Page 1 of 2

IC342 - second image with Star Tools

PostPosted: Fri Jan 31, 2014 2:35 am
by markcasazza
I'm starting to get more comfortable with StarTools. Please find my version of IC342 at http://www.astrobin.com/75661/

My process at a high level was
Stack in Nebulosity
Stretch in StarTools
Separate luminosity layer in PS via L layer in LAB color mode
Screen Mask Invert luminosity
Use Reveal and noise reduction is StarTools to work magic on the luminosity layer
Enhance saturation with Block method in PS
Combine

I look forward to revisiting this with a triplet and more time in the future.

Comments welcome.

Mark Casazza
http://casazza.net
Home of the Clear Sky Alarm Clock and Tonight's Sky

Re: IC342 - second image with Star Tools

PostPosted: Fri Jan 31, 2014 11:56 pm
by admin
Hi Mark,

You got some excellent detail In the galaxy, however it seems you have lost all color in the process?
Could you tell me more about the 'Block Method'?
If you're interested SMI (as well as PIP) can be performed in StarTools as well, so that StarTools' Tracking module can take into account these procedures when performing final noise reduction.

Re: IC342 - second image with Star Tools

PostPosted: Sat Feb 01, 2014 2:09 am
by markcasazza
Ivo,

I am eager to learn more and share what I do know. I'd be glad to learn how I can do more of this in StarTools. By the way, thank you for logging what is done to each file. That makes this so much easier. I have been adapting Scott Rosen's LLRGB method (http://www.astronomersdoitinthedark.com ... torial.php); that is the base process. Here is the process I used in detail

Preprocessing is done with Nebulosity
Dark master is created for every distinct temperature that I have lights for with 50+ darks
Flats have bias already applied
Lights are pre-processed by applying the appropriate dark and flat
Images are then normalized and debayered.
I then do a manual 2 point alignment and stack the result.
Before I leave Nebulosity I do a color balance (my camera is a full spectrum T1i with a UV/Ir filter in place)

Next is StarTools where I did this:
File loaded [C:\Users\mark\Pictures\Lights\IC 342\Processing\IC342 Stkd v2 cb.fit].
---
--- Develop
Parameter [White Calibration] set to [Use Stars]
Parameter [Gamma] set to [1.00]
Parameter [Skyglow] set to [0 %]
Parameter [Digital Development] set to [93.65 %]
Parameter [Blue Luminance Contrib.] set to [100 %]
Parameter [Green Luminance Contrib.] set to [100 %]
Parameter [Red Luminance Contrib.] set to [100 %]
Parameter [Dark Anomaly Headroom] set to [5 %]
Parameter [Dark Anomaly Filter] set to [6.2 pixels]
--- Wipe
Parameter [Mode] set to [Correct Color & Brightness]
Parameter [UNKNOWN] set to [No]
Parameter [Precision] set to [256 x 256 pixels]
Parameter [Dark Anomaly Filter] set to [Off]
Parameter [Drop Off Point] set to [50 %]
Parameter [Corner Aggressiveness] set to [100 %]
Parameter [Aggressiveness] set to [75 %]
--- Develop
Parameter [White Calibration] set to [Use Stars]
Parameter [Gamma] set to [1.00]
Parameter [Skyglow] set to [0 %]
Parameter [Digital Development] set to [89.00 %]
Parameter [Blue Luminance Contrib.] set to [100 %]
Parameter [Green Luminance Contrib.] set to [100 %]
Parameter [Red Luminance Contrib.] set to [100 %]
Parameter [Dark Anomaly Headroom] set to [5 %]
Parameter [Dark Anomaly Filter] set to [Off]
--- HDR
Parameter [Small Detail Precision] set to [Max]
Parameter [Channels] set to [Brightness Only]
Parameter [Algorithm] set to [Optimize Hard]
Parameter [Dark/Bright Response] set to [Full]
Parameter [Detail Size Range] set to [21 pixels]
Parameter [Noise Suppression] set to [Off]
--- Wavelet De-Noise
Parameter [Scale 1] set to [90 %]
Parameter [Scale 2] set to [90 %]
Parameter [Scale 3] set to [90 %]
Parameter [Scale 4] set to [90 %]
Parameter [Scale 5] set to [0 %]
Parameter [Mask Fuzz] set to [1.0 pixels]
Parameter [Scale Correlation] set to [3]
Parameter [Color Detail Loss] set to [7 %]
Parameter [Brightness Detail Loss] set to [7 %]
Parameter [Redistribution Kernel] set to [4.5 pixels]
Parameter [Read Noise Compensation] set to [Off]
Parameter [Smoothness] set to [75 %]
File saved [C:\Users\mark\Pictures\Lights\IC 342\Processing\IC342 Stkd v2 cb ST.tiff].

Then in Photoshop I opened the file and did
A color balance with Levels
A quick stretch; this revealed a overdoing of vignette correction (the edges were too bright) but I liked the look of the galaxy so I sent this back to StarTools.

File loaded [C:\Users\mark\Pictures\Lights\IC 342\Processing\IC342 Stkd v2 cb ST test.tif].
---
--- Wipe
Parameter [Mode] set to [Correct Brightness Only]
Parameter [UNKNOWN] set to [No]
Parameter [Precision] set to [256 x 256 pixels]
Parameter [Dark Anomaly Filter] set to [1 pixels]
Parameter [Drop Off Point] set to [0 %]
Parameter [Corner Aggressiveness] set to [100 %]
Parameter [Aggressiveness] set to [75 %]
File saved [C:\Users\mark\Pictures\Lights\IC 342\Processing\IC342 Stkd v2 cb ST test.tiff].
** Note here I set the Mode to Brightness Only because the default mode turned the image red and I just finished a rough color balance

Back in Photoshop I converted the file to LAB color and pulled out the Luminosity data

** Work on the eventual Lumonisity layer **
Created a new file in PS with the grey scale image
Did a standard Screen Mask Invert, the result was rather bright, but I figured I would try to get galactic detail in StarTools. The actions below were with a lasso mask drawn around the galaxy. the process really blew out the already bright stars.
File loaded [C:\Users\mark\Pictures\Lights\IC 342\Processing\IC342 02-Lumonisity.tif].
---
--- Wavelet Sharpen
Parameter [Intelligent Enhance] set to [Yes]
Parameter [Scale 1] set to [100 %]
Parameter [Scale 2] set to [100 %]
Parameter [Scale 3] set to [100 %]
Parameter [Scale 4] set to [100 %]
Parameter [Scale 5] set to [100 %]
Parameter [Mask Fuzz] set to [18.5 pixels]
Parameter [Amount] set to [100 %]
Parameter [Small Detail Bias] set to [75 %]
--- HDR
Parameter [Small Detail Precision] set to [Low (Fastest)]
Parameter [Channels] set to [Brightness Only]
Parameter [Algorithm] set to [Reveal DSO Core]
Parameter [Dark/Bright Response] set to [Full]
Parameter [Detail Size Range] set to [250 pixels]
Parameter [Noise Suppression] set to [98 %]
--- Wavelet De-Noise
Parameter [Scale 1] set to [90 %]
Parameter [Scale 2] set to [90 %]
Parameter [Scale 3] set to [90 %]
Parameter [Scale 4] set to [90 %]
Parameter [Scale 5] set to [0 %]
Parameter [Mask Fuzz] set to [1.0 pixels]
Parameter [Scale Correlation] set to [3]
Parameter [Color Detail Loss] set to [20 %]
Parameter [Brightness Detail Loss] set to [20 %]
Parameter [Redistribution Kernel] set to [4.5 pixels]
Parameter [Read Noise Compensation] set to [Off]
Parameter [Smoothness] set to [75 %]
File saved [C:\Users\mark\Pictures\Lights\IC 342\Processing\IC342 02-Lumonisity.tiff].

** Combine the Luminosity layer with the original color data **

Back in PS I applied this as a luminosity layer and liked the results. Next was to increase the color saturation. The idea here is to blur the color layer and enhance the color.
The Luminosity layer will provide the detail. To this end I did the following:
Turn off the Luminosity layer
Two sets of curves
One extremely over aggressive noise reduction (I use Boundary Noise Reduction)
Next the "Block Method" to increase color saturation
Select all, Copy Merged, Paste as a new layer (twice)
The first layer is set to Soft Light
The second is set to Luminosity
This brought out the galaxy color nicely
Turn the Luminosity layer back on (to test the result) - I was happy
Do a very non-aggressive Boundary Noise Reduction
Done!

I considered the image was too red, but looking around this galaxy seems to be red because we are seeing it through a lot of dust, so I left it.
A friend with different color taste tweaked it and his result is here: https://www.dropbox.com/s/wn1nujy0yewoy ... axyHP4.gif

Mark Casazza
http://casazza.net
Home of the Clear Sky Alarm Clock and Tonight's Sky

Re: IC342 - second image with Star Tools

PostPosted: Sat Feb 01, 2014 10:41 pm
by admin
Thanks Mark,

Your processing flow seems unnecessarily long (as well as detrimental to your signal fidelity and colors).
Here are some suggestions to get more out of your data (and StarTools).

markcasazza wrote:Before I leave Nebulosity I do a color balance (my camera is a full spectrum T1i with a UV/Ir filter in place)

Don't. :)
Color balancing makes it impossible to recover the best possible (lowest noise level) luminance data. Color balancing amplifies (or attenuates) the noise in the red, green, and blue channels from which the luminance data is derived. Changing these noise levels before luminance data can be derived exacerbates noise. Even more so, if you choose 'Linear, was bayered, not whitebalanced', StarTools can properly weight your channels so that the influence of the bayer matrix on your CCD is taken into account for another small boost in luminance fidelity (because the green channel is twice more reliable than the other channels).
markcasazza wrote:Next is StarTools where I did this:
... Then in Photoshop I opened the file and did
A color balance with Levels

Color balancing of stretched data (apart from pure arbitrary aesthetics which can't be argued with) doesn't make mathematical sense and really isn't advisable (it will *never* lead to a 'correct' coloring in the sense of something that would be achievable through 'old school' photographic means). Instead, I would run the Color module before switching Tracking off, which will let you do color balancing as if the data was linear (Color recalculates the effect your stretching had). I highly recommend you have a read here on what's going on. If you don't like the Color Constancy color model which keeps colors and features comparable across the image (which is a more scientific way of rendering color and helping the viewer see more, but less 'photographic'), you can go for the more 'artistic' look that Photoshop users are more used to. You'll also notice a selection of different color composition ways that StarTools can emulate at the click of a button, without having to spend a lot of time setting up the layers in Photoshop.

A quick stretch; this revealed a overdoing of vignette correction (the edges were too bright) but I liked the look of the galaxy so I sent this back to StarTools.

To find this problem early on, use AutoDev to evaluate Wipe's output. Wipe should be really be used only once and while the data is still linear. At this point you have lost StarTools' ability to track noise propagation or any ability to recover real color.

Back in Photoshop I converted the file to LAB color and pulled out the Luminosity data

Unfortunately the L component has at this stage been polluted with the changing of colors and noise will be much higher than it needs to be (this is why StarTools separates color processing and luminance processing automatically all the way through and recombines color at the last moment). If you want StarTools to output a luminance-only (grayscale) image with just the CIE L*a*b L component. Run the Color module before switching tracking off and just set Saturation to 0%.

** Work on the eventual Lumonisity layer **
Created a new file in PS with the grey scale image
Did a standard Screen Mask Invert, the result was rather bright, but I figured I would try to get galactic detail in StarTools. The actions below were with a lasso mask drawn around the galaxy. the process really blew out the already bright stars.


If I read this flow correctly, the main reason to bring the StarTools result into PS, was to apply SMI.
SMI (and PIP/Power of Inverse Pixel) for that matter are very, very old (and crude) tools which have much better, more modern cousins (ex. the 'Brighten Dark' algorithm in the HDR module).
That said, if you really want to, you can replicate these effects using the Layer module.
Use of the Layer module while Tracking is on is best avoided (as it can gravely interefer with Tracking's ability to keep track of noise), but in the case of SMI or PIP it's ok.
It's best applied before running the Color module and swithcing Tracking off, after you've done all other operations (Decon, Sharp, HDR, etc.) to bring out your detail.

Here's how to do it;
  • Launch the Layer module.
  • Set 'Layer Mode' to 'Screen'.
  • Set 'Brightness Mask Mode' to 'Where foreground is light, use background'.
And believe it or not, that's it!
There is a slight modification to the SMI process, giving you more control, while keeping stars from blowing out. Instead of blurring the inverse mask, the mask is modulated by a non-linear response ('Brightness Mask Power' - in essence applying a gamma correction in PhotoShop to the luminance mask, instead of applying a filter). This keeps the stars from blowing out, while not destroying small faint detail. The luminance mask that is normally required is automatically generated for you and you can control its non-linear response with the 'Brightness Mask Power' parameter. You can also overdrive the effect by pushing Blend Amount beyond 100%. Because Tracking will keep tabs on the increased noise, you don't have to worry about the the SMI process increasing noise either; Tracking will take care of the increased noise prevalence once you perform final noise reduction.

Since we're discussing the 'golden oldies' of 90s image processing techniques :P, you might be interested in an alternative to SMI as well,called PIP (Power of Inversed Pixel). Simply switch the 'Layer mode' to 'Power of Inverse'. PIP is a bit milder.
Once you;re happy with the results, 'Keep' the result. You've just applied SMI in a few clicks, while you're still able to avail of Tracking's targeted noise reduction.

File loaded [C:\Users\mark\Pictures\Lights\IC 342\Processing\IC342 02-Lumonisity.tif].
---
--- Wavelet Sharpen
--- HDR
--- Wavelet De-Noise

Again, all these operations are best done when Tracking still has a chance to track noise evolution and color recovery.

If you'd like to share the data with me, I can show you what a PS-less workflow in StarTools would look like (including SMI if you wish).

Hope this helps!

Re: IC342 - second image with Star Tools

PostPosted: Tue Feb 04, 2014 12:15 am
by Rowland
This is an excellent post. LLRGB is an interesting process.

It's nice to see people thinking outside the square.

Nice image, Mark.

Re: IC342 - second image with Star Tools

PostPosted: Tue Feb 04, 2014 2:52 am
by gminder
Ivo,

This thread is a real keeper. I love it when you go into depth to explain a concept and to show how ST does just about anything that any other software will do.

Gary

Re: IC342 - second image with Star Tools

PostPosted: Wed Feb 05, 2014 12:20 am
by markcasazza
Ivo,

I agree, this is very helpful in figuring out how to do these things. This only my second image processed in any serious way with StarTools and I am very much still learning. You can tell that I am inserting ST into a process I have been using for a while. As I learn more capabilities in ST I will surely replace even more of that process.

I would be honored if you wanted to show a full ST solution; you can even use it in a future video if you think it would be helpful. I posted the just stacked file in my DropBox account where you previously picked up a file. Either way I plan to try what you have suggested. I have to agree everything you said makes sense.

Thanks for the great product support.

Regards,

Mark Casazza
http://casazza.net
Home of the Clear Sky Alarm Clock and Tonight's Sky

Re: IC342 - second image with Star Tools

PostPosted: Thu Feb 06, 2014 10:31 pm
by admin
Hi Mark,

I had a look at the data you sent me (thanks!).

It's really challenging data when it comes to color - very little color is recoverable. I'd love to know why that is - I've never seen data quite like this! Would you be able to share the data as it comes out of Nebulosity without doing a color balance? To make matters more complicated in the color domain, the chromatic aberration of your scope causes most detail to be either purple or blue. I also notice a strange curve going through the image - seems like some sort of secondary reflection in the optical train?

Regardless, luminance detail wise, it's fairly easy to achieve an image comparable in detail to the one you posted using said methods (HDR Brighten algorithm, instead of SMI). In fact, it was easy to achieve just using AutoDev, which does a great job brining out the faint arms all by itself (just put a Region Of Interest over the galaxy).

Re: IC342 - second image with Star Tools

PostPosted: Fri Feb 07, 2014 2:02 am
by markcasazza
Ivo,

That is the un-color balanced, just stretched, data! My preprocessing workflow is standard for Nebulosity: Apply darks, flats, and bias; normalize; debayer; align (I use 2 point alignment to adjust for translation and rotation); stack. Nebulosity does some stuff when it stacks that DSS does not. I'll just quote the manual:

During any of the Align and Combine methods, Nebulosity can mathematically stack
images in one of two ways. By default, an Adaptive Stacking technique is used (see
Preferences menu). Some people worry a great deal about whether to add (sum) or
average their frames during the stacking process. Each technique has its ups and
downs. If you have 3 images in which the same pixel reads 100, 100, and 101,
summing gives you 301, whereas averaging gives you 100. Internally, in Nebulosity, the
average would be 100.33333 (as it should be), but when saved, it would become 100
as the images are saved in "integers" (aka whole numbers, not "floats" which let you
have fractional bits as well). This makes one think that adding is best, but another
example shows the problem there.
Let's now say that a bright pixel reads 32000, 32010, and 32100 in our three images.
The sum is 96110 here where the average is 32036.666. When saved, this would
become 65535 if summing were used and 32037 if averaging were used. Let's have
another pixel - even brighter - reading 64000, 64010, and 64100 in the image. Once
saved, the sum would make this 65535 and the average would make it 64037. Here,
we see the problem with simple summing. You can saturate the image pretty easily,
especially if you start with 16-bit images. Here, one pixel should be twice as bright as
the other and yet it ends up equally bright (65535) if adding is used, since this is the
highest possible value.
Nebulosity uses an Adaptive Stacking technique that avoids the weaknesses of both. It
can be viewed as always being somewhere in between adding and averaging your
data. The output (the stack) will always have a maximum value of ~65535 so that you
are always using the full range of your data. This is enabled by default and for most
uses will be optimal. (Note, it is not used when the Fixed Combine is selected as this
tool is often used for dark frames). Unless you have a real reason to, you should leave
this on (see Preferences menu). If you turn it off, Nebulosity will compute a straight
average when stacking.


If you'd like to see the lights stacked another way, that is no big deal.

Yes, that scope has hot blues. The new scope will be ordered any day and that will be a thing of the past, but this one has been a great scope to learn the craft.

You are the first person to notice the huge arc. That is very unusual for this scope. I have not looked at all my lights to see if it was only in a few or in every one. I thought there was a very bright star outside the field, but I've ruled that out as well. I also assume some unusual reflection in the image chain. Happily that is the first and only time I've seen that kind or artifact.

I have yet to play with the image again. I think I will spend some time on Sunday with it.

Regards,

Mark Casazza
http://casazza.net
Home of the Clear Sky Alarm Clock and Tonight's Sky

Re: IC342 - second image with Star Tools

PostPosted: Fri Feb 07, 2014 2:18 am
by admin
Hmmm... :think: Most peculiar. Those Nebulosity settings shouldn't affect color.
Have you done any M42 images by any chance with the same gear? (for a color comparison)