Jump to content


Photo

I am doing something very very wrong

  • Please log in to reply
87 replies to this topic

#51 bunyon

bunyon

    Gemini

  • *****
  • Posts: 3224
  • Joined: 23 Oct 2010
  • Loc: Winston-Salem, NC

Posted 07 October 2013 - 08:18 PM

I'm no expert, but I'd try just stacking the lights and see how it looks. That is, no calibration. That would tell you if the calibration frames are a problem.

#52 akulapanam

akulapanam

    Mariner 2

  • -----
  • Posts: 223
  • Joined: 26 Aug 2012

Posted 07 October 2013 - 09:53 PM

The Autosave.tif file is itself problematic. It contains 16-bits of real data + 16-bits of zeroes within a 32-bit tif wrapper (quite why it does this is unclear).


Some editing/AP packages will not open the Autosave.tif file at all.


Why this is so is easy - its a private TIFF format that DSS uses to cache intermediate data during stacking and for reloading once the stacking is done - its not designed for users to play with. Hence many programs can't deal with it.

Because of a bug in DSS, the "File - Save As" command does not create a high-quality file. It creates a 16-bit tif wrapper with 8-bit data inside it.


I find this troubling as this is not my experience at all. At least the evidence when opening a 16 bit in Images plus is contrary to this - e.g. the histogram data statistics do not concur and there are no gaps (combing) which is what you would expect with an 8 bit depth file maskerading as a 16 bit file.

If this is really the case then posterization of brightness data after even a basic stretch would be very evident (like it is in GIMP which is still stuck at 8 bits). I don't see this after any sort of stretching in Images Plus ... so it is now greatly puzzling me why we are just hearing of this.

Which makes me think that if there is a bug it may be down to specific types of RAW file input (i.e. from a particular camera model) The only thing I can think might be an issue is if some has been using a newer model camera and hasn't picked up the DSS beta with the newest DCRaw program - i.e. the problem actually lies with using an too old version of the DCraw program that DSS uses.

It would be interesting if more information was available - camera used etc


I've just done an intensive Google on this issue and I can find nothing else reported or said by anyone. The nearest is Sander Pool (DSS Yahoo Groups guru) mentioning that post processing in GIMP is limited by it 8 bit depth. I would have though if anyone was going to make a song+dance of this bug Sander would!

:shrug:


There are several threads on astronomyforum that detail the issue with photos. I'm sure that their is at least one in DSS on yahoo because I believe I posted in the thread. When I get a sec I'll post links for you but you will have to register.

#53 Tonk

Tonk

    Fly Me to the Moon

  • *****
  • Posts: 7078
  • Joined: 19 Aug 2004
  • Loc: Leeds, UK, 54N

Posted 08 October 2013 - 03:05 AM

Eventually this bug should get fixed. Until it does, for the best possible data out of DSS you need to use the Autosave.tif file that it creates automatically after stacking is completed.


So has Luc acknowledged this bug and given a indication of when it will be fixed? Do you have a link to the bug report?

#54 shawnhar

shawnhar

    Fly Me to the Moon

  • *****
  • Posts: 5358
  • Joined: 25 Jun 2010
  • Loc: Knoxville, TN

Posted 08 October 2013 - 12:45 PM

I'm with Tonk, I think this is mis-informed info, someone is confused about 8bit vs 16bit. I tried the "fix" of converting the autosave from DSS to a 16bit tiff and saw no difference whatsoever at 100% zoom between that and the regular output tiff.
8 bit sucks, even when in a 16 bit wrapper, you will know 8 bit instantly when you stretch because it looks like Nintendo, the 1st one.

#55 Tonk

Tonk

    Fly Me to the Moon

  • *****
  • Posts: 7078
  • Joined: 19 Aug 2004
  • Loc: Leeds, UK, 54N

Posted 09 October 2013 - 06:42 AM

I think this is mis-informed info, someone is confused about 8bit vs 16bit.


+1. The so called bug is bogus. Just done my own checks via the data "fix" - like you no difference found.

#56 ICallHimGamblor

ICallHimGamblor

    Vostok 1

  • -----
  • Posts: 127
  • Joined: 24 Jun 2013

Posted 09 October 2013 - 07:36 AM

So did my new biases fix my calibration issue? It looks better when I auto-dev in StarTools, but I still can't get much data out of it.

#57 shawnhar

shawnhar

    Fly Me to the Moon

  • *****
  • Posts: 5358
  • Joined: 25 Jun 2010
  • Loc: Knoxville, TN

Posted 09 October 2013 - 11:29 AM

I tried it and it has the same gradients...

#58 ICallHimGamblor

ICallHimGamblor

    Vostok 1

  • -----
  • Posts: 127
  • Joined: 24 Jun 2013

Posted 09 October 2013 - 12:04 PM

In that case, I am kinda stuck. I don't know what to try next. Any suggestions would be greatly appreciated.

Can I go longer on subs? Should I bump ISO to 800? How important is the pointer that I read that the histogram of the subs should be about 1/3 of the spectrum?

#59 pfile

pfile

    Gemini

  • -----
  • Posts: 3163
  • Joined: 14 Jun 2009

Posted 09 October 2013 - 12:59 PM

the gradients in this image are completely wild... are you sure that the subs were not taken thru clouds or that some of the subs have trees or a house or something in them?

#60 shawnhar

shawnhar

    Fly Me to the Moon

  • *****
  • Posts: 5358
  • Joined: 25 Jun 2010
  • Loc: Knoxville, TN

Posted 09 October 2013 - 02:13 PM

1st, stack without the flats, I would start there, have taken more than my share of bad flats, they are easy to get wrong.
2nd, I have taken pics where the histo was over half and ones where it was on the left wall. The ones one the left wall will be noisy as all get out, the ones over half will have a horrible washed out look that can be difficult to deal with, but it's better than being jammed up against the wall or limited to 2 minute subs. I have also dropped to iso 400 to get 10 minute subs and they came out fine.
Try iso 400 and go as long as you can without going over halfway on the histo and you will be fine.
Those gradients on the other hand...you gotta figure out what is causing that. Start with the flats, if that ain't it, maybe what is in your area that is casting light on your lens? My money is now on the flats.
Other than the gradient I think you are getting good data.

Attached Files



#61 srosenfraz

srosenfraz

    Soyuz

  • *****
  • Posts: 3994
  • Joined: 06 Mar 2011
  • Loc: United States

Posted 09 October 2013 - 05:19 PM

The new bias frame you posted still doesn't look right. Its level is at 56, where it should be at 0. Now, its possible that the bias is ok, and what you're posting has had some stretch applied, but there's something not right about the Master offset file you posted. That may or may not be the issue, but I think you'll want to resolve the discrepancy.

@Shawn - since you're familiar with DSS, would you mind giving ICallHimGamblor directions on how to save an unstretched Master offset tiff file?

@ICallHimGamblor - Would it be possible to post a raw sub (the .cr2 file itself)? That will give us a much better idea as to what your raw data looks like before anything is done to it.

#62 Rankinstudio

Rankinstudio

    Mercury-Atlas

  • *****
  • Posts: 2586
  • Joined: 05 Apr 2008
  • Loc: Southern UT

Posted 09 October 2013 - 06:43 PM

I've been having a lot of issues with DSS as well. It looks to be reversing my flats at times, hollow stars, ect. I have found a few settings that seem to help get consistent results.

Under registration
Check automatic detection of hot pixels
Under advanced, reduce noise using median filter
Then adjust the slider and do multiple test runs until you end up with a number of stars that looks like what is in your image.

In the stacking tab use "Standard" under Result
Do a Median combine for lights, darks, flats, bias
Under alignment choose Bilinear
Under cosmetic, uncheck everything

For the best stretch, I save it out as 32bit rational then do a linear stretch before converting it to 16bit.

Seems to make it work right for me for some reason.

Cheers,

#63 ICallHimGamblor

ICallHimGamblor

    Vostok 1

  • -----
  • Posts: 127
  • Joined: 24 Jun 2013

Posted 09 October 2013 - 09:02 PM

Thanks, everyone... everyone is being so helpful.

Scott: I uploaded a single random sub here: https://docs.google....dit?usp=sharing

David: Thanks.. I will try restacking and re-saving tomorrow when I have a little more time.

Right now, I'm going to try stacking in my Nebulosity trial to see if it makes a difference.

#64 shawnhar

shawnhar

    Fly Me to the Moon

  • *****
  • Posts: 5358
  • Joined: 25 Jun 2010
  • Loc: Knoxville, TN

Posted 10 October 2013 - 07:43 AM

@Shawn - since you're familiar with DSS, would you mind giving ICallHimGamblor directions on how to save an unstretched Master offset tiff file?

In the folder where your flats are, DSS will generate a master flat when you stack them, same with the darks and bias, but that ain't the problem.
Your single sub has those crazy gradients as well.
What is the light like around you? Streetlight behind some leaves shining in the scope? Gotta figure this out 1st...

Attached Files



#65 ICallHimGamblor

ICallHimGamblor

    Vostok 1

  • -----
  • Posts: 127
  • Joined: 24 Jun 2013

Posted 10 October 2013 - 07:57 PM

I am in the city, and there are definitely streetlights around, though none with any direct light. There is one nearby behind some leaves.

However, due to trees I am shooting mostly straight up. I bet I started my imaging of this set at 85 degrees above the horizon.

Am I basically just screwed completely because I live in the city? I am pretty sure there are folks on this forum getting decent results in metropolitan areas.

#66 pfile

pfile

    Gemini

  • -----
  • Posts: 3163
  • Joined: 14 Jun 2009

Posted 10 October 2013 - 08:46 PM

you may need a light pollution filter, in which case you should be able to get images of emission nebulae at the very least. i have taken these images with a modified DSLR + lenses and telescopes as well, in what is a red zone according to the maps:

Posted Image

Posted Image


Posted Image

i used the Astronomik CLS filter. it is pretty unkind to galaxies, because it "deletes" a lot of wavelengths that galaxies emit. so with CLS it's easier to stick with Ha/OIII rich targets.

it can be done, don't give up :)

anyway, those gradients look like clouds or even dew on your scope/lens; i don't think i've ever seen such a complex gradient in any of my subs. usually the gradient is mostly linear in nature.

rob

#67 ICallHimGamblor

ICallHimGamblor

    Vostok 1

  • -----
  • Posts: 127
  • Joined: 24 Jun 2013

Posted 10 October 2013 - 09:01 PM

I have uploaded a random sub from a week earlier on a completely different target, Hickson 93.

https://docs.google....dit?usp=sharing

Does it have the same issues? If so, we can eliminate clouds and dew and stuff, right?

#68 srosenfraz

srosenfraz

    Soyuz

  • *****
  • Posts: 3994
  • Joined: 06 Mar 2011
  • Loc: United States

Posted 10 October 2013 - 09:22 PM

Suggestion - you're trying to image some very challenging targets from very light polluted skies. While its possible to do, it might serve you better to start with some brighter targets and refine your routines and techniques for image capture and processing. Targets like Stephan's Quintet and Hickson 93 can make for some beautiful and impressive images. But, they also demand tight focus to capture the small details plus a lot of exposure to give a good SNR due to their relatively low surface brightness. When you're imaging from light polluted skies, you'll typically need to increase that total exposure time to many, many times what it takes at a dark sky (i.e., a target I can capture well in 1 hour from my blue/green zone sky will take 8 hours for comparable SNR in a red zone). As such, if you try working more with objects with a higher surface brightness, you can capture them in a more reasonable amount of time and start seeing some positive results.

You're not completely screwed because you're in the city, but it does make things more difficult. Still, there's a long list of bright targets (high surface brightness) that should be comparatively easy for you to capture (Mr. Messier developed a pretty good list for starters).

Hope this helps...

#69 ICallHimGamblor

ICallHimGamblor

    Vostok 1

  • -----
  • Posts: 127
  • Joined: 24 Jun 2013

Posted 10 October 2013 - 09:43 PM

Scott: I completely understand what you are saying, and I agree. I just have very little interest in those targets. I don't mind at all spending 8 hours on a neat Hickson group, as long as I know something useful will come out of it.

I do have some subs of the Deer Lick Group and NGC 6946 that are equally as disappointing in my opinion, and those are pretty bright, large galaxies. In fact, NGC 6946 came out worst of them all.

If the answer to all my problems is just "you need more exposure time", then that's fine. I'm good with that. But it sounds like I'm doing something wrong if my bias frames don't come out black and I'm getting psychotic gradients on three minute subs.

But while I'm learning, I will also seek out some easier targets. Your advice is definitely taken.

#70 pfile

pfile

    Gemini

  • -----
  • Posts: 3163
  • Joined: 14 Jun 2009

Posted 10 October 2013 - 11:26 PM

so just as a sanity check, did you stack without any calibration frames and check the results?

it is hard to see the gradients in a single sub. what i can see though is that the vignetting in each channel is very different, which is a little strange for a reflecting telescope. still there are refracting elements in an SCT which might account for that.

also your camera temperature is way up there at 97F (36C). this can really make it difficult to collect dim targets. you may want to investigate building a cooler box to keep the thermal signal down.

rob

#71 shawnhar

shawnhar

    Fly Me to the Moon

  • *****
  • Posts: 5358
  • Joined: 25 Jun 2010
  • Loc: Knoxville, TN

Posted 11 October 2013 - 06:29 AM

I have uploaded a random sub from a week earlier on a completely different target, Hickson 93.

https://docs.google....dit?usp=sharing

Does it have the same issues? If so, we can eliminate clouds and dew and stuff, right?

Seems to have the same gradient.
Take a raw with the lens, pointed at a surface that is uniformly colored, like a wall and see if it still has it.

Attached Files



#72 Tonk

Tonk

    Fly Me to the Moon

  • *****
  • Posts: 7078
  • Joined: 19 Aug 2004
  • Loc: Leeds, UK, 54N

Posted 11 October 2013 - 07:39 AM

That look like plain old vignetting as far as the brightness drop off to the corners is concerned (that's sorted by flats) - the dark band along the bottom is often due to the spongy strip padding that runs on one edge to damp out the mirror slap - that too is cured by flats

- or are you referring to the weak coloured gradients on top. No idea what doing that and flats would not likely solve those

#73 ICallHimGamblor

ICallHimGamblor

    Vostok 1

  • -----
  • Posts: 127
  • Joined: 24 Jun 2013

Posted 11 October 2013 - 09:27 AM

Oh man.. I really hope we are back to just having a problem with my flats.

#74 shawnhar

shawnhar

    Fly Me to the Moon

  • *****
  • Posts: 5358
  • Joined: 25 Jun 2010
  • Loc: Knoxville, TN

Posted 11 October 2013 - 10:21 AM

It's the color gradient that is causing the issue. I am not really sure what is causing it or how to deal with it, never seen gradient like that on one sub except shooting through clouds. The gradient appears to be the same on each non-calibrated exposure, suggesting it is internal to the camera or the scope.

#75 ICallHimGamblor

ICallHimGamblor

    Vostok 1

  • -----
  • Posts: 127
  • Joined: 24 Jun 2013

Posted 11 October 2013 - 11:09 AM

Is there a camera setting that could affect this? That might also explain why my biases are not black.






Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics