AtrousWT Process

[esraguin]

Well its time to call for help in my image processing,first my equipment Meade DSI11 col camera, Lxd75 mount, SC8in Scope.  Images are captured in Raw format in Nebulosity, Debayering in Nebulosity1, Stacking, then BackGround Adjust.  Then take image into Pixinsight.

Now my understanding of my Image is that there are four parts, 1 Stars 2 Nebulosity 3 Background 4 Noise.
In the many tutorials on PI i have learned so many things and understand a bit of most, so much so my imaging is now starting to be enjoyable because of Pixinsight.  In the four processes above i get maybe three out of four right or two out of the four depending on the image.

But there is one process AtrousWT in which the more i read about it, the more i think its the key to taking the image apart at different levels and sorting that part without destroying other key areas.  I have read, read, read info on this subject, but because of my lack of brain power, just dont get it.  Then i put on the big hat with the D on it and stand in the corner.  I have a feeling AtrousWT is the most important tool on PI and I WANT TO UNDERSTAND SOME OF IT, so any help in this matter would be greatly appreciated.

Alex

[Nocturnal]

Hi Alex,

why do you think AWT is the most important tool in PI? It's important sure but pointless without histogram transforms for example. Can you post some of your images you feel would benefit from AWT? I personally use HDRWT more often as it's a little less complicated. I still don't understands how it works though I mean I know what is *does*, it highlights small brightness differences in images with a high dynamic range such as galaxy cores and some nebulae. I also understand wavelets divide the image into layers with different detail scales. But then what it does is unclear to me.

[mmirot]

That's funny I find the HDRWT more usufull and easier too.

I don't alway get the best results with ATrous and it takes longer to get a good result. 

Sometimes unsharp mask works nicely on deepsky images.

On galaxies and planets ATrous multiscale processing has greater benifits.

It would be nice to here from the some of old the hands when they perfer ATrous.

Max

[Nocturnal]

I just finished part 3 of my PI article series for Astro Photo Insights and it has a section on HDRWT. Nothing deep, mind you. Just the basics on how to narrow down on the right set of parameters (try, fiddle, try, fiddle). I felt bad not being to explain the magic of wavelets better than that but I think it's useful as an introduction anyway.

[Harry page]

Hi

I think Atrous is something Juan needs to do a tutorial on and or instructions for use as I sometimes get a result using this
but have not got a clue how or why and therefore can not reproduce it.
So Juan are you up for it


Regards Harry

[Carlos Milovic]

By the way, HDRWT uses the ATWT to do it's magic so, saying that ATWT is one of the most fundamentals and powerfull processes in PI is very close to reality.

Again, once I have some time (sorry, but I can't see it happening until december, to be truly) I'll search for an old article I wrote abput convolutions and wavelets that should explain all the basics. I think that it was in spanish only, so that's the reason I'm delaying this task until I have more time.

[Simon Hicks]

I thought that lots of the PI processes used ATW somewhere in them. I know HDRWT does, I suspect that StarMask does, deconvolution (wavelet regularisation), ACDNR (edge detection/protection?), DarkStructureEnhance?, HDRComposition, StarHaloReducer, etc. Maybe I've got some of these wrong, but I'd suspect there are other processes / scripts that use ATW somewhere in them. They seem to be very useful little tools. And I'd like to know more about them as well.

[Nocturnal]

I suspect there's a general Wavelet engine somewhere that gets used all over the place. If I was bored I'd review the PCL docs to make sure

[esraguin]

Thankyou all for ur comments and advice, This all started when last week i imaged part of the Veil Neb, the stacked image looked quite good, wispy nebulosity through the image ,stars looked good ,background not to bad.
So into Pixinsight applied HDRWT nebulosity looked fine,star mask on so stars were fine, background rubbish,so now started to sort background nebulosity went to mush.and so on.But at the weekend will give it another go.

This Tutorial in PixInsight LE Tutorials...Multiscale Processing of Deep-Sky Images with PixInsight By Vicent Peris seems to answer my questions but trying to follow it i get so far then it loses me,because of my inability to understand,this tutorial struck a cord with me as regards working on Deepsky images,so a better understanding of
AtrousWT Process is something im looking forward to in the future

Alex

[Cheyenne]

By the way, HDRWT uses the ATWT to do it's magic so, saying that ATWT is one of the most fundamentals and powerfull processes in PI is very close to reality.

So.. I have a question, could someone maybe describe how one would mimic HDRWT with ATWT?  I think it would be educational in understanding ATWT.

[Nocturnal]

Good question Cheyenne

[Juan Conejero]

To everybody in this thread:

Now that PI 1.5.9 seems stabilized, I'm going to take a small break in development. Among a long list of pending tasks, there are of course new video tutorials. Some of the first videos will be a small collection about the ATrousWaveletTransform tool. So, patience (once more ), it's a matter of weeks.

I wouldn't say that ATWT is the most important tool in PixInsight since there are many essential tools, as Sander has pointed out, such as HistogramTransformation, PixelMath, etc. However, it is true that all "advanced" processing tools in PixInsight are either based on wavelets, or use wavelets in some key parts of the underlying algorithms. Since the beginning (back in 2003-2004), we always have made a strong bet on multiscale processing. Multiscale techniques and procedures are a completely new paradigm, a new vision of the data and their analysis that changes the way we understand and implement most image processing tasks. In PixInsight, there are wavelets almost everywhere: along with the ATWT and HDRWT tools, where the use of wavelets is evident, we use them in not-so-obvious tasks such as image registration (structure/star detection and multiscale cross-correlation) and image integration (noise estimation for optimal image weighting and goodness-of-fit evaluation). We'll use them also for image calibration: noise estimation for iterative dark frame optimization.

[mmirot]

I think this might be a good point for advertising PI.

The PI plateform is very different from other packages in this respect.

Also, the approach to to color seems quite a bit different compared to PS.

[dhalliday]

Well...I feel better this AM after reading this...I thought it was just me !!!

Dave