Jump to content


Photo

I am doing something very very wrong

  • Please log in to reply
87 replies to this topic

#1 ICallHimGamblor

ICallHimGamblor

    Vostok 1

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

Posted 05 October 2013 - 10:01 AM

For some reason, I can't ever extract useful data from my stacked images. Last night, I took three hours of lights of Stephan's Quintet (60x180sec ISO400). I took 20 darks, I have 25 flats and 15 biases, and still can only raise the cluster just barely above the image floor. I have a two-hour stacked image of the fireworks galaxy (NGC 6946) that I can't raise above a wisp of a hint of an image.

I thought I was using StarTools improperly, but this is what Ivo said: "This is not really something that's a post-processing problem that ST can deal with. It seems like a stacking problem (incorrectly applied flats/darks/bias) or possibly an acquisition problem."

I am using a LP filter and am dithering using BackyardEOS + PHP.

Here is the stacked image: http://www.astrobin.com/58749/

And here is a random sub: http://www.astrobin.com/58912/

Can someone do an autopsy on this image to figure out what I am doing wrong? :confused:

Thanks.

#2 Chris_H

Chris_H

    Aurora

  • *****
  • Posts: 4752
  • Joined: 24 Nov 2003
  • Loc: Norway

Posted 05 October 2013 - 10:28 AM

I think you are expecting too much at this focal length. You are also slightly out of focus which doesn't do you any good at all at this image scale (fine detail will be lost). Can't help you with startools though as I don't use it. But if it has levels and/or curves (and it should if it's made for astro!), I would play with them.

#3 mmalik

mmalik

    Fly Me to the Moon

  • *****
  • Posts: 5322
  • Joined: 13 Jan 2012
  • Loc: USA

Posted 05 October 2013 - 10:34 AM

If your sky allows, do 5 or 10min subs; raise the ISO to 800 or 1600. As a control test, do an ICNR in case you problem might be calibration. Use this... LPS filter if using something else. Get evals of ImagesPlus and PixInsight to stack and process, for comparison. Thx

#4 srosenfraz

srosenfraz

    Soyuz

  • *****
  • Posts: 3942
  • Joined: 06 Mar 2011
  • Loc: United States

Posted 05 October 2013 - 12:39 PM

It looks like both your stacked image and your raw sub have had a stretch applied to them. It would be helpful to see the unstretched of each. Perhaps you could upload to someplace like Dropbox a raw sub (.CR2 file), as well as an unstretched 16 bit tiff?

You didn't mention what software you're using for stacking and calibrating. If you're using DSS, you'll want to save your final stack as a 16 bit tiff and be SURE that its not applying a screen stretch. You can tell if this is done correctly (unstretched), because the stacked tiff should be VERY dark (background 8 bit levels typically less than 5).

If you can post those two files, we'll probably have a better idea what you have happening.

Also, can you tell us more about ALL of the details about this image - what camera, is it modded, what scope (aperture and f/ratio), sky conditions (urban, rural, Bortle scale, Color zone, or SQM reading - some indication of how dark your sky is), what software you are using for stacking and calibration.

#5 ebacon

ebacon

    Explorer 1

  • -----
  • Posts: 74
  • Joined: 13 Feb 2013
  • Loc: Rochester Hills, MI

Posted 05 October 2013 - 02:12 PM

Will an unmodded camera detect Stephan's Quintet? At mag 14.8 it seems like it would be down in the noise.

I also shoot with an unmodded camera. I am in a red zone (according to Dark Sky Finder) and have found that many DSOs barely scratch through the noise floor.

#6 srosenfraz

srosenfraz

    Soyuz

  • *****
  • Posts: 3942
  • Joined: 06 Mar 2011
  • Loc: United States

Posted 05 October 2013 - 02:18 PM

Modding a camera increases the red sensitivity and has little (or no effect) at the blue end of the spectrum. Most galaxies tend to be either blue or white/yellow in general color, so they usually respond well to unmodded cameras. You typically won't capture the Ha regions very well, but the majority of the galaxy is unaffected.

The galaxies in Stephan's Quintet are mostly blue and yellow, so one would expect an unmodded camera to work almost as well as a modded camera for a target like this.

Ojbects that are primarily Ha (very red) benefit significantly from modding the camera.

#7 TimN

TimN

    Apollo

  • -----
  • Posts: 1414
  • Joined: 20 Apr 2008
  • Loc: Ontario, Canada

Posted 05 October 2013 - 04:30 PM

I don't think there is enough data there. I agree with Mikes earlier comment - bump up the iso to 800 and try at least 5 minute subs.

#8 ICallHimGamblor

ICallHimGamblor

    Vostok 1

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

Posted 05 October 2013 - 04:45 PM

It looks like both your stacked image and your raw sub have had a stretch applied to them.


I have not stretched anything... those are the pics directly from the camera and DSS converted to JPG.

I am not sure what that tell you. :confused:

#9 ICallHimGamblor

ICallHimGamblor

    Vostok 1

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

Posted 05 October 2013 - 04:49 PM

If your sky allows, do 5 or 10min subs; raise the ISO to 800 or 1600.


I thought I read somewhere on here that subs should have a histogram that lies at 1/3 of the histogram graph. If I can go further than that, I would love to, but the histogram on the sub I posted looks right on that point. Let me know if I can go further...

#10 ICallHimGamblor

ICallHimGamblor

    Vostok 1

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

Posted 05 October 2013 - 04:50 PM

I think you are expecting too much at this focal length.


Oh yeah... I am also using the standard celestron focal length reducer, so I am shooting at 6.something...

#11 Jamienite

Jamienite

    Lift Off

  • -----
  • Posts: 22
  • Joined: 30 Nov 2011
  • Loc: Western Massachusetts

Posted 05 October 2013 - 05:02 PM

I agree with the above comments, 180 seconds is not enough for such a dim object. I've tried the quintet on a night where I wasn't tracking well and could only manage 2-3 minutes. I got results very similar to yours.

#12 Jamienite

Jamienite

    Lift Off

  • -----
  • Posts: 22
  • Joined: 30 Nov 2011
  • Loc: Western Massachusetts

Posted 05 October 2013 - 05:09 PM

Oh, and bump up to iso800 for sure. ;)

#13 shawnhar

shawnhar

    Fly Me to the Moon

  • *****
  • Posts: 5294
  • Joined: 25 Jun 2010
  • Loc: Knoxville, TN

Posted 05 October 2013 - 05:29 PM

Restack the subs
In DSS, save picture to file, in the save options, make sure "Embed but DO Not apply" option is checked.
Upload that tiff to Dropbox or something we can download from
You might be surprised...

#14 srosenfraz

srosenfraz

    Soyuz

  • *****
  • Posts: 3942
  • Joined: 06 Mar 2011
  • Loc: United States

Posted 05 October 2013 - 06:01 PM

It looks like both your stacked image and your raw sub have had a stretch applied to them.


I have not stretched anything... those are the pics directly from the camera and DSS converted to JPG.

I am not sure what that tell you. :confused:


I understand that you haven't consciously stretched anything - you probably have software doing it without you realizing it. From the camera, if you pull it into most software as a RAW, it'll apply a default stretch to it. When you save it as jpeg, it'll include that stretch. If you take it from the camera as a Jpeg, the camera will have applied the default stretch.

With DSS, the default behavior is to apply the screen stretch unless you tell it to save it as a 16 bit tiff and do NOT save the screen stretch with it. That's what it looks like you have there. Again, an unstretched image is usually VERY dark.

It would be quite valuable to see unstretched data, as you're asking about what you may be doing wrong to end up with less than optimal results. With unstretched data, we can look at it, process it, etc. to have a much better idea what the potential is.

With 3 hours of Stephan's Quintet, I would expect that you should normally have quite usable data. But, 3 hours is usually plenty enough time to have a good enough SNR to allow for a quality image - particularly at f/6.3. My image of Stephen's Quintet at f/6.3 on a C8 should be pretty comparable to yours. Mine was at ISO 1600, yours at ISO 400 - No big deal. If I had shot this at ISO 400, it would have looked more or less the same. I used 3 minute subs as well, and have a little over twice as much time. That means that my SNR is about 1.5x better than yours. That's better, but its not night and day either.

I suspect there's a decent image hidden in your data....

#15 ICallHimGamblor

ICallHimGamblor

    Vostok 1

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

Posted 05 October 2013 - 06:27 PM

Wow... I had no idea stretching was being applied to my image.

Here is the 'no adjustments' version.

http://www.astrobin.com/58959/

#16 srosenfraz

srosenfraz

    Soyuz

  • *****
  • Posts: 3942
  • Joined: 06 Mar 2011
  • Loc: United States

Posted 05 October 2013 - 06:37 PM

That looks like the right stuff. The only thing is that we need that as a 16 bit tiff. I don't think you'll be able to post that large a file to Astrobin - that's why the suggestion to use something like Dropbox (free for up to 2 GB of total storage).

#17 Jamienite

Jamienite

    Lift Off

  • -----
  • Posts: 22
  • Joined: 30 Nov 2011
  • Loc: Western Massachusetts

Posted 05 October 2013 - 06:49 PM

Wow! I had now idea about this dss auto stretch. Thanks Scott. Also, looking at your Quintet image, I see that 3 minutes is plenty of exposure! Great image!

#18 ICallHimGamblor

ICallHimGamblor

    Vostok 1

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

Posted 05 October 2013 - 06:56 PM

Here is the TIF: https://docs.google....dit?usp=sharing

#19 srosenfraz

srosenfraz

    Soyuz

  • *****
  • Posts: 3942
  • Joined: 06 Mar 2011
  • Loc: United States

Posted 05 October 2013 - 07:04 PM

Wow! I had now idea about this dss auto stretch. Thanks Scott. Also, looking at your Quintet image, I see that 3 minutes is plenty of exposure! Great image!


Thank you very much. And you're welcome. Also, thanks to Shawn for the precise directions on how to save it correctly out of DSS.

One thing to note about the unstretched data - if you data is underexposed (i.e. ISO 400 with 3 minute subs from a dark sky), it can often be difficult to stretch this data using levels and curves (it can be done, its just very tricky). For these types of images, stretching with DDP (Digital Development Processing) tools makes all the difference in the world (as well as for well exposed images, too). For well exposed images (back of the camera histogram mountain well histogram well separated from left axis), stretching with levels and curves is the same as usual.

#20 mmalik

mmalik

    Fly Me to the Moon

  • *****
  • Posts: 5322
  • Joined: 13 Jan 2012
  • Loc: USA

Posted 05 October 2013 - 07:30 PM

There seem to be calibration/gradient issues with your image. My go at your TIF below:

ImagesPlus:
1. RAW File Conversion (Skipped)
2. Align TSR/Combine/Crop (Only Crop)
3. Digital Development
4. Multi Point Flatten Background-Planar
5. Multiresolution Smooth/Sharpen
Saved as 16 bit TIFF for processing in PixInsight

PixInsight:
6. HDR Multiscale Transform
7. ACDNR (Noise Reduction)
Saved as 16 bit TIFF for processing in ImagesPlus/Photoshop

ImagesPlus/Photoshop:
8. Geometric Transform/Scale [0.3] (ImagesPlus [Skipped])
9. Smoothing & Noise Reduction (ImagesPlus)
10. ProDigital ‘Deep Space NR’ and HLVG (Photoshop)
11. Saturation (Block: 1 pass) and Noise Reduction (Photoshop)
12. Star Size & Halo Reduction, and Saturation (ImagesPlus)
13. Levels, Curves, Contrast, Brightness, Vibrance, and Color Balance (Photoshop)
Saved as JPG for posting


Processing instructions here...

Attached Files



#21 srosenfraz

srosenfraz

    Soyuz

  • *****
  • Posts: 3942
  • Joined: 06 Mar 2011
  • Loc: United States

Posted 05 October 2013 - 11:52 PM

I agree with Mike about there being something wrong with this data. I have the same color gradient issues when I try to do a DDP stretch, and the image has extremely low contrast.

You've fundamentally done all the right things - captured a good amount of data, have reasonable number of calibration frames, etc. It feels like something may have gone haywire along the way.

A few questions -

1) With your flats - is the orientation of the camera unchanged from the time you take your flats until the time you start capturing?

2) When you captured your flats, did you have the CLS filter in place (and left it the same until you were capturing data)?

3) When applying your calibration frames in DSS - are you sure you have all of the correct frames in the right place (i.e., you didn't swap dark frames for flats, or some other type of mistake)?

4) Would you mind posting one raw .CR2, your Master Bias, Master Flat, and Master Dark frames?

#22 ebacon

ebacon

    Explorer 1

  • -----
  • Posts: 74
  • Joined: 13 Feb 2013
  • Loc: Rochester Hills, MI

Posted 06 October 2013 - 06:53 AM

I am learning from this thread, too.

What are calibration frames? Is that a generic term for the groups of flats, bias, and darks?

Thanks!

#23 Footbag

Footbag

    Fly Me to the Moon

  • *****
  • Posts: 5750
  • Joined: 13 Apr 2009
  • Loc: Scranton, PA

Posted 06 October 2013 - 08:29 AM

I am learning from this thread, too.

What are calibration frames? Is that a generic term for the groups of flats, bias, and darks?

Thanks!


Yes. Flats, darks and bias are calibration frames.

#24 shawnhar

shawnhar

    Fly Me to the Moon

  • *****
  • Posts: 5294
  • Joined: 25 Jun 2010
  • Loc: Knoxville, TN

Posted 06 October 2013 - 11:23 AM

I'm gonna guess one of 2 things....
1st, this image was taken through clouds, maybe high thin clouds you couldn't see, but that is the most likely case. I have had the same thing happen, flat's can't fix it because the gradients are changing frame to frame.
2nd and hopefully less likely, you have crazy multiple light sources around like neighbor lights or you are shooting through/close to light sources. Flats can't fix that either because the target is moving, therefore the gradients are moving frame to frame.
I gave a go in Star Tools but like Ivo said, very difficult to doa anything with it in post processing, those gradients are pretty rough. I tried using masks for the galaxies and "isolate" in the Life function.

Attached Files



#25 petemumbower

petemumbower

    Surveyor 1

  • -----
  • Posts: 1639
  • Joined: 12 Jul 2009
  • Loc: West Michigan

Posted 06 October 2013 - 12:56 PM

How does the each frame look after calibration before stacking? I want to agree that maybe high thin clouds are present.






Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics