Jump to content

  •  

CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.

Photo

Wavelet explanation

This topic has been archived. This means that you cannot reply to this topic.
8 replies to this topic

#1 Alex McConahay

Alex McConahay

    Fly Me to the Moon

  • *****
  • topic starter
  • Posts: 7402
  • Joined: 11 Aug 2008

Posted 12 October 2009 - 09:31 PM

I'm just a beginner in solar system photography with a video based system.

I am using Registax 5.

I cannot find a good description of what wavelets are and how they work.

I have found people who say "Use only level two and three." And one that says "Use a bell shaped curve on them." And more than a dozen that say "Use what works best and looks good."

Can anybody point me to a source for how these things work. Do you start with the sixth slider and come down? Or do you do level one and move up? And why?

PS....I can find nothing worthwhile using the search function here in CN.

Alex

#2 orenabah

orenabah

    Ranger 4

  • -----
  • Posts: 318
  • Joined: 21 May 2008

Posted 13 October 2009 - 01:16 AM

There is no "concrete explanation". You need to use trial and error. The wavelets can always be dragged backwards if you overdo it or don't like what you see.

Sometimes they don't need much adjustment, and other times I shove all six to the far right. It really varies, depending on the object you're imaging and how good the data is.

In a nutshell: It's all trial and error.

#3 LeedsGreen

LeedsGreen

    Mariner 2

  • -----
  • Posts: 224
  • Joined: 11 Sep 2008

Posted 13 October 2009 - 07:23 AM

Hi Alex,

In the large, I think Paul is correct, it really is trial and error. However, as a rule I start with wavelet levels 3 and 4 and push those up to 10 to see what happens. Depending on results, I then decide if "Initial Layer (IL)" setting should remain at 1 or go up to 2, 3 or even 4. I have noticed there really is no rule for which IL I use and this can vary even on the same camera/web-cam used on the same night.

Anyway, once I have chosen the IL, I then adjust the other sliders but as a rule, I notice my wavelets settings often have a bell-shaped look to them.

It is easy to spend ages fiddling away (and part of the fun) but what I also do is save a few versions of results and open in another image program to view results and then pick best before post-production begins (i.e. levels, contrast, colour, sharpness, de-noising etc)!

All great fun when the image looks a good 'un!

Hope this provides some help,

Nick

#4 Alan Friedman

Alan Friedman

    Vanguard

  • *****
  • Posts: 2072
  • Joined: 30 Aug 2007

Posted 13 October 2009 - 08:17 AM

I work on a Mac platform hence I do not use Registax. But a good way to understand wavelets or other pyramidal contrast tools is to experiment with unsharp masking in Photoshop. Open an image and apply unsharp masks to it, varying the pixel radius from .5 to 5 pixels. Wavelets does this but applies multiple passes at different pixel radii in the same action. As others have said, experimentation is key. In general, the lower the contrast of features in the image, the larger pixel diameter the processing can tolerate without returning a nasty result.

Alan

#5 Mike Phillips

Mike Phillips

    Soyuz

  • *****
  • Posts: 3693
  • Joined: 21 Feb 2006

Posted 13 October 2009 - 10:01 AM

It's some fancy math (Fourier transforms or something) on the back end, but it's use is simple and yes, trial and error.

My advice, write down what you did as you will want to know how to do it again later after so much fooling around!

Mike

#6 Mitchell Duke

Mitchell Duke

    Gemini

  • *****
  • Posts: 3384
  • Joined: 22 Apr 2008

Posted 13 October 2009 - 10:37 AM

I agree! I dont use the wavletts in registax they always seem to give me mixed results. IRIS is a great program for its processing power, and its freeware. In IRIS I use the fine, and finest wavletts which brings out the small details. The key is to play with it to get the best detail without a lot of noise. Hope this helps

#7 Alex McConahay

Alex McConahay

    Fly Me to the Moon

  • *****
  • topic starter
  • Posts: 7402
  • Joined: 11 Aug 2008

Posted 13 October 2009 - 02:29 PM

Wow,

I was beginning to suspect that there would be no firm answer. But why have six sliders if it is all just trial and error?

Can I take it that the lowest level, the sixth slider down, is what happens if you use a larger pixel radius in Photoshop Unsharp Masking, and the highests level (the first slider down) is the same as a smaller pixel radius?

Alex

#8 RedIrocZ-28

RedIrocZ-28

    Surveyor 1

  • -----
  • Posts: 1554
  • Joined: 18 Sep 2005

Posted 14 October 2009 - 09:54 AM

Alex, yes. The 6th down has a larger radius.

What you should do is take a lunar .avi and stack it, then click and hold on the button where the 1.0 appears. Click on the button on the #1 wavelet slider, and then all the way down to #6. Notice how the image of the red and green goes from very grainy and pixelated when you hold down on the 1.0 button on #1, vs the button on #6?

I normally use very little on #1 and #2. It can bring out a noisy and pixelated look to an image. And I use #3-6 quite a bit sometimes. It really, seriously, all depends on what the image can withstand with regards to not overshaprening it.

I once had a jupiter imaging run that was 8 seeing, and budging the #1 slider to a mere 4 brought out the finest details and no more processing was really necessary. On a recent imaging run, I got great Jupiters but I had to push the 3 and 4 sliders over 20, with 1 and 2 set to 3.0 and 5.0.

Its all relative ;)

#9 Alex McConahay

Alex McConahay

    Fly Me to the Moon

  • *****
  • topic starter
  • Posts: 7402
  • Joined: 11 Aug 2008

Posted 14 October 2009 - 08:48 PM

thanks,

I'll give that a try.

Alex


CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.


Recent Topics






Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics