Registax Wavelets

[PeterCPC]

Hi All

I looked in the topics but could not see anything on this.

I wonder if anyone could give me guidance on the issue of wavelets in Registax. I looked online and found a few suggestions but these did not work for me and made the image very noisey.

Has anyone got any suggestions what to set the 6 wavelet filters to as a starting point for Jupiter say. Is it best just to leave them at the default setting? They would seem to be a powerful tool to sharpen images but you can easily overdo it I've found. 

Any suggestions would be welcome.

Peter

[Zakalwe]

I only use the top two wavelets . Click on Dyadic (under Wavelet scheme), select Gaussian (under Wavelet filter) and the whack the first two up near max. That should do the trick. If you need more then push the Sharpen up a bit using the up and down arrows under Sharpen.

Be careful with Wavelet 3...it will fatten up the details in a millisecond. If you need to use it, then the merest whiff will do it. I normally leave it alone.

If the image starts to get noisy then up the DeNoise a tad. Anything more than.10 will start to lose detail though. You can get smaller increments by typing the value directly into the DeNoise box rather than using the up and down arrows. Or, alternatively, hold down the Alt key and use the left and right arrow keys.

You shouldn't need to touch the remaining wavelets.

[RikM]

That's pretty much the same as I do for white-light solar. The top two sliders look after the fine detail. There doesn't seem to be any 'standard' scheme though. Just tweak and see.

[Steve Ward]

As said there is no set method as every set of data is different.

I personally would advise against using the "Dyadic" mode to start with , the effects of the sliders is much more pronounced and less controllable in my opinion, "Linear" and "Gaussian" gives a much 'gentler' touch .

The noisier the data the worse the final image , make sure you only use the very best frames to stack in the first place , rubbish in = rubbish out , too much noise and you will find that no amount of fiddling in Wavelets will help.

[PeterCPC]

Thanks chaps - I'll give it a try.

Peter

[Bizibilder]

For Registax 6 I have found that this scheme works quite well (there is no perfect scheme - you just have to play!):

Set Linear and Gaussian.

Start with slider 6 and move it about until you find the optimum position - ignoring any noise artifacts that this generates (this comes with practice!)

Then use the up arrow to increase the value in the denoise box (the one above slider 6).  You should increase this until all the noise you just produced with the slider disappears!   A value of 0.1 or 0.15 should do.

Then repeat the process with slider 2 and its denoise box (Again the one above the slider - You may need to increase the denoise to 0.3 or maybe a little more).

Then slider 5, slider 3 and finally slider 4 (plus their associated denoise boxes - 0.15 should be enough for all of these).

Only then should you try slider 1 - often this will not actually produce any great enhancement after all the other sliders have done their job - often you don't need to do anything with slider 1.

Here is another published tutorial:  http://www.astronomie.be/registax/linkedwavelets1.html  that tells you do do things another way!!  Just to add to the confusion of wavelets!

[Zakalwe]

So much of this gig is down to personal preferences, isn't it? Its no wonder that beginners find it so confusing..there's never a definitive "right" way to do it. :confused:

That's an interesting routine Roger. I tried it on some of my data, but I didn't like it. I might give it a bash on Lunar stuff...I find that the bottom wavelets gets good results on the fine details on the Moon.

The way that I use Came from Christohe Pellier http://www.planetary-astronomy-and-imaging.com/en/finalising-processing-registax6/

There's some really good tips on his site for planetary imaging,