PDA

View Full Version here: : Does anyone want to try processing my DS FITS


kinetic
31-03-2011, 07:23 AM
Hi,

I've processed a data set for the last 2 days and thrown my entire
usual regime at it. The results are pretty good, one is my best result of
NGC 4945 so far as a luminosity shot, for example.
But I know that there is much more in this data set and the limitation
staring me in the face is me. :)

I'd really appreciate if anyone could have a quick tweak, using the
RAW cropped result FIT as a starting point and see what they can
retrieve from it.

I think this has been done or requested a few times before for planetary
results but I couldn't find a Deep Sky example.

If I was to upload the cropped Pre-adjusted FIT to the upload section
(about 1.2Mb Edit: oops 7.5Mb) it would probably be a very educational exercise for
not only me of what could be achieved with a data set by several
approaches, maybe?:shrug:

I'm at work right now and don't have the FITs on board this laptop but
if it is a positive response I will attempt to uplaod them once at home
tonight.

also, please mods put this in the appropriate forum if this doesn't
belong here.

Steve

irwjager
31-03-2011, 03:17 PM
Hi Steve,

Looking good! I'm sure there's a few people who would give the reprocessing a go (myself included :) ).
Let us know!

Ivo

Octane
31-03-2011, 03:34 PM
In for the win.

H

rcheshire
31-03-2011, 08:39 PM
Sounds like fun!

kinetic
31-03-2011, 08:53 PM
Ok, thanks guys,

hopefully this has worked....

I have uploaded a zip file with 3 files in it here:
http://www.iceinspace.com.au/uploads/4945.zip

result_best440fr_1_6x_crop.fit
This is a stack result only cropped, no further processing.
This stack is each of the 440 best frames Bspline upsampled 1.6x after
dark subtraction and bad pixel removal.

result_best440fr_1_6x_crop Deconvolved30.fit
Same as above but a deconvolution applied in CCDStack with 30 iterations.

result_best440fr_1_6x_crop Deconvolved120.fit
Same as the first file but 120 iterations deconvolution.

My attachment at the start of this thread is a result JPG that
has the following processing applied:

levels adjustment
gentle first curve, highlights ending linear.
2nd gentle curve, same
levels adjusted for black
aggressive 3rd curve
levels adjusted finally for black.

Laplacian sharpen.
Saved as 16bit BMP.
Edge preserving smooth noise reduction in PSP

resized 75% down
Edit: forgot 180 rotated to make North up
saved as JPG

Steve

irwjager
01-04-2011, 12:41 PM
My quick effort, all processing done in StarTools.
Work flow looked something like this;

* Cropped stacking artifacts.
* Digital development (e.g. 'levels') and normalization.
* Low-level noise filtering (median filter with luma mask)
* Wavelet sharpening, keeping 1st octave as-is (I prefer using a different sharpening algorithm for the smallest details).
* Blended back in some detail that was lost due to clipping (because of Wavelet sharpening).
* Blended sharpened result with non-sharpened result (because I overdid it :P )
* Fractionally binned the result to 71% gaining some noise reduction. No discernable loss of detail as the raw image seemed slightly oversampled for the focus/atmospheric conditions.
* Rounded stars, getting rid of the egginess (new module to be released soon).
* Sharpened smallest details while keeping stars masked out.
* Tightened stars and made them less pronounced.

All-in-all 15 minutes work.

I used the non-deconvoluted image as it seems the decon just added grain and no further detail.

Looking forward to other people's interpretations!

DavidU
01-04-2011, 01:12 PM
Nice & tight...

kinetic
01-04-2011, 04:28 PM
Thanks Ivo, that has certainly got my attention.:)
Very nice, thanks for the processing details.

Steve

gregbradley
01-04-2011, 05:44 PM
I had a go at this one. You got some nice data there.

Greg.

kinetic
01-04-2011, 06:34 PM
Thank you Greg, that is pretty impressive.
I don't even have to be 10ft from the monitor :)
What routine did you apply to it?

gregbradley
01-04-2011, 06:51 PM
0. I aligned and median combined the 3 files you had using CCDStack. I ran data reject on hot/cold pixels and interpolate rejected pixels.
It seems to make an image a bit freer of artifacts although perhaps just a habit.

1. I deconvolved the image, 40 iterations - deconvolution is easily overdone.

2. levels and curves to bring it up to view. Keep the last part of the curve straight so as not to blow out the highlights.

3. Gradient Xterminator to reduce the slight gradient.

4. Clone tool set to 25% to get rid of that blotch in the background left side - maybe a dust donut from the camera.

5. I did soft light layer at about 25% to darken the background a tad and increase contrast.

6. I did selective sharpening/contrast boost by duplicate layer set to soft light, high pass filter set to maximum (not just 7 pixels but 100)
then layer/layer mask/hide all blended the sharpened increased contrast layer in with a white brush set to about 25% or less as the enhanced layer was quite bright.

7. I then set opacity of that layer to about 70% and ran a gaussian blur of about 20 pixels to the mask to prevent sharp boundaries.

8. I then duplicated that layer twice to enhance the effect more.

9. I flattened the layer.

10. I used an inverted mask to apply Noise Ninja only to the dim areas.
I used a Noel Carboni reduce deep space noise action to reduce background noise.

11. I used the blur tool to smooth out some noisy areas at the edge of the galaxy that were still noisy.

12. I used the smudge tool to nudge a few of the eggy stars to be more round (about 5 or 6 of them).

13. I used the sponge tool set to burn to enhance the dark regions of the galaxy and set it low.

14. Somehow the image was 180 degrees back to front so I rotated it.

15. I cropped the image as there was a stacking black edge on some of the sides.

16. A very slight curves to boost it a tad (hardly at all).

17. Reduced the file size and set it to 8 bit image and save as a jpeg, 5 times scanned and maximum resolution.

Thats it. Sounds like a lot but I suppose it took 20-25 minutes to do. I am pretty used to these tools though.

I tried a gradient removal technique I sometimes use but its fussy and there is something about it I don't 100% understand so I need to restudy it but sometimes it is useful if you have a blitchy background.

Processing goals were to bring out the detail in the galaxy whilst keeping the stars nice and soft without creating harsh stars and reducing background noise whilst not softening the look of the image.

I tend not to use sharpening tools like unsharp mask or smart sharpen as they too often enhance noise and damage stars.

Greg.

kinetic
02-04-2011, 07:50 AM
Thank you Rowland, Greg, Ivo, all.
My initial sequence is using mostly what Louie (Atalas)
has setout in one of his online tutorials. (http://atalas.businesscatalyst.com/Levels_and_Curves_for_Photoshop.htm l)
He suggests first gentle curves and levels then a more
aggressive one.
The thing that I cannot really understand is what shape curve
to apply according to the subject matter.:shrug:
Obviously, my effort has made the tonal range fairly bland
and not very contrasty, and all of your efforts look way better
with that.
My aggressive curve is typically the 'quickmega' shape in Nebulosity,
fairly steep down near the black.:shrug:
I love the noise reduction in all of the attempts, way better and smoother:thumbsup:
This gives me a lot of hope with processing my backlog of 'best ofs'.

I wish I had given you guys a better looking galaxy to try out on.
She certainly is an ugly galaxy:) Imagine being told you live in
an ugly galaxy.

Thank you guys so much for the efforts:thumbsup:

Steve

rcheshire
02-04-2011, 08:19 AM
Sorry Steve. I removed my first attempt as you were replying - we crossed paths. Here it is again. Not sure what you think, this is the work of wavelet manipulation and a little deconvolution, exponential transformation and curves. It's actually very beautiful.

irwjager
02-04-2011, 11:58 AM
Wow. I'm inclined to say you win this round Rowland :)
The galaxy looks a bit 'fluffy' or hairy though in places - is that true detail or is it the result of some sort of noise reduction?

rcheshire
02-04-2011, 03:28 PM
It's starting to look like Pizza. I think morphological transformation might be responsible for the fluffy look.

DavidU
02-04-2011, 04:47 PM
Steve, is the camera a DS 11 mono?

scopemankit
02-04-2011, 07:49 PM
my effort using paint shop pro 5

kinetic
03-04-2011, 09:48 AM
Chris did you try the FITS as a starting point?

Everyone's results shown below.
No resizing done....clicking through the images like a slideshow,
the transition between sizing isn't obvious too much.

Steve

scopemankit
03-04-2011, 08:43 PM
No, Sorry.

Here is a new one starting with Fits LIberator and progressing to Astro art both using fits. To Convert to JPEG I used a full screen mode and took a screen print into PSP.