DBE trouble

[rdryfoos]

Still having trouble with DBE.  Attached are 2 images --1 is with no DBE, the second is the same image after DBE.  This always happens to my LRGB stacks/images.  Narrowband I have down pat.  But this is frustrating.  What is going on?  Vicent, if you could clarify why DBE makes the image look like this I would be most appreciative.  thanks--REodd

This will take 2 posts do to size restrictions

1--no DBE

[rdryfoos]

Here's the DBE image.  How can I process when it looks like this?  This is the same issue I had with NGC 1333--

[Geoff]

What parameters (sample size, number of samples, ....) are you using for dbe?
Maybe upload a fits image (before dbe)so others can try processing it.

[rdryfoos]

It does not matter how I change the settings--I have tried many, many various combinations of box sizes, smoothing factor, shadows relaxation, normalized, not normalized, etc.  This always happens to my RGB and Lum data--always, but it never happens to my narrow band data.  Could it be my skies?  I find it hard to believe that no one in the state of CT can shoot broad band data.  Here are the fits files for RGB and Lum prior to DBE.  Images have been cropped to be the same size.

I don't know how to upload the drop boxfiles.  They are on my drop box but How do I upload the link to this forum?

What parameters (sample size, number of samples, ....) are you using for dbe?
Maybe upload a fits image (before dbe)so others can try processing it.

Thanks--I hope this problem can be fixed.  Until it is I can only shoot NB
Rodd

[mmirot]

Up the Tolerance to about 2.5 to 2.8

I up the sample radius to 18.
Did a standard generation , added samples missing areas. Avoid samples with bright stars, star halos and object.

Set  Target to replace and subtraction.

Hit the check box.

Max

[rdryfoos]

Up the Tolerance to about 2.5 to 2.8

I up the sample radius to 18.
Did a standard generation , added samples missing areas. Avoid samples with bright stars, star halos and object.

Set  Target to replace and subtraction.

Hit the check box.

Max

Well--you certainly got it to work.  Thanks.  I will try your suggestions. 

[rdryfoos]

OK--new problem.  After I use DBE the use of the STYF causes the problem.  I can manually stretch to go non linear, but I don't want to--I want to use the stf to look at it to see if I am ready.  If color, I want to do BN and CC before using the histogram.  Why does the STF cause the issue that is evident in the second image?

[Duncan]

I've seen something like this before. What's happening is that because so much of the image is background, the Auto-STF is stretching the image much more after DBE. Essentially it's cranking up the stretch until it can see some variation in the background.

The fix is to manually move the sliders around in STF after DBE (and doing auto-STF settings first). I usually find that I have to move the mid-point slider a long way to the right in this situation.

Duncan

[rdryfoos]

I've seen something like this before. What's happening is that because so much of the image is background, the Auto-STF is stretching the image much more after DBE. Essentially it's cranking up the stretch until it can see some variation in the background.

The fix is to manually move the sliders around in STF after DBE (and doing auto-STF settings first). I usually find that I have to move the mid-point slider a long way to the right in this situation.

Duncan

Thanks Duncan--it does seem that manual stretching solves (helps with) the problem.  But that makes it more problematic to identify dark background pixels to use for the BN preview.  Also-it only happens with LRGB--not narrowband, regardless of how dim the target is

[Geoff]

Have you tried ABE to see if there is the same problem?

[Duncan]

Thanks Duncan--it does seem that manual stretching solves (helps with) the problem.  But that makes it more problematic to identify dark background pixels to use for the BN preview.  Also-it only happens with LRGB--not narrowband, regardless of how dim the target is

I've typically found that it really depends on how much of the image is background; if I have some reasonable sized nebula in the image the auto settings for STF are fine, irrespective of whether it's LRGB (or one channel) or narrowband.

But images that have sky background as a large proportion of the pixels (like a cluster of smaller galaxies) I usually find STF doesn't do a good job at auto-setting, especially after a successful DBE/ABE that leaves the background essentially flat and dark.

No big deal - in those cases it just requires manual adjustment of the STF sliders. I have found it helps to duplicate the original pre-DBE image so as to have a comparison reference. Adjust the mid-point STF slider looking at the features of the galaxies until they look similar, and then adjust the black point slider.

I'm not sure I understand the issue with dark background pixels - could you explain a little more? I have always been able to get a manually adjusted set of STF settings that shows a flat (but not completely black) background.

Duncan

[rdryfoos]

Thanks Duncan--it does seem that manual stretching solves (helps with) the problem.  But that makes it more problematic to identify dark background pixels to use for the BN preview.  Also-it only happens with LRGB--not narrowband, regardless of how dim the target is

I've typically found that it really depends on how much of the image is background; if I have some reasonable sized nebula in the image the auto settings for STF are fine, irrespective of whether it's LRGB (or one channel) or narrowband.

But images that have sky background as a large proportion of the pixels (like a cluster of smaller galaxies) I usually find STF doesn't do a good job at auto-setting, especially after a successful DBE/ABE that leaves the background essentially flat and dark.

No big deal - in those cases it just requires manual adjustment of the STF sliders. I have found it helps to duplicate the original pre-DBE image so as to have a comparison reference. Adjust the mid-point STF slider looking at the features of the galaxies until they look similar, and then adjust the black point slider.

I'm not sure I understand the issue with dark background pixels - could you explain a little more? I have always been able to get a manually adjusted set of STF settings that shows a flat (but not completely black) background.

Duncan

  Well--when you choose your preview for background neutralization, I was under the impression that you want to use pixels that are chromatically even (I use the readout numbers at the bottom of the screen).  The BN process subtracts the value of the preview pixels from the background so you don't want a preview that is skewed toward any chromal value.  Also--I was under the impression that dark pixels are better than bright ones--as you want a dark background.  Please correct my assumptions if they are incorrect.

[Duncan]

Well--when you choose your preview for background neutralization, I was under the impression that you want to use pixels that are chromatically even (I use the readout numbers at the bottom of the screen).  The BN process subtracts the value of the preview pixels from the background so you don't want a preview that is skewed toward any chromal value.  Also--I was under the impression that dark pixels are better than bright ones--as you want a dark background.  Please correct my assumptions if they are incorrect.

According to my understanding, the background neutralization process is unaffected by STF; STF is just showing a visualization whereas the neutralization process is only concerned with the actual linear pixel values.

What I think is going on with the DBE result is that, in the case of an image with a small proportion of pixels that are DSOs rather than sky background, the algorithm chooses slider values that excessively stretch the image and make it appear that the background is very uneven when it really isn't. Because there's "nothing there" in most of the pixels in the image, and the STF auto-stretch is trying to pick values that show something, you end up with slider settings that excessively stretch the image. Moving them around manually solves the problem. None of that should affect subsequent linear process steps, though.

Duncan