2-Minute Tutorial: Removing Gradients with DBE (OK - I really blew the 2-minute part big time!)
Вставка
- Опубліковано 6 бер 2023
- This is another in a series of concise tutorials on working with Pixinsight. It is geared towards new astrophotographers and those just learning Pixinsight.
This video demonstrates how to use DynamicBackground Extraction (DBE) to gradients in your images.
THIS VIDEO IS almost 15 minutes long, and I questioned whether I dared release this under the "2-Minute Tutorial" Series because this is so off the mark timewise.
But I see the 2-Minute Tutorial series as being:
1) shorter - at least shorter than some of the super long tutorial videos out there.
2) Focusing on one narrow topic
3) Suitible for people still learning Pixinsight.
So with apologies, here is the worlds longest 2-Minute Tutorial!
WARNING: This tutorial is a little longer than others - the "2-Minute" name should be interpreted as "Shortish videos" rather than literally - especially in this case! :-)
These videos are associated with my Website:
CosgrovesCosmos.com
Which covers all aspects of my personal journey into Astrophotography. There you will find images, gear, tips, and techniques!
Is there a 2-minute tutorial that you would like to see? Leave me a comment, and I will work on it!
Please support the channel by liking the video, subscribing, and ringing the bell!
Thanks,
Pat - Наука та технологія
You are a very good teacher. This covers all I need to know to get started.
Thank you - I really appreciate the feedback!
All the best,
Pat
Those videos are really helpful and to the point!
Thank you so much for saying this - very much appreciated!
Great video again Pat, I’ve hit a wall on my workflow so going to split channels with dbe and other processes to see if that can take my images to ‘the next level’.
Question about dbe, what would you do if nebulosity covers the whole image?
Thanks Ian!
You ask a very interesting question.
If your field is mostly nebula, but some patches are clearly background sky - you can try placing sampling squares in those regions manually. Then use just these to create a model and see if it makes sense with what you see in the image. If it does, try using it to drive the fix. This can work, and I have used this on images successfully before.
On the other hand, if the nebula is truly spanning the entire image with no true background sky showing - can you even tell if there is a gradient? Maybe not. If that is the case, then just skip DBE and move on.
In the last case, there is no way that either DBE or ABE can truly create a model by sampling. If you try and force it to, it will create a model based on sampling the nebula and trying to subtract it. You don't want to subtract valid signal data from your image - this will cause distortion.
If you can see no background sky but still see what looks like a troubling gradient that is not part of the nebula, you might try manually correcting this in Photoshop using a gradient filter. I have never had to do this yet myself, but I have heard from people who have used this method successfully.
Hope that helps!
All the best,
Pat
Very cool video Cos. Do you like DBE more than GraXpert? What is the latest recommendation on when to run Dynamic Crop? I heard that the new SPCC doesn't like an early crop
Thanks very much Craig!
To be honest, I have not tried GraXpert. I am very used to DBE, and I find that I have a lot of control over how it acts. I seem to quickly handle my gradient issues with it and have not felt compelled to look for an alternative. This is unlike NR or Deconvolution. I was getting good results with the traditional tools there, but they seemed time-consuming and sometimes hit or miss. But then I found NoiseXTerminator and BurXTerminator and found them to be 1) easier 2) Quicker, 3) able to achieve better results - and I was sold. I will try it at some point, but so far have not felt the need.
From my perspective, DynamicCrop is the very first tool I run. Stacking gives you ragged edges, and these are not true signals - they are just bad or messed-up data. I do not want that data to influence any processing downstream, so I always do this first. SPCC is a statistical modeling tool; leaving bad data in would tend to mess up the results, so I do not know why you would not want to do crop first.
I see crop and DBE as corrective measures for your data. You are cleaning up undesirable signals that have not been handled fully by the calibration process.
So Crop+DBE before SPCC. Then BlurTerminator, Then a light linear NR with NoseXTerminator is my recommendation for Linear processing.
Hope this helps!
All the best,
Pat
Just adding to Pats reply, before spcc run image solver if you’ve cropped the image.
Ahhh - Yes - that is a good point Ian.
The new paradigm in Pixinsight is that WBPP will provide the astrometric solution. This is only useful if you also use WBPP auto crop capability. Then the master image is already cropped and has the astrometric solution.
If you use WBPP to add the astrometric solution and don't use the auto crop, you would be forced to use Dynamic Crop to trim the edges outside of WBPP. This would invalidate the astrometric solution, so you would have to manually run the ImageSolver script to add a new solution back in.