DBE trouble

[rdryfoos]

Lately I am having peculiar issues with DBE and ABE when I use it.  Sometimes the use of DBE completely ruins my stack, or my image when applied after integration of the stacks.  Using DEB turns the background mottled, and there are lines and patterns visible.  Zoomed out it looks grainy, zoomed in it looks scary.  It is not the calibration frames because I will shoot 2 images on the same night and 1 comes out decent (DBE does not ruin it), and ones is ruined.  It is not Mure Denoise, which I always use, because I processed the same image with and without Mure Denoise and the use of DBE ruined both.  Prior to DBE, the image looks very smooth and except for the gradients, would be good.  I have attached a red stack for NGC 1333 (21 10 minute subs, STT 8300, Televue np101is).  Image 1 is the red stack prior to DBE, Image 2 is the red stack after DBE.  Why the mottling?

I really need help with this because it is preventing me from processing.  It is extremely frustrating.

[rdryfoos]

PS--trhe extreme JPEG compression has impacted the first image--it looks much smoother in its natural FITs or XISF format.

[rdryfoos]

That's why there needs to be a manual or book written by the PI team--so when posts of inquiry go ignored it is not so irritating.  Yes I have read Warrens book and watched his videos....don't cut it....too general, no room for questions.  That's why there is a forum people.

[ChoJin]

maybe you could learn being patient too. You often sound very aggressive...

[rdryfoos]

maybe you could learn being patient too. You often sound very aggressive...

Then weeks would go by instead of just days. 

[Juan Conejero]

Both images are normal results. The "mottling" is just noise that becomes more visible after a stronger stretch, which is feasible after removing large-scale gradients. The same noise exists in the first image; it is just less evident. DBE is not the main problem here, but more basic concepts such as signal, noise, brightness and contrast.

[rdryfoos]

Both images are normal results. The "mottling" is just noise that becomes more visible after a stronger stretch, which is feasible after removing large-scale gradients. The same noise exists in the first image; it is just less evident. DBE is not the main problem here, but more basic concepts such as signal, noise, brightness and contrast.

  Well--some very experienced imagers and I have been looking at the Use of DBE on some of my ages that are pretty decent images.  There is no question that DBE and ABE are producing strange results for me--images that have a high SNR.  After all, many RGB targets are faint and they come out pretty darn good for many folks--NGC 1333 among them (think dust in the iris, the ghost etc). However, the image I posted was a bad choice, first of all it was the wrong stack--but I will not go through the trouble of pursuing this here, which seems very strange to me as it is a PI issue, and this is a PI forum.  But the discussion would be long (I know, I have had it before).  As far as a more strong stretch--not so.  I use DBE and the STF set at the same level as before the DBE--I know I am not the guru that you are....but this is not a normal result.  I have gotten decent (normal results).so I do know what they look like.

[MikeOates]

Ticking the Normalize option in DBE gives a smoother result, perhaps you normally use that and for some reason not this time?

Mike

[rdryfoos]

No--I don't usually check that box.....somewhere along the line I believe it was recommended not to check that box.  Maybe that will solve it--i will try this data to see.  The following are true for me

1) Still having memory read errors 50% of teh time and can't use DBE without a system restart
2) When DBE works Narrowband never has an issue with DBE
3) LRGB has the issues--typically in more light polluted skies

Could it be a light pollution issue?  I even had this problem when I heavily denoised the stack (Mure Denoise, followed by removing the 1st wavelet layer using MSMT (also tried removing first 2 layers), followed by a heavy dose of TGV denoise. 

Sometimes using DBE (or ABE) leaves the background mottled and patterned--the black pixels are way too black, and there are lines visible between pixels.

[jkmorse]

If your data is as good as you say, then you should hold off on any aggressive denoising until after stretching and get away from just applying the STF basic stretch.  For whatever reason, sometimes Mure does not play well with DBE, especially if you try to push Mure too hard.  As Juan says you are aggravating noise in the linear phase which gets blown way out of proportion when you stretch your image.  Processing requires a delicate hand, not a hammer and going overboard is messing up your results by trying to do too much with too little.

As far as your machine constantly crashing, that seems like an issue with your system since no one else is having these issues and the PI team can't reproduce them.

Take a deep breath before you flame me in a reply.  I am only trying to help.

Best,

Jim

[rdryfoos]

If your data is as good as you say, then you should hold off on any aggressive denoising until after stretching and get away from just applying the STF basic stretch.  For whatever reason, sometimes Mure does not play well with DBE, especially if you try to push Mure too hard.  As Juan says you are aggravating noise in the linear phase which gets blown way out of proportion when you stretch your image.  Processing requires a delicate hand, not a hammer and going overboard is messing up your results by trying to do too much with too little.

As far as your machine constantly crashing, that seems like an issue with your system since no one else is having these issues and the PI team can't reproduce them.

Take a deep breath before you flame me in a reply.  I am only trying to help.

Best,

Jim

 

Jim,
Your right about the computer--the frustrating thing about it is this is my 2nd high end laptop with the issue--Both windows with intel Core i7 6500 quad processors.  I am not a computer guy so there is no chance of me fixing it, and all the computer technicians at my disposal can't address the issue with Pixinsight because they know nothing about PI and refuse to work on others software anyway.  I am stuck with it--but I can struggle through with shut downs and restarts when needed.  eventually I get it to work

As far as Mure Denoise--I use the darkbiasnoise est script as explained in the tutorial, get my noise number, use the appropriate settings, again as described in tutorial, and am generally amazed and very satisfied with Mure denoise.  I have closely inspected before and after and don't believe I am degrading data.  I run MD several times with various settings until I get the one that is perfect.  It might be settings in DBE need to be changed--I have experimented heavily and could not find a solution.  I no longer use the STF basic stretch--I either use Masked stretch or manual hist trans after a very low level STF transfer (I change the settings so they are repeatable for my lum or Ha insertions, but very low, so I manually stretch after.

Since most of the DBE issues involve LRGB images--I am thinking it is a light pollution/sky glow issue.  That's all I can think of .
Thanks,
Rodd