Username:

Password:

Altair AstroDIO DehumidifiersAtik CamerasModern AstronomyDavid HindsNe3 Filters
Solar Gravity 2018

Author Topic: Shooting and processing DSLR images using DSS and Photoshop  (Read 32112 times)

0 Members and 1 Guest are viewing this topic.

Offline dciobota

  • Daniel
  • Galactic Poster
  • ******
  • Posts: 9722
Shooting and processing DSLR images using DSS and Photoshop
« on: 07:08:29, 04 December, 2007 »
EDIT:  Many thanks to Leigh for converting the whole article written below to a pdf file.  If there are images missing below or parts missing, get the pdf from here:
https://docs.google.com/viewer?a=v&pid=explorer&chrome=true&srcid=0ByZXuV0vUfw7Nzk1ZTJmNmItOWRjZS00OGYxLWFiMzEtYzFjMzkyN2IyYjdl&hl=en_US

NOTE:  The pics look pretty awful when viewed in google docs from the link.  Use the download button to download and view the file in your pdf viewer, looks much better.


 After over two years of shooting deep sky objects with dslr's from my back yard, I think I finally figured out how to get decent images out of them.  I have been asked several times recently to help with processing and describe a typical workflow from shooting the image to making the final result.  So, in this thread I'm going to start describing my workflow, and hopefully others who plan to use dslr's and have to work with fairly light polluted back yards can do as well (or better) than me in producing decent images.

 First off, I'm not an imaging guru, nor am I a processing guru.  I view myself as a pretty average Joe with average skills... I sincerely believe that anyone starting in this hobby can do this, and I encourage all who read this to keep at it.  May take a while, but hopefully after reading this, not as long as it took me. ;)

 I will also not talk about telescope kits, guiding and the like.  Everyone has a preference in that regard (and different budgets as well).  For this article, I will only talk briefly about camera settings and the actual taking of subs.  Since I own canon cameras, it will be related to them, but setting things such as bulb exposures, iso and white balance pertains to all dslr's, so no matter which camera you own, this should be useful to you as well.

 Ok, onto the article.  Btw, this is the rough workflow I used to produce the Horsehead and Flame I posted earlier today.

 Before I talk about taking subs, I want to mention filters, because they do affect the image processing and the taking of the subs themselves.  I use the astronomik CLS, but other filters work just as well.  Actually, depending on the amount and kind of LP you have, you should choose the appropriate LPR filter.  The CLS is for mild to moderate LP, while the IDAS LPS (another favorite) is for mild LP.  

 To give you an idea of the LP I deal with, the first attached pic is a 30sec iso1600 exposure of the Alnitak area, using a Canon 400D and a scope running at about f5.4.  No filter.  As you can see, lp is creeping in already, washing out the nebulas.  The second attachment is a sub from today's posted pic.  It's a 10min iso1600 exposure, from approximately the same area of the sky and about the same f ratio (f5.9 in this case).  It's also with the 40D, which has a very similar sensor response to the 400D I used last year.  This sub however has the cls filter installed. At this point, a different type of lp is creeping in, due to the white lights used at a plastics plant nearby.  The cls filter cannot deal with the lp from those lights unfortunately.  :(

 At this point you notice that the nebulas in the 10min shot are really showing not much better than the unfiltered 30sec shot.  But, the lp in the first shot is much harder to subtract than the second, especially for Ha emitting nebulas.  Check out the third image, it's a version of the second one with the background lp subtracted.  Notice how much clearer the horsehead appears now.  It was impossible to get that kind of detail from the unfiltered image.

 So, now you know how beneficial lp filters are, and also even more importantly, how much data resides in a filtered image, even though it looks totally washed out.  Believe me, all the subs taken for the horsehead pic were pretty much exactly the same as the one posted here.  Now, onto part 2, camera settings.







PLEASE REMOVE ME

Offline dciobota

  • Daniel
  • Galactic Poster
  • ******
  • Posts: 9722
Re: Shooting and processing DSLR images using DSS and Photoshop
« Reply #1 on: 07:41:18, 04 December, 2007 »
  Ok, part 2, camera settings.  This will be fairly brief, but it is important.

  Firstly, every camera has a "sweet spot", since iso, dynamic range, and exposure length are all related.  For example, most Canon cameras maintain their best dynamic range (that is the span of signal between just above noise level to total whiteout, or blowing out as it's called) from iso100 to 400 or 800.  So, traditionally, most of my shots have been at iso400 or 800.  At that point, exposure is determined by how much background lp you're willing to allow in your subs.  The higher the background brightness, the less dynamic range, since the background lp is part of the "noise" of the image.  

 But you don't want the background to be totally dark either.  Why?  Because at that point camera noise takes over and destroys the fainter details.  I usually choose my background brightness point depending what I shoot.  For nebulas, which are mostly Ha, I can let the background be fairly bright, since the color is not within the Ha band.  For galaxies, which cover most of the spectrum, it's more difficult.  I choose a fairly dark background at that point (about 20% of total white) and hope for the best.  Galaxies are always very tough for me.  In practice, the exposure works out to about 10mins at iso400 or 800, depending which section of the sky I image.  First attachment shows a 10min iso800 sub I took of the Sculptor galaxy (never posted the finished pic, it's really kinda sucky).

 So, iso from 400 to 800, exposure 5-10mins (you can get by with less, especially if you increase iso setting).  I also shoot raw+large jpeg.  I use only the raws when I stack, and I use the jpegs to verify tracking, focus and other exposure problems.  Very handy if you have the disk space.

 Last setting, white balance.  I find two options that work equally well, but in all cases, it is imperative that you take flats.  In the case of filters, flats not only correct for things like vignetting and dust, but color balance as well.  But you must use the exact same white balance when you take the subs and the flats as well.  I'll discuss taking flats in the next section.  Anyways, the two white balance options that work best are daylight and custom.  As far as custom, you can either leave it without a calibration image (in which case it assumes the camera default grey calibration), or, what works best with modded cameras, use a custom calibration image made with a grey card.  Like I said though, do the flats using the exact same white balance setting or the colors will come out unbalanced after stacking.

 That's it really for camera settings.  Next, I'll talk about taking flats.


PLEASE REMOVE ME

Offline dciobota

  • Daniel
  • Galactic Poster
  • ******
  • Posts: 9722
Re: Shooting and processing DSLR images using DSS and Photoshop
« Reply #2 on: 07:43:12, 04 December, 2007 »
  Man, I didn't realize how late it is... will continue as soon as I get the chance.  Stay tuned!
PLEASE REMOVE ME

Offline dciobota

  • Daniel
  • Galactic Poster
  • ******
  • Posts: 9722
Re: Shooting and processing DSLR images using DSS and Photoshop
« Reply #3 on: 07:04:36, 10 December, 2007 »
Ok, I'm back.  :)

 Taking flats.  Very easy to do with a dslr, imo.  This is the way I do it, but it's by no means the only way.

 First, pick a nice sunny day... heck, pick a nice cloudy day, makes no difference really as long as it's not raining or something like that.  All you need is a uniform patch of sky, clear of trees, power lines, thing like that.

 Second, you'll need to have the scope set up exactly the same as you would when taking subs.  That means same focuser position, adapters and filters.  

 Third, you need a way to diffuse the light coming in through the objective.  The sky may have clouds or small non uniform variations, so a diffuser will make it as uniform as possible.  You may use a light neutral color material, just as long as there are no creases in it or imperfections that might affect the flat.  Some use a white t shirt, I prefer a sheet of typing paper folded in thirds, with the middle part over the objective and the outer parts holding agains the dewshield... a U shape basically.  Works pretty well, and to counteract winds I sometimes use a rubber band.

 The, set the camera to Av mode (that's where the camera automatically adjusts the shutter speed).  Set the iso to 100, and metering mode to average.  This ensures that the camera will expose for the entire illuminated field.  Oh, VERY important, make sure the white balance mode you use is exactly the same as the one used to take subs... that is very important, to prevent odd colors or color variations in the final image.  Make sure exposure compensation is set to 0 (the exposure meter is on the middle graduation), then click a shot.  I take both raw and jpeg calibration images, just in case.  Some people recommend several flats, and it can't hurt to do that (it reduces possible noise).  I'm lazy, I just take one, since iso100 is pretty darn clean to start with.

 That's it!

The flat, in DSS, accomplishes two things.  One, it adjusts the image for optical issues like vignetting and dust specs.  Two, it also corrects for color offsets.  How?  When you take an image through say, a blue filter, the resulting sub has a blue tint.  Witness the Horsehead sub above taken with the cls filter.  Taking a flat of a uniform diffuse field with the same filters not only shows vignetting, but also shows the same color bias as the subs.  So, when DSS divides the flats, not only is vignetting corrected but the color balance as well.  Cool, huh?  :)

 For a typical flat, see the image below.  ALso note that a proper flat will have values centered around the middle to lower third of the histogram.  A little slop is ok, but for severe vignetting, it's best to be close.  On most cameras you can check the histogram after the shot is taken to make sure it came out ok.  If not, you can vary the exposure compensation from 0 until the shot histogram is centered correctly.

 Ok, now you have flats.  You can use them any time, for any subs taken with the same setup.  Flats are temperature and iso independent.  However, some folks experience non symmetric vignetting and/or sagging in the optical train.  In that case, I strongly recommend you take a set of flats not only with the same setup, but with the camera in the exact same orientation as well, to account for those conditions.  Same with dust bunnies btw, they can indeed shift from session to session.

 Now, onto the meat of the article, workflow with DSS.


PLEASE REMOVE ME

Offline dciobota

  • Daniel
  • Galactic Poster
  • ******
  • Posts: 9722
Re: Shooting and processing DSLR images using DSS and Photoshop
« Reply #4 on: 00:58:36, 06 January, 2008 »
  Lol, this is starting to look like a Harry Potter writing project.  ;)

  Ok, using DSS!  

  First off, the version I'm currently using is 3.1.0, so that's the version this article will pertain to.  Future versions may have changes and improvements that may make it work differently, so bear that in mind.  I may update this article if there are significant changes and of course, if I have the time.  Also, bear in mind that I use an unmodded canon dslr (40D) and an astronomik cls filter.  Technically, it shouldn't matter what filter(s) you use nor which dslr you use, provided that you followed my advice regarding the taking of flats.  That, imo, is an essential part of making DSS work properly.  The majority of issues people report about DSS are usually color related, and the proper use of a flat along with the correct use of color balance settings in DSS will address most if not all of those issues.

 Ok, I'm assuming you have basic knowledge of DSS and have read at least the online manual provided, so you are familiar with the various popup dialogs.  There are really two that are most important, the raw/dpp settings and the stacking parameters settings.  The first one is accessible through a link on the main panel (lower left side), while, oddly, the stacking parameters are only accessible from the stacking popup dialog when you register or stack the images.  Once set though, the options are retained in subsequent runs of the program.  I only set them once, and rarely change anything afterwards.

 The first pic shows the raw/dpp settings.  Notice that I use camera white balance.  This is important.  As I mentioned before, I shoot the flats, the lights and whatever else using the exact same white balance setting, then instruct DSS to use that same setting (it's extracted from the raw files) when calibrating and stacking.  Cannot stress enough how important this is to avoid odd colors in the final image.

 Also, I use AHD for color interpolation.  It is not critical you do, but I find it produces more accurate star shapes towards the edges.  This is especially true when a lot of field curvature is present in the image.  But, bilinear is faster (much faster), so if you're strapped for time or processing power, bilinear does almost as well.

 Brightness setting.  If you have low signal subs (very dark), the stars in it may be too dark for DSS to detect for registration and alignment.  If you find that DSS rejects too many of your subs, it may be due to this (or the star detection threshold set too high).  One remedy is to increase the brightness multiplier from 1.0 to something higher.  This multiplies the entire brightness of the image though, so be careful not to burn out important parts of it.  Alternatively, you can lower the star detection threshold value (in the Registration dialog, the Advanced tab)... just keep in mind that setting it too low can start picking up noise as stars.

 Ok, next section, the stacking parameters.


PLEASE REMOVE ME

Offline dciobota

  • Daniel
  • Galactic Poster
  • ******
  • Posts: 9722
Re: Shooting and processing DSLR images using DSS and Photoshop
« Reply #5 on: 01:25:12, 06 January, 2008 »
  Ok, stacking parameters.
  
  I only set three tabs in the stacking parameters dialog.  Believe it or not, I don't use bias or dark subs.  I messed with both of them, and imo, for the length of subs I shoot (10mins) and spread over several nights with varying temps, I used to shoot about as many darks as subs... so, I used the in camera long exposure noise reduction, which not only corrects what darks would (hot pixels, amp glow) but also reduces noise in the sub itself.  And this brings me to note an important point: darks DO NOT help in reducing noise.  In fact, unless you use quite a few (10-20 is recommended), they actually add to the noise in the final image.  This pertains to dslrs of course.  Dedicated ccd's may work differently in that regard.  Again, this is just an empirical finding of mine over the years I've been using dslrs (from the days of the 300D actually).  As to bias frames, I never found a benefit nor liability from using them, so I just decided to keep things simple and not use them.

 Ok, onto the tabs.  The first one is the result tab.  I use the mosaic setting.  Why?  Because sometimes it's hard to pic the best frame to stack against.  Mosaic lets me see the entire area I've covered with my subs, letting me crop the best section.  It does use more memory and cpu time though.  Also, I checked the rgb channels alignment.  This was a feature I asked Luc (and many many thank to him for the prompt implementation); it aligns the color channels after stacking is done, in the final pic.  This is important if you have what is called lateral aberration, where the stars towards the edges show misalignment between red green and blue.  Checking the box corrects it nicely and takes very little extra processing time.

 The second tab is the light settings tab.  I use median-kappa since it seems to give me better detail than straight average or median and also is very good at removing hot pixels (or smoothing dead pixels).  I also check the rgb channels background calibration, to neutralize the background color.  If you use the other one (per channel) it may give you an odd hued background in the final pic.  The settings I use always give me a totally neutral grey background.

 The third tab is the alignment tab.  This is the one that determines how to stretch each sub to fit over the others.  It is especially important if your optics have distortions, like astigmatism or field curvature.  Choosing bicubic make sure that you get the most accurate alignment, provided the subs have enough stars.  If not, then depending on the number of stars, it uses one of the less accurate methods.  So choosing bicubic is like a priority setting for stacking alignment accuracy.

 Ok, next, DSS registration and stacking workflow.  :)






PLEASE REMOVE ME

Offline dciobota

  • Daniel
  • Galactic Poster
  • ******
  • Posts: 9722
Re: Shooting and processing DSLR images using DSS and Photoshop
« Reply #6 on: 01:52:09, 06 January, 2008 »
  Workflow.  This one's easy, just a few comments needed.

  First off, I load lights and flats.  Since I only use one master flat for a particular setup, I don't need to load the lights into separate groups.  But, if you shoot flats with every session, and you have sessions extending for several nights - or, sets of lights and darks that go together - then it's best to split them into groups.  Just click on the next empty group tab and load them in there.  DSS takes care of calibrating and stacking them properly.

  Also, if my sessions extend over several nights, I usually save the file list.  That way, the next night I load the previous files list add the new lights, then save again.  It really saves the hassle of reloading everything, especially if you have groups.

 That's it really.  After I load all files and click on check all, I go to register, I choose to stack 100% of all subs (if you have trailing or subs with clouds, you may want to choose less than 100%), then press OK and watch the show.  ;)

 When DSS is done and display an image on the screen, I first note where it saved the autosave.tiff file (if you loaded data from several folders, it may vary where it saves the tiff file).  The I just close DSS.  I don't bother to do any adjustments or save what's on the screen.  The autosave.tiff is the only file I post process. 

 Oh, one other thing.  The autosave.tiff file is a 32bit rational tiff file.  Most programs can read it or convert from it.  Earlier versions of photoshop though can't (I think photoshop 7 can't, not sure).  In that case, your next best bet is to save as a 16bit tiff, which loses a bit of data from the stacking.  To save it that way, you can load in the autosave.tiff into DSS by clicking on the Load picture file... on the main panel (middle), then when the image is finished drawing on the screen, click on Save picture to file... and choose 16bit tiff as the format.  There is a toggle box wher you can choose to either apply or embed changes... the changes mentioned are a default brightness curve that DSS applies.  My recommendation is to choose embed instead of appy.  That way the file data is exactly as it came from stacking, and you can apply your own curve in post processing.  Ymmv though.

 Ok, next a very rough workflow in CS (CS3 in this case).
PLEASE REMOVE ME

Offline dciobota

  • Daniel
  • Galactic Poster
  • ******
  • Posts: 9722
Re: Shooting and processing DSLR images using DSS and Photoshop
« Reply #7 on: 02:32:53, 06 January, 2008 »
 Photoshop workflow.  Note that this pertains to CS3, but the older photoshops can do some of these steps as well.

 This actually incorporates parts of a post I made to Leigh's questions, so I chose my M42 image as the example for post processing.

First image is the original 32bit tiff brought into photoshop.  I just cropped and resized for this example.

 Then I ran autolevels. The result is the second image.  Notice it blew out the core more (mine was already blown to start with).

 Then, I selected local adaptation when converting to 16 bit mode.  Notice the tone histogram has a hump in the middle, and the conversion "curve" is actually a straight line running across from lower left to upper right.  What this means is that the bulk of your image data now sits in the middle of the histogram.  Bring the lower left corner of the conversion curve to where the tone histogram starts, at the foot of the "hump".  You will have to play a bit because that sets the black point of your background basically.  Don't make it totally black, a nice dark grey is good... this will allow a bit more play in the post processing later.  Now, the curve will depend on how much contrast you want, which parts you want more contrasty than others, and how much highlight you want to preserve.  The upper right of the curve is where you control highlights, clicking somewhere close to the upper right part of the line and pulling down gently will recover them.  If you want to brighten up the fainter bits and make them more contrasty, click somewhere along the lower end of the curve and pull up to brighten, down to darken.  You'll really have to play and get it right to your taste.  The third pic shows the result on mine, I curved the tone curve downward a bit to recover highlights, and set the black point right at the beginning of the histogram (where it just starts to rise).

 So now you have a fairly decent image to work with.  At this point, the workflow will largely depend on the amount of gradient in the pic, noise, artefacts, etc, so it's really hard to generalize.  I usually use a very few tools and actions:

 Noel's:
 Light pollution removal
 Select brighter stars
 Increase star color
 Local contrast enhancement

 Tools:
 Curves
 Smart sharpen
 Gaussian blur
 Surface blur
 hue/saturation

 Only rarely do I use anything else.  I also use layers, selections and masks.  There are really many many tutorials for post processing, my skills are nowhere near to those guys.

 Well, that finally concludes this article.  I hope it helps, please post any questions or point out anything I might have missed.

 Clear skies,

Daniel






PLEASE REMOVE ME

Offline spaismunky

  • Galactic Poster
  • ******
  • Posts: 7302
  • Space is ace-not bobbins
Re: Shooting and processing DSLR images using DSS and Photoshop
« Reply #8 on: 02:46:21, 06 January, 2008 »
Fabulous! :big_clap: :big_clap:
It's like watching a suspense. :lol:
mojo :D
Meade LX90 10"

Offline dciobota

  • Daniel
  • Galactic Poster
  • ******
  • Posts: 9722
Re: Shooting and processing DSLR images using DSS and Photoshop
« Reply #9 on: 05:58:40, 06 January, 2008 »
 Lol Mo, I should've made weekly installments with a cliffhanger at the end.  :D

PLEASE REMOVE ME

Offline minkus

  • Starting Member
  • **
  • Posts: 57
Re: Shooting and processing DSLR images using DSS and Photoshop
« Reply #10 on: 09:29:13, 06 January, 2008 »
Nice one Daniel. Very helpful.   :D

I was quite surprised about your comments regarding the use of darks - I'll have to experiment with my 350D.

Cheers

Mick
Pulsar 7' Obsy, Losmandy G-11, Celestron C11, WO Megrez 80 FD, Canon 350D, Meade DSI Pro, other bits and bobs.

Offline dciobota

  • Daniel
  • Galactic Poster
  • ******
  • Posts: 9722
Re: Shooting and processing DSLR images using DSS and Photoshop
« Reply #11 on: 09:45:17, 06 January, 2008 »
 It does indeed go against the grain, everyone else advised me against it.  I would be very interested to know the results of your testing, whichever way it goes. 

PLEASE REMOVE ME

Offline refractions

  • Starting Member
  • **
  • Posts: 85
Re: Shooting and processing DSLR images using DSS and Photoshop
« Reply #12 on: 13:36:31, 06 January, 2008 »
Hi,

First of all many thanks for the fantastic advice.  Now then, can i be a bit cheeky, and ask - is there any chance you could put all this into a word document or a pdf?

Iain
As I look into the night sky at the stars I cannot help but ask myself, Where's the roof gone?

LX200 12", Revelation APO, Canon 300d, 350d - both Ha mods. ROR obs.

Offline Stuart

  • Poster God
  • *****
  • Posts: 1266
Re: Shooting and processing DSLR images using DSS and Photoshop
« Reply #13 on: 20:18:02, 06 January, 2008 »
Excellent, I'm sticking this al down on a word file for reference and I'll try it out on my data from the days of clear skies.

Thanks Dan (you already got your k for this when you sent me the pm by the way)

Offline koootzz

  • Administrator
  • Poster God
  • *****
  • Posts: 2948
Re: Shooting and processing DSLR images using DSS and Photoshop
« Reply #14 on: 20:25:08, 06 January, 2008 »
Excellent write up, thanks Daniel  :urock:

Now I just need to do it justice and get myself a decent image  :D

Leigh
Leigh
vixen 60s, modded 1000D, modded 300D, DSI I, 10" dob (work in progress), some other bits and a stupendous talent for attracting cloud!

 

ukbuysellRemote Imaging from AustraliaSharpSkyblank APTUKAI on Facebook
TinyPortal © 2005-2012