Jump to content


Photo

Achieving color in galaxy shots

  • Please log in to reply
48 replies to this topic

#26 akulapanam

akulapanam

    Mariner 2

  • -----
  • Posts: 224
  • Joined: 26 Aug 2012

Posted 07 February 2013 - 10:57 PM

Try this one.

http://sdrv.ms/XUuCjo

#27 srosenfraz

srosenfraz

    Soyuz

  • *****
  • Posts: 3998
  • Joined: 06 Mar 2011
  • Loc: United States

Posted 07 February 2013 - 11:28 PM

Try this one.

http://sdrv.ms/XUuCjo



That works - I can download that one. I'll probably give it a try tomorrow.

#28 srosenfraz

srosenfraz

    Soyuz

  • *****
  • Posts: 3998
  • Joined: 06 Mar 2011
  • Loc: United States

Posted 08 February 2013 - 10:25 PM

I did a quick processing on the sample data you posted. Obviously, the noise is very high since there's less than 3 minutes of total exposure and the data was captured from a red zone. As such, I was limited to how much I could stretch it and how much can be shown in the image. However, I think you can see that the galaxy's spiral arms have a blue color to them, while the core is heading towards a yellow/red color. These are typical colors I expect from M81.

When I compare it to your original image, it appears that your image lacks much of the blue, and the core is much redder than my processing of your 4x40 subs. Also, when I processed your original 32 bit Tiff, I had to push blue into the galaxy in order to have similar colors.

My gut suspicion is that there's something you're doing in your pre-processing that is altering the colors or removing some of the color. I'd be hard pressed to tell you what it is, but I didn't have to do anything fancy with my processing of the 4x40 subs to bring out blue in the galaxy.

If it'll help, here's the basic workflow I did with your data:

1) Used Images Plus 5.0 for automatic processing (generate master dark, generate master flat, convert CR2s to FITs, calibrate lights, normalize lights, align, and combine). This is a lot of steps, but IP does it all automatically after you point it to the data files.

2) Initial DDP stretch in Images Plus 5.0 - settings of Break Point 10,000, Background Weight .92, Scale function nth Root X^n, Root .820

The rest of the processing was in Photoshop CS5:

3) Gradient Xterminator

4) Levels - set initial grey point, set initial black point.

5) HLVG - Medium setting. Removes some of the green pixel noise.

6) Mild stretching curve

7) Carboni Deep Space Noise Reduction

8) Another mild stretching curve

9) Saturate using Block Method - 3 times

10) Levels - set final white and black points


Other than saturating and HLVG, there's nothing in this workflow that would have affected the color balance in the image (the HLVG would have taken out some green, so the image is very slightly more magenta). Certainly there's nothing here that would have pushed blue into the spiral arms.

I'm not sure if this is helpful, but hopefully it may give you some ideas about where to look for your lost color. Please let me know if you have any questions or thoughts.

#29 pfile

pfile

    Gemini

  • -----
  • Posts: 3163
  • Joined: 14 Jun 2009

Posted 08 February 2013 - 10:56 PM

btw i agree, i processed the image in PI and with enough saturation boost you can see the right colors, albeit very faint.

bottom line is that when i shot this at f/7 from a red zone, i had to do 10h of integration time to get a decent result. 3 minutes is just too short to expect much...

#30 akulapanam

akulapanam

    Mariner 2

  • -----
  • Posts: 224
  • Joined: 26 Aug 2012

Posted 08 February 2013 - 11:35 PM

Interesting it must be something in DeepSkyStacker because that is the only pre-processing I use. Any ideas?

#31 akulapanam

akulapanam

    Mariner 2

  • -----
  • Posts: 224
  • Joined: 26 Aug 2012

Posted 08 February 2013 - 11:46 PM

Here are my DeepSkyStacker settings:

Lights
Stacking Method: Kappa-Sigma Clipping
Per Channel Background Calibration

Darks
Median Kappa-Sigma Clipping
Hot Pixel Detection

Flats
Median Stacking Method

Result is set to "mosaic mode"

#32 srosenfraz

srosenfraz

    Soyuz

  • *****
  • Posts: 3998
  • Joined: 06 Mar 2011
  • Loc: United States

Posted 09 February 2013 - 01:56 AM

I don't use DSS, so I probably won't have the greatest suggestions. There's lots of folks on the list that use DSS (probably more than any other program), so hopefully someone can take a look at your settings and make suggestions.

The one thing about which I would wonder in your settings is the "Per Channel Background Calibration". Not knowing what it does, perhaps that could be related? Is this a default setting? Also, I've heard folks talk about issues with Deep Sky Stacker doing a screen stretch and if you save the file incorrectly, it'll use the screen stretch and create some color problems.

Again, I haven't used DSS, so I'm just speculating that perhaps there's something within that issue could be related to your color issue.

#33 srosenfraz

srosenfraz

    Soyuz

  • *****
  • Posts: 3998
  • Joined: 06 Mar 2011
  • Loc: United States

Posted 09 February 2013 - 02:00 AM

btw i agree, i processed the image in PI and with enough saturation boost you can see the right colors, albeit very faint.

bottom line is that when i shot this at f/7 from a red zone, i had to do 10h of integration time to get a decent result. 3 minutes is just too short to expect much...



Just to be clear - akulapanam had indicated in his original post that his image was made from about 2 1/2 hours of integration. The 3 minutes of subs he posted here was just because of the constraints of trying to transfer raw data over an internet connection.

Also, I'm glad to hear you saw similar results when processing in PI - I think it confirms that there is actually useful color in the data.

#34 pfile

pfile

    Gemini

  • -----
  • Posts: 3163
  • Joined: 14 Jun 2009

Posted 09 February 2013 - 03:55 PM

oh sorry, i got confused... as usual.

#35 akulapanam

akulapanam

    Mariner 2

  • -----
  • Posts: 224
  • Joined: 26 Aug 2012

Posted 09 February 2013 - 06:03 PM

I switched it to the RGB option instead of the Per Channel and it didn't have an effect :-( See below

http://sdrv.ms/Y1MeYV

#36 pfile

pfile

    Gemini

  • -----
  • Posts: 3163
  • Joined: 14 Jun 2009

Posted 09 February 2013 - 11:35 PM

do you have 'set black point to zero' checked? that was the culprit in someone else's data that i was looking at recently.

#37 akulapanam

akulapanam

    Mariner 2

  • -----
  • Posts: 224
  • Joined: 26 Aug 2012

Posted 10 February 2013 - 12:32 AM

No I double checked that and I played around with white balance on the same page without any results.

#38 akulapanam

akulapanam

    Mariner 2

  • -----
  • Posts: 224
  • Joined: 26 Aug 2012

Posted 10 February 2013 - 09:38 PM

Scott if you don't mind try this file out and see if you can do anything with it. It is looking a little better to me. I used imagemagick to get around the dss saving issue and I switched dss to standard mode from mosaic. The mode shouldn't make a difference but it might help.

http://sdrv.ms/154NAcz

#39 mmalik

mmalik

    Fly Me to the Moon

  • *****
  • Posts: 5344
  • Joined: 13 Jan 2012
  • Loc: USA

Posted 11 February 2013 - 06:48 AM

My go at your AS2 data...

Processing Workflow: (Instructions here...)
•Crop in ImagesPlus 5.0
•Digital Development in ImagesPlus 5.0
•Multi Point Flatten Background-Planar in ImagesPlus 5.0
•Multiresolution Smooth-Sharpen (Finest/Fine 40/20) in ImagesPlus 5.0
•Geometric Transform/Scale (Factor 0.5) in ImagesPlus 5.0
•Smoothing in ImagesPlus 5.0
•Saturation (Block I), and 'Noise Reduction' in Photoshop Extended CS6
•Star Size and Halo Reduction in ImagesPlus 5.0
•Hue-Saturation-Luminance in ImagesPlus 5.0
•Saturation (Block II), Curves, Contrast, and Saturation in Photoshop Extended CS6

Attached Files



#40 Dave Venne

Dave Venne

    Sputnik

  • *****
  • Posts: 38
  • Joined: 12 Nov 2007
  • Loc: Burnsville, MN, USA

Posted 11 February 2013 - 05:58 PM

Your image has some real difficulties with the background--light pollution and some vertical banding is making it very uneven. But there is enough signal to produce some reasonable color--as Scott's quick processing showed. Here's Mikes image, with some judicious channel mixing. I'm not entirely happy with the tinge of the background, but the galaxy center has a rosier glow and the spiral arms are bluer, both of which resemble the colors found by much better imagers than myself.

Attached Files



#41 akulapanam

akulapanam

    Mariner 2

  • -----
  • Posts: 224
  • Joined: 26 Aug 2012

Posted 11 February 2013 - 09:40 PM

Those are looking better. I'm still not sure why DSS is suppressing the color so much compared with the imagesplus stack. I'm going to try and get a free trial and compare the results with 200+ frames.

No clue on the vertical banding I have never had that problem before.

#42 srosenfraz

srosenfraz

    Soyuz

  • *****
  • Posts: 3998
  • Joined: 06 Mar 2011
  • Loc: United States

Posted 13 February 2013 - 12:01 AM

Scott if you don't mind try this file out and see if you can do anything with it. It is looking a little better to me. I used imagemagick to get around the dss saving issue and I switched dss to standard mode from mosaic. The mode shouldn't make a difference but it might help.

http://sdrv.ms/154NAcz



Sorry, I didn't have a chance to look at this until tonight. This file was easier to work with in that the blue color was inherently there. I enhanced it via Lab Color and Selective color adjusments, but I didn't have to force blue into it by stretching the blue channel (like I had to do with the earlier version). Here's my quick and dirty on it:

Posted Image

I posted my .psd file if it'll help you (40 MB).

#43 akulapanam

akulapanam

    Mariner 2

  • -----
  • Posts: 224
  • Joined: 26 Aug 2012

Posted 13 February 2013 - 01:03 AM

Thank! That really does look much better. The key seems to be using imagemagik to take the autosave and convert it, really "unbox it", to 16 bit. I just downloaded your psd file and I'm going to try and learn how to do your processing routine.

#44 srosenfraz

srosenfraz

    Soyuz

  • *****
  • Posts: 3998
  • Joined: 06 Mar 2011
  • Loc: United States

Posted 13 February 2013 - 02:53 AM

Thank! That really does look much better. The key seems to be using imagemagik to take the autosave and convert it, really "unbox it", to 16 bit. I just downloaded your psd file and I'm going to try and learn how to do your processing routine.


Glad to hear you may have found a better way convert it. Let me know if you need any explanation on the .psd file. I left it all as separate layers to show all steps, but you could flatten the file after each step or a group of steps. Good luck with it!

#45 akulapanam

akulapanam

    Mariner 2

  • -----
  • Posts: 224
  • Joined: 26 Aug 2012

Posted 16 February 2013 - 06:38 PM

I have been reviewing what you did. The only area I'm having a hard time understanding is how you did the lab color b channel increase.

#46 srosenfraz

srosenfraz

    Soyuz

  • *****
  • Posts: 3998
  • Joined: 06 Mar 2011
  • Loc: United States

Posted 16 February 2013 - 09:29 PM

I have been reviewing what you did. The only area I'm having a hard time understanding is how you did the lab color b channel increase.


Here's my process for doing that:

1) Duplicate the image (Image | Duplicate)

2) Flatten the image (if it isn't flat).

3) Convert to Lab Color (Image | Mode | Lab Color)

4) Click on the Channels tab and the click on the b channel so that it is highlighted in blue. Then, click on the eyeball next to "Lab" so that all channels have eyeballs (Lab, lightness, a, and B). This makes it so that you see all the channels (i.e., the image looks normal), but you're only working on the b channel.

5) Increase the contrast (Image | Adjustment | Contrast). This is one of the few times that you want to use Contrast instead of curves. You are not increasing the contrast of the image - you are increasing the contrast of the b channel. The b channel affects blue and yellow. So, increasing the contrast in the b channel increases the saturation of blue and yellow. Increase the contrast (saturation) until you're satisfied with the result. If needed, you can do a second or third pass of increasing contrast to bring out even more color.

6) You can do the same thing with the a channel (increase the contrast). This increases the saturation of the green and magenta. Most astronomical images will have very little green (the general exception being planetary nebulae), so the effect of increasing the contrast on the a channel is to saturate magenta. This can be useful for bringing out Ha regions in galaxies or emission nebulae.

7) Once you're satisfied with your adjustments, click on the "Lab" channel. This should highlight all 4 channels (Lab, lightness, a, and B).

8) Copy the image to the clipboard (Edit | Select All; Edit | Copy).

9) Click back on the original image and then click on the Layers tab. Edit | Paste your Lab color adjusted image as a layer on top of the stack.


That's the basics. A couple of things to keep in mind with this:

a) Lab Color is good for many images, but keep in mind that you can't increase blue without increasing yellow (and, similarly magenta and green). So, while you're increasing the contrast in the b or a channel, keep an eye on both colors you're saturating to make sure that one doesn't become overpowering (i.e., if increasing the b channel makes your image look to yellow, its time to not increase it so much).

B) Oftentimes, you'll want to use Lab color for PART of an image and not the entire image. Using masks can be very effective in bringing out the color in just the areas you need.

Hope this helps.

#47 Tonk

Tonk

    Fly Me to the Moon

  • *****
  • Posts: 7078
  • Joined: 19 Aug 2004
  • Loc: Leeds, UK, 54N

Posted 17 February 2013 - 05:41 AM

astronomical images will have very little green (the general exception being planetary nebulae)


You have totally forgotten about comets - can be the greenest of all astro objects!


On colour saturation in general. The alternative to doing L*A*B colour treatment is Photoshops Match Colour tool. It isn't immediately obvious how to do this given the tool is actually designed to balance colour in one image to match another, but if you set it so the two (source and target)images are the same image then the colour saturation controls work superbly and avoids increasing chromatic noise etc while giving excellent colour enhancement from the midtones towards the highlights (my two recent comet postings use this technique)

Another trick is to use a Hue/Saturation layer to selectively desaturate precise colour ranges after a general saturation step. I set it to remove greens and magentas in stars after using a "match colour" saturation - though if your initial colour balance after debayering is good this desat step rarely has to be done.

#48 srosenfraz

srosenfraz

    Soyuz

  • *****
  • Posts: 3998
  • Joined: 06 Mar 2011
  • Loc: United States

Posted 17 February 2013 - 11:17 AM

astronomical images will have very little green (the general exception being planetary nebulae)


You have totally forgotten about comets - can be the greenest of all astro objects!



Whoops - you're absolutely right. My apologies to you and all the comet photographers. Sorry to have been a DSO snob.

:-)

#49 Tonk

Tonk

    Fly Me to the Moon

  • *****
  • Posts: 7078
  • Joined: 19 Aug 2004
  • Loc: Leeds, UK, 54N

Posted 17 February 2013 - 11:53 AM

:) :)






Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics