Jump to content


Photo

Just can't figure out post processing

  • Please log in to reply
30 replies to this topic

#1 sternenhimmel

sternenhimmel

    Vostok 1

  • -----
  • Posts: 102
  • Joined: 05 Jun 2013

Posted 06 August 2013 - 04:29 PM

I've read many tutorials and watched many videos on how people post process their astro images. However, following these techniques I always seem to get a very unsaturated result with lots of noise. I've tried the curves/levels method, alternating between a high pass and adjusting the black point. I've also tried stretching the data just using subsequent level adjustments. I've also tried variations on these techniques and they all seem to produce noisy images with little color data. I then have to bump up the saturation which introduces a lot more noise.

Frankly, it could just be that I lack enough data, or that i shot at too high of an ISO and the noise is a direct result. I've attached the stacked image (TIFF) if anyone wants to take a stab at what little data might be there.

It's a composite of:
23 x 120 second ISO 1600 frames
21 Darks, 20 bias
14 x 120 second ISO 800 frames
8 Darks
1 x 300 second ISO 800 frame
2 Darks

Total time: 1 hr 18 min

UPDATED AGAIN Unaltered Stacked 16-bit TIFF:
https://skydrive.liv...thkey=!APVix...

The one success I've had is in gradient removal, but the uploaded file is unaltered.

#2 Falcon-

Falcon-

    Aurora

  • *****
  • Posts: 4932
  • Joined: 11 Sep 2009
  • Loc: Gambier Island, BC, Canada

Posted 06 August 2013 - 04:56 PM

I notice the file is described as "with settings applied" I am guessing you are using Deep Sky Stacker. Would you be able to provide a stack TIFF with the DSS adjustments *not* applied?

Generally speaking DSS's adjustments are not advantageous, they may be why (or part of why) colour is so suppressed.

Edit: Also, could you tell us what lens, what f/-stop the aperture was set to, and what camera you used?

#3 sternenhimmel

sternenhimmel

    Vostok 1

  • -----
  • Posts: 102
  • Joined: 05 Jun 2013

Posted 06 August 2013 - 05:07 PM

The camera was a T1i, unmodded. The lens was a 24-105mm f/4L stopped down to f/5.6 and soht at 105mm.

The file shouldn't have any settings applied. I saved two versions, one with adjustments to the luminosity and color channels, and one where I reset these values and saved the stacked file, which is the file I uploaded. Perhaps it says "with settings applied" because I reset the values?

#4 sternenhimmel

sternenhimmel

    Vostok 1

  • -----
  • Posts: 102
  • Joined: 05 Jun 2013

Posted 06 August 2013 - 07:33 PM

To clarify, I'm not looking for someone to do the work for me. I'm happy doing the post processing and I'd love to get a grasp and an intuition for it. I'm really just looking for someone to take a quick look and give me an idea of where my realistic goals for this photo should be.

#5 frito

frito

    Apollo

  • -----
  • Posts: 1183
  • Joined: 05 Oct 2012
  • Loc: Fremont, CA

Posted 06 August 2013 - 09:30 PM

I took a go at it, not going to put much time into it, your data is not linear, you need to upload the autosave.tif 32 bit file DSS outputs automatically, anything you save manually in DSS is modified, that being said I no longer use DSS because Pixinsight does a better job with stacking and pre-processing. I could not get any decent color data out of the file provided and its very noisy, Noise reduction is not easy and is best done via methods during capture (sensor cooling, taking darks and using lots of bias frames etc) you image has a big gradent in it that i tried to hide by not streching it as far as it might be able to go. (once again don't want to spend time fixing it)

I'm no expert by any means at post processing but with that data you have provided i can see why you are struggling with it.

Attached Files



#6 Ivo Jager

Ivo Jager

    Vendor ( Star Tools )

  • -----
  • Vendors
  • Posts: 155
  • Joined: 19 Mar 2011
  • Loc: Melbourne, Australia

Posted 06 August 2013 - 09:34 PM

Hi,

This is what I came up with in StarTools. It does appear the data has been modified (it has been white balanced at the very least, making true color recovery hard and inhibiting noise reduction), while it also seems to have been stretched in some way (inhibiting again noise reduction and making deconvolution impossible).

Don't despair - it sounds you're doing everything right!

The trouble is that there are some features in DSS that are there to 'help' people get quick results, but in the end they do more harm than good if you really want to get the most out of the data. It can actually be tricky to turn them off I have found - I'm gathering some info and data so I can write a small tutorial/checklist...

Attached Files



#7 mmalik

mmalik

    DSLR camera modifications

  • *****
  • Vendors
  • Posts: 5478
  • Joined: 13 Jan 2012
  • Loc: USA

Posted 06 August 2013 - 09:39 PM

Couple of pointers:

1. Try following instructions in this... doc if they help.

2. It looks like you have picked a very difficult target to image and to process. Try something more discrete and concise than Andromeda.

3. You may try some unprocessed/combined FITs here... to refine your processing skills.

Hope this helps!

#8 TimN

TimN

    Surveyor 1

  • -----
  • Posts: 1703
  • Joined: 20 Apr 2008
  • Loc: Ontario, Canada

Posted 06 August 2013 - 09:56 PM

I recommend that you try StarTools. Its easier - was for me - than some of the others and can do a very good job. Its free to try and if you like it its only $60

#9 sternenhimmel

sternenhimmel

    Vostok 1

  • -----
  • Posts: 102
  • Joined: 05 Jun 2013

Posted 06 August 2013 - 10:02 PM

These are 3 excellent responses that I would like to address individually:

Frito and Ivo (and Falcon): You all appear to be correct. I realized earlier that when I opened the Autosave file that the curve was linear, whereas when I opened the TIFF file I saved, the curve was non-linear. I was unaware that the DSS defaults were non-linear and this does explain a lot.

Frito: There is a large gradient as a result of the wide field and that the object started near the horizon during the beginning of imagining. The good news is that my trial version of GradientXTerminate seems to do an excellent job at removing said gradient.

Ivo: Wonderful. I'm glad to see that, even working with non-linear data, you were able to extract significant detail and obtain good color. I'm unfamiliar with StarTools though...

mmalik: That's quite the guide you've got there! Are ImagePlus and PixInsight preferred tools for the people doing serious astrophotography? What makes Adromeda a difficult target? I always thought it would be an easy target due to it's large size and relative brightness.

#10 frito

frito

    Apollo

  • -----
  • Posts: 1183
  • Joined: 05 Oct 2012
  • Loc: Fremont, CA

Posted 06 August 2013 - 10:12 PM

Oh i'm quite familiar with gradents caused by light pollution, deal with it all the time on my own images. gradentXterminator is a great plugin for photoshop, i used up my trial and decided not to buy it because i'm no longer using photoshop for my astro stuff, pixinsight is just plain superior. Startools is a great easier to use cheaper alternative to both, i played with it, was set back at first at how basic it seemed but it can do some great things for sure but with pixinsights all inclusive and mass array of tools i'm definately going the pixinsight route once my trial licence is up, its well worth the 200 dollars esp if one does not own photoshop to begin with (and i do, but the plugins to make photoshop useful and eaiser on you quickly add up to 200 dollars)

the issue with andromeda is the color, its pretty subtle. here is a good comparison for you below is my most recent andromda i imaged last friday from a DARK site (bortle 2-3)

Posted Image

and attached was imaged from a orange zone "darker site" i often go to. granted i got much more data from the dark site and it was also more well placed when i imaged it from there it still shows the drastic differnece. its not as easy of a target as one would think, M13 would be something i would call an easy target but for really widefield like you are doing you should hit up sections of the milky way this time of the year i would think.

Attached Files



#11 mmalik

mmalik

    DSLR camera modifications

  • *****
  • Vendors
  • Posts: 5478
  • Joined: 13 Jan 2012
  • Loc: USA

Posted 06 August 2013 - 10:15 PM

"Are ImagePlus and PixInsight preferred tools for the people doing serious astrophotography?"

Yes! You could get their trial versions free to learn.

"What makes Andromeda a difficult target? I always thought it would be an easy target due to it's large size and relative brightness."

Quite the contrary; Andromeda's dynamic range and it's distribution makes it difficult to acquire/process properly. Try something like dumbbell nebula to practice. Thx

#12 sternenhimmel

sternenhimmel

    Vostok 1

  • -----
  • Posts: 102
  • Joined: 05 Jun 2013

Posted 06 August 2013 - 10:27 PM

I'm convinced, I'll take a look at PixInsight. PS was just familiar to me and I already had it, so I started there. I was also very encouraged by the impressive results I was seeing in various tutorials.

Anyway, I'm now working with a linear image and am already getting better results! I replace the TIFF file in the original thread with the unaltered version I'm using now, just in case anyone else wants to take a stab.
https://skydrive.liv...thkey=!AH_hR...

#13 frito

frito

    Apollo

  • -----
  • Posts: 1183
  • Joined: 05 Oct 2012
  • Loc: Fremont, CA

Posted 06 August 2013 - 10:30 PM

If you get a Pixinsight trial i highly recommend harry's astro shed's tutorials, they will get you started.

http://www.harrysast...nsighthome.html

#14 sternenhimmel

sternenhimmel

    Vostok 1

  • -----
  • Posts: 102
  • Joined: 05 Jun 2013

Posted 06 August 2013 - 10:33 PM

Awesome frito, thanks!

Also beautiful shots of Adromeda you've got there. Clearly what I need to do is invest in some better equipment ;).

#15 frito

frito

    Apollo

  • -----
  • Posts: 1183
  • Joined: 05 Oct 2012
  • Loc: Fremont, CA

Posted 06 August 2013 - 10:37 PM

i'm using a pretty basic setup to be honest, low amount of money invested thus far.

66mm Semi-APO scope
Canon 350D (Rebel XT)
used CG-5 AS-GT mount unguided.

thats it. total investment under 1k. starting off with just a DSLR and a lens is easier though but if you do step up to using a telescope start with a small widefield APO refractor like mine, it will help a lot, your camera alone on my setup would increase the quality of my images i have no doubt due to its higher resolution.

#16 sternenhimmel

sternenhimmel

    Vostok 1

  • -----
  • Posts: 102
  • Joined: 05 Jun 2013

Posted 06 August 2013 - 10:52 PM

I don't want to derail this read too much, but I will say this:

I own an Orion 8" f/4 Newtonian, but it lives at my parents' place in Arizona. That kind of scope would probably require guiding and a sturdy mount. Plus, it's not exactly the most portable scope.

I have been looking at the Stellarvue 70ED http://www.astropix....ESTS/SV70ED.HTM, which would be comparable to the scope you're using. Part of me wonders whether or not I could get away with 1 or 2 minute exposures at that focal length if I improve the performance of my barn door tracker...

#17 frito

frito

    Apollo

  • -----
  • Posts: 1183
  • Joined: 05 Oct 2012
  • Loc: Fremont, CA

Posted 06 August 2013 - 11:02 PM

your talking a pretty big weight and balance differience going from a DSLR w/ lens to a small refractor with lens, I could be wrong but i've never heard or seen anyone try to use a barn door tracker with an actual telescope before. I honestly thing it would be more trouble than its worth, just get a used CG-5 they can be had for under 400 these days, with an autoguider it would not be out of the question that you could image on a CG-5 with your 8" F/4 but an 8" SCT is a better route to go at that size.

#18 frito

frito

    Apollo

  • -----
  • Posts: 1183
  • Joined: 05 Oct 2012
  • Loc: Fremont, CA

Posted 06 August 2013 - 11:03 PM

oh and you can't derail your own thread, if anyone derailed it it was me bringing up equipment LOL.

#19 srosenfraz

srosenfraz

    Skylab

  • *****
  • Posts: 4199
  • Joined: 06 Mar 2011
  • Loc: United States

Posted 07 August 2013 - 12:40 AM

Here's a very quick and dirty processing of your data in Photoshop:

Posted Image

Full Resolution version. If you're interested to see my layers, my Photoshop .psd available here (25MB)

It still looks to me like your stacked tif has had a prior stretch applied to it. Nonetheless, your data seems quite usable. A more careful processing could probably produce a very nice image.

#20 mmalik

mmalik

    DSLR camera modifications

  • *****
  • Vendors
  • Posts: 5478
  • Joined: 13 Jan 2012
  • Loc: USA

Posted 07 August 2013 - 12:43 AM

PS was just familiar to me and I already had it, so I started there.


You can and will still be using PS, just little differently when you use PI or IP. A combinant processing technique here.... Thx

#21 Ivo Jager

Ivo Jager

    Vendor ( Star Tools )

  • -----
  • Vendors
  • Posts: 155
  • Joined: 19 Mar 2011
  • Loc: Melbourne, Australia

Posted 07 August 2013 - 12:48 AM

That's a lot better Sternenhimmel. The data is is still pre-whitebalanced, but at least it is no longer stretched.
I was able to achieve more correct colors now for the star field and the galaxy (non-whitebalanced data would make this even easier and would reduce noise further).

Attached Files



#22 sternenhimmel

sternenhimmel

    Vostok 1

  • -----
  • Posts: 102
  • Joined: 05 Jun 2013

Posted 07 August 2013 - 01:10 AM

I'm learning things that's for sure!

As both of you who took a stab at the new file pointed out, the data is probably still a little pre-processed and I think I know why. Photoshop never lets me open the Autosave files from DSS. It says there's not enough RAM, which I find hard to believe given I've allocated nearly all 16GB of RAM to photoshop. Thus, I opened up the autosave file in DSS and saved it as a 16-bit TIFF. The curve was linear, but maybe DSS applied something else of which I'm unaware (like white balance).

Now it seems like the thing to do is work with the Autosave file, completely unadulterated. When I start working with PixInsight this won't be a problem, but, why on earth can't I access my autosave file with Photoshop? I'll look into turning off all DSS settings that alter the stacked composite.

As for a small refractor on a barn door tracker: I suppose I underestimated the weight of such a refractor. 4.5lbs would be a heavy load that would require some design changes. But even then, I don't think I'd obtain any decent performance without moving to close-fitting metal parts, and at that point I'm approaching making my own equatorial mount -- probably not worth it.

I'm seriously considering a more substantial setup now.

#23 Falcon-

Falcon-

    Aurora

  • *****
  • Posts: 4932
  • Joined: 11 Sep 2009
  • Loc: Gambier Island, BC, Canada

Posted 07 August 2013 - 01:28 AM

The autosave file is a 32bit-per-channel image (96bit per pixel). Photoshop normally works in 16bit-per-channel or even 8bit-per-channel mode.

There are really not too many apps that CAN work on 32bit-per-channel data. Thankfully you can save out of DSS in 16bit-per-channel *without* the adjustments applied - as I recall there is an option in the file-save dialog itself (the one where you set the file name and location as you are saving it) as to if the saved image will have the adjustments applied or not. Saving with them not applied in 16bit-per-channel mode should make Photoshop happier and have the same non-modified data as the Autosave.tif image has.

#24 Falcon-

Falcon-

    Aurora

  • *****
  • Posts: 4932
  • Joined: 11 Sep 2009
  • Loc: Gambier Island, BC, Canada

Posted 07 August 2013 - 02:05 AM

Here is my quick run with your data in PixInsight. This is what I did:

- DynamicBackgroundExtraction
- ColorCalibration
- HistogramTransformation (medium stretch)
- TVGDenoise
- TVGDenoise again (only to Chrominance)
- CurvesTransformation (mostly Saturation with a bit of Luminance)
- HistogramTransformation (minor stretch)
- StarMask
- (applied new star mask to image)
- MorphologicalTransformation (Erosion tool, to reduce intensity of the stars vs the galaxy)
- (disabled star mask)
- DynamicBackgroundExtraction (again, to remove some residual bumpiness/gradient in background)
- CurvesTransformation (small saturation boost to mid-levels)
- HistogramTransformation (very small adjustment)

I would not consider that process a cookbook - I was just sorta randomly trying things and not spending too much time adjusting settings for optimal results or anything.

I would say you have good data here. As already said M31's very subtle colours and large brightness range make it actually a fairly hard object to process. The same camera/lens setup used here will do very nicely on other targets I think! (Sections of the Milky Way and M42 come to mind.)

Attached Files



#25 Falcon-

Falcon-

    Aurora

  • *****
  • Posts: 4932
  • Joined: 11 Sep 2009
  • Loc: Gambier Island, BC, Canada

Posted 07 August 2013 - 02:07 AM

The full frame:

Attached Files








Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics