Jump to content

NLCbanner2024.jpg.2478be509670e60c2d6efd04834b8b47.jpg

importance of post-processing

glowingturnip

By glowingturnip
in Getting Started With Imaging

Recommended Posts

glowingturnip

glowingturnip

Posted
Posted

Please forgive the self-indulgence and length of this post, but as a relative newcomer to the hobby, I thought I'd share my experience with just how important post-processing is.

I don't get out to my dark-sky site as often as I like, but I am a bit of a computer geek too, so enjoy the post-processing aspect as much as the actual telescopey bits, and am not averse to remastering something.  I use Maxim DL, DSS and Photoshop Express mostly.

I've been working with a 2 year-old image of the orion nebula - 5x 3min lights at 100 ISO (I know !), 1 flat (I know !) and 6 darks (more on those later), so not a great set of raw data, but at least my focussing and guiding seem to have been bang on (on this occasion at least).

So what follows is three iterations of my post-processing as my skills have developed, all from the same raw data and all of which I considered to be the 'finished article' and have spent time as the wallpaper on my office pc.

1)

I thought this was a great image, loads of detail, some lovely purple shading in there from that Hydrogen Alpha etc.  But wait a minute, it is *very* purple, isn't it - the stars are purple too, it looks a lot more purple than any other pics I've seen, it's even a lot more purple than the raw light frames, and there's a nasty step gradient in colour in the middle of the nebula where i've stretched one of the colour channels too much.  Hmm...  Turns out the Maxim DL Colour Conversion settings for my Nikon D80 were well off (should be scaling 100/120/100 in case you're interested), and have since learned how to colour balance properly anyway.

2)

That's a lot better !  more detail in the fainter parts of the nebula, can see a bit of the running man nebula, and the horrible colour cast is gone.  Still a bit of a dark background gradient in there though, expecially between the two nebulae, and I'm not over-convinced that is nebulosity. 

Actually, looking at the raw lights, they don't have that gradient either - turns out that all the darks have an ugly gradient in them, no idea how, maybe light was leaking into the camera when I took them (note to self, take darks when it's..  err..  dark).  Seems next to impossible to edit out a gradient from a dark file without debayering it and without losing all that hot pixel data, so was forced to use other darks taken at different ISO's and exposure times.  Maxim DL point-blank refuses to use a different ISO dark, DSS was willing to try, but the final result looked ugly but the intermediate calibration files from DSS weren't too bad though, so I used those and stacked in Maxim for the next image - worked quite well.  Still left a bit of 'amp-burn' showing in the corners and I had to be a little more 'creative' than I'd like to remove those, but they are fortunately in black background areas and have been mostly cropped away anyway.

I also figured out in the meantime how to deconvolve and use digital development processing in Maxim too, so...

3)

That's better !  Lovely detail in the nebula, can see the running man clearly, pinpoint stars, great detail in the dark nebulosity near the trapezium and M43 and I'm pretty sure that the dark background stuff there now is actually nebulosity and not poor processing.
I find it quite amazing the difference between these pics from the exact same raw data, which is why I thought I'd post this up.  Next step - start all over again with some decent data this time!
Cheers,
Stuart
P.S. In case you're wondering, I found this deconvolution script for Maxim to be much better than the built-in deconvolution - http://www.princeton.edu/~rvdb/images/deconv/deconv_MaximDL.html.  For the Digital Development Processing, I start off with auto for both background and midrange and no filters, then nudge the background number lower till the black pointer falls just below the histogram, and nudge the midrange slightly (usually upwards) until it looks nice.
P.P.P.S I'd be more than happy, time permitting, to help post-process other people's raw images if you don't have the inclination, or access to some of the Maxim tools.  I know there are plenty of other helpful people on this forum who would help too.
Link to comment
Share on other sites
  • Replies 29
  • Created
  • Last Reply
slickfonzy

slickfonzy

Posted
Posted

Fantastic, the difference between the three images is immense and the last one looks great.  I've only got as far as observing previously, and as someone who is returning to the hobby and keen to start imaging it's great to see how much detail can be brought out in post-processing.  I have a lot to learn!

Ian

Link to comment
Share on other sites
ollypenrice

ollypenrice

Posted
Posted

The process of learning that you describe has no ending. It is simply what this thing is all about. Making better and better pictures. The data came from the stars and if your guiding was sub pixel and you were in focus then what happens next is in post processing, and who knows where the limit lies? At present, nobody does. 

You will want to improve it still further, so 3 minutes is far too long for the Trapezium. Think 10 to 15 seconds. Then you'll want to explore layer masking methods of blending. And then you'll want to get after the faint outlying dust with very long subs. Etc etc. But this is the fun! Ive had about five serious pops at M42 (you have to! It's like Mount Everest for a mountaineer.) Each time I've thought, 'Got it,' but a wait of 18 months has been enough for me to see lots that needs improving. A little over a year ago I was happy with mine, but now I'm not. There is insufficient differentiation of colour in the faint stuff. At the moment I don't know how to deal with that but I hope to find out. When I know something new I'll have another go. And another, and another...  http://ollypenrice.smugmug.com/Other/Best-of-Les-Granges/i-rNfQT5R/0/X3/M42%20WIDE%202FLsV3-X3.jpg

The progress of your image is excellent and shows a very good eye. Bravo.

Olly

Link to comment
Share on other sites
Stevie P

Stevie P

Posted
Posted

thanks for the comments guys, really appreciate it.

But Olly, that picture of yours is astounding !  (bow)

Stevie, will have a look at home - am at work at the mo and the IT filter's blocking it

Hi

Thanks for taking a look

Details are here :-

http://stargazerslounge.com/topic/194615-would-any-kind-image-processor-have-10-mins-to-spare/

Steve

Link to comment
Share on other sites
  • 2 weeks later...
Russe

Russe

Posted
Posted

and here is a 2 session stack, calibrated with darks/flats/bias

56*300s + 81*300s at ISO 1600 with a Canon 450D & astronomik LP filter on a 130pds with coma corrector

The stack is not yet cropped, because I decided to go for the maximum available data and then cut it down in PS5.

https://www.dropbox.com/s/cwgtl46q4ltf0qi/M33%20session%209%20and%2010%2010%202013.TIF

Link to comment
Share on other sites
glowingturnip

glowingturnip

Posted
  • Author
Posted

Hi Russ,  here's my attempt at your M33, hope you like it ?  I'm quite pleased with it, though it is a little noisy:

post-30803-0-41086700-1382037391_thumb.j

I think it looks pretty good with a close crop like this:

post-30803-0-61623200-1382037407_thumb.j

The processed tif is here: http://www.stupey.com/Astronomy/M33_session_9_and_10_10_2013_processed.tif

I used the second image you put up, I could only get so far with the first one, because it looked a little 'damaged', for want of a better word - did you by any chance save that one from DSS with adjustments applied ?  That seems to garble the image a bit, better to save as embedded but not implied and do all your processing elsewhere.

Processing workflow used:

- initial levels and colour balancing in photoshop - using the colour histogram window and the levels window together, stretch the data to make it visible, but leaving plenty of space either side of the histogram for later steps, then using middle slider in each colour channel, get each colour histogram peaking in the same place, about 1/3rd along, ie manual colour balance.  tweak for colour aesthetics.  Bit of an epiphany for me actually, doing this at this stage preserved a lot of detail i was losing in the next stages, where some of the colour data to the far left of the histo was getting severely clipped (my first try with this image I lost all of the red channel !), explains a lot of the colour balance issues I've been having

- gradientxterminator

- levels again

- deconvolve in maxim

- ddp in maxim (just a fancy stretch)

- further curves boost for midtones, they didn't really pop even after ddp

- layer treatments in photoshop, each layer split into luminance and colour layers, all colour layers blurred and colour noise reduced:

  - background layer, saturation -30, lower contrast and levels in luminance - couldn't blur at all though, there are many faint stars in there that just winked out as soon as I touch the blur button

  - star layer - saturation 2x +20, bit of unsharp mask, levels to brighten them

  - galaxy layer with blurred galaxy mask, saturation +15, high pass overlay layer, 5px, 50% opacity (much prefer sharpening like this that unsharp mask, I think that's always too harsh), contrast to bring out the spiral arms

- flattened image again, and various levels and other fiddling

- curves to drop the very low end a bit to kill some of the noise

- 24hr eyeball filter (very important)

- very light surface blur, noise was (and is) still quite heavy

Link to comment
Share on other sites
Russe

Russe

Posted
Posted

Processing workflow used:

- layer treatments in photoshop, each layer split into luminance and colour layers, all colour layers blurred and colour noise reduced:

- background layer, saturation -30, lower contrast and levels in luminance - couldn't blur at all though, there are many faint stars in there that just winked out as soon as I touch the blur button

- star layer - saturation 2x +20, bit of unsharp mask, levels to brighten them

- galaxy layer with blurred galaxy mask, saturation +15, high pass overlay layer, 5px, 50% opacity (much prefer sharpening like this that unsharp mask, I think that's always too harsh), contrast to bring out the spiral arms

Very nice result. You managed to get the outer fainter region much nicer than I did!!!

The 4 steps above I'm not sure how to do. Do you have a link to a tutorial?

Thanks a lot for your great effort!

The first pic was as unchanged as the second. But, the second has twice the amount of data from the second session. Or is that the one you meant?

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

Go to topic listing

  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...

Important Information

We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue. By using this site, you agree to our Terms of Use.