Registax vs LR deconvolution shoot-out on Jupiter + little animation

[michael.h.f.wilkinson]

As suggested by Freddie, I just tried to compare LR deconvolution to RS6 wavelet sharpening on my best Jupiter data to date. Both run on the same drizzle 1.5x stacks, and both with the same post-processing in GIMP (unsharp masking plus curves). I did not rescale the images to 640x640 this time

LR deconvolution in Astra Image 4.0:

Registax 6 linear scheme, Gaussian wavelets: 0.20 denoise setting on first wavelet band, all wavelets nearly maxed out:

To my surprise, I must conclude that RS6 seems to have it. Thanks to Freddie for suggesting checking RS6 out. I stacked 5 images out of the batch into a little animation:

[JamesF]

I agree, though the additional sharpening in R6 has brought out more noise, particularly in the blue channel by the looks of it.  And LR seems to have created a limb artefact that isn't present with R6.

I'm intrigued by the two features present near the north west limb in only one frame of your animation.  Any idea what it might be?

James

[michael.h.f.wilkinson]

I agree, though the additional sharpening in R6 has brought out more noise, particularly in the blue channel by the looks of it.  And LR seems to have created a limb artefact that isn't present with R6.

I'm intrigued by the two features present near the north west limb in only one frame of your animation.  Any idea what it might be?

James

I agree the RS result is a bit noisier, but I do not mind that too much. I could spend some time finding better settings to cure that.

I know what the spots are: dust on the sensor

[Uplooker]

I agree, though the additional sharpening in R6 has brought out more noise, particularly in the blue channel by the looks of it.

James

Hi James,

Not wanting to hijack the thread, would you mind elaborating on your comment about more noise, particularly in the blue channel? How have you discerned this?

Ian

[Space Cowboy]

That Registax sharpened image with a bit more noise control is the way to go Michael. You should try Image Analyser its very good for dealing with noise.

[michael.h.f.wilkinson]

OK, taking on board some advice, I have changed the colour balance using auto balance in RS6, altered the noise control settings, ditched unsharp masking (which was stretching the noise quite a bit), adding a contrast stretch and subtle increase of saturation, and we arrive here:

Scaled down:

Animation:

[Freddie]

Looking good. I think this is the best I have seen from you.

Edited February 8, 2015 by Freddie

[michael.h.f.wilkinson]

Thanks, Freddie. I also think this is the best Jupiter I have done so far. It is certainly the best to come out of the ASI120-MC I have. The DMK21 with filter wheel is usually a clearly better performer, but I think it might have been hard pressed to improve on this.

[JamesF]

Hi James,

Not wanting to hijack the thread, would you mind elaborating on your comment about more noise, particularly in the blue channel? How have you discerned this?

Ian

To me it appears that the lighter bands in the image have some blue speckling in them that looks more like noise (in that it seems quite random) than a feature of the atmosphere.  I get a similar thing in my own images.  It does seem to depend what monitor I'm using to view the images though.  Some seem to show it well whilst on others it's barely noticeable.

James

[neil phillips]

That's  the best I have seen from you by a long shot. Way to go Michael

When you do the denoise technique on registax 6 what I do is resize the image on registax under resize. to see clearly all the noise. if its still bad you can up the denoise. the aim is to control it with the least amount of denoise as possible. Each session is different you don't want to stick to twenty. Play it by ear, sometimes 25 is needed. especially blue channels

Really good to see these shots they are very nice indeed.

[michael.h.f.wilkinson]

Thanks Neil. I tend to experiment a bit with low noise control settings. The value I used now was 0.25. I might try Perona-Malik anisotropic diffusion to see what that does in terms of noise control.