Ah repros, love'em (and hybrids )

I can see some improvements in the new version for sure but as is often the case with such exercises it looks like a mix of the two might be perfect

I do like the detail in the latest version (seems your decon is nicely handled in the arms, with no sign of the dreaded bright worms ) but I don't mind the colour balance and although a bit strong, the sharpness in the original.

I'm thinking use a blend of the two for the colour component and then the latest version (with just the stars from the original blended in 50/50) for the luminance just an idea

Great shot

Mike

Thanks for the feedback Mike.
I actually thought I'd gone a bit too purple with the old version but I see what you mean about the sharpness and the star sizes in the old one. I think I'll keep plugging away with it in PI as I certainly don't mind a bit of extra practice


EDIT: I've had a bit more of a play with the image and, while I'm sure I'm not done with it yet , I'm happier with it.

[Nortilus (Josh)]

awesome...makes my m83 look like it needs more work...

[Amaranthus (Barry)]

That latest reprocess is looking terrific Hugh. Lots of HII regions and the deep background galaxies also show up clearly. Nice!

Do I see a few hot pixels or are those just really red background stars?

[pluto (Hugh)]

Originally Posted by Nortilus

awesome...makes my m83 look like it needs more work...

Thanks Josh

Originally Posted by Amaranthus

That latest reprocess is looking terrific Hugh. Lots of HII regions and the deep background galaxies also show up clearly. Nice!

Do I see a few hot pixels or are those just really red background stars?

Cheers Barry, and yes there could be a few hot pixels. When I processed my second image in PI (M8) I learned how to do cosmetic correction properly, and realised I hadn't been doing it properly before, so there could still be some nasty bits in this M83.

[multiweb (Marc)]

I like the details you've achieved but not the colors. The star colors are all gone. Two bright stars next to each other are yellow and deep blue so is the other one up. I feel you've focussed your processing on the galaxy alone which should have a lot of yellow and brown as well. Now it has a red/blueish cast over it. But the whole starfield is now whitish.

[pluto (Hugh)]

Originally Posted by multiweb

I like the details you've achieved but not the colors. The star colors are all gone. Two bright stars next to each other are yellow and deep blue so is the other one up. I feel you've focussed your processing on the galaxy alone which should have a lot of yellow and brown as well. Now it has a red/blueish cast over it. But the whole starfield is now whitish.

Thanks for the feedback Marc, I'm not happy with the star colours either!
I agree that I've pushed the galaxy toward purple, probably too far, and I need to work on the star colours

[gregbradley]

That's looking quite good. My only suggestion is its a bit too magenta. Colour is a bit tricky with M83 in my experience. If you can use selective colour tool in PS or equivalent in PI to reduce the magenta and make it more of a shade of blue you'd be closer. It is a slightly magenta type coloured galaxy but I don't think that much.

Greg.

[pluto (Hugh)]

Originally Posted by gregbradley

That's looking quite good. My only suggestion is its a bit too magenta. Colour is a bit tricky with M83 in my experience. If you can use selective colour tool in PS or equivalent in PI to reduce the magenta and make it more of a shade of blue you'd be closer. It is a slightly magenta type coloured galaxy but I don't think that much.

Greg.

Thanks for the advice Greg. I think I'll revisit it in a week or so with fresh eyes, and a bit more PI knowledge, and have another go at getting the colour sorted, especially in the stars.

[multiweb (Marc)]

Keep the galaxy. Just do a star mask to redo the field with a separate pass, colours + saturation.

[Geoff45 (Geoff)]

Originally Posted by pluto

Hi all,

Having become a bit frustrated with the limitations of my current processing workflow I've decided to give PixInsight a try.
I'm really liking it so far, it's much more logical and straightforward than PS,

Nicely done Hugh. A bit more practice with PI will have you churning out masterpieces. I really can't understand why some people find it difficult and unintuitive. My best moment in AP was when I could
1. Use ONE piece of software to calibrate, register, stack and process.
2. Turn my back on all the magic tricks that are necessary in PS

I agree with OPs remarks about too much magenta and lack of colour in the stars. There's actually quite a lot of blue in the outer regions. See here http://www.astrobin.com/93786/
Geoff