wavelet sharpening - which one?

[devilgas]

i'm looking at doing wavelet sharpening on my solar images. i had downloaded the trials for astra & topaz, but both have now run out so i can't remember which was the better. i've also tried gimp with the wavelet sharpening plugin & also had a pop at registax.

which of these (or something different) would people recommend? i've struggled with registax (i'm using an slr at the mo) and the gimp plugin is very simplified. now it might be that i'll never achieve what i'm seeing on the forum whilst using an SLR, but i feel i need something to get more detail from my images. i'd prefer the outlay to be low / zero

any suggestions greatly appreciated!

[PeterW]

I have use registax and AviStack. I tend to apply a little of the first wavelets and then less ofnthe next one. I test out how much to apply by adding loads till the image looks rubbish and the backing it off till I can't see the artefacts. Recently someone suggested running an unsharp mask on the resultant image, seemed to help,things slightly. See how you get on, there should be plenty of online help,about the best settings, I have just played a bit. Lunar images are a good starting point as craters are easier to see differences on than the sun.

All the best

Peterw

[markt]

I use Registax 5.1 specifically for the wavelets function, it's a free download...

[devilgas]

thanks both. currently have regi 6, so will download 5.1

[ScubaMike]

A guy I know that produces stunning solar images uses Registax too.

He uses most of the bottom wavelet slider, then progressively less of each one above. Makes a kind of parabola shape.

Seems to work for him.

[Cschur]

ID go for registax 6, so far it has the best most solar imagers are boasting about. But as far as the sliders, I only use the top one or two. The rest just make big shadows and distort the details. Set the multiplier for .3 instead of the default of .1 as well, and just try the top slider first. The goal is to supply enough sharpening to not get any dark halos around bright objects!

[ScubaMike]

Hmm, just tried using top slider most. Distorted my images terribly.

Lots of image breaking up.

Tried with bottom slider most and got acceptable results

Maybe there is something I'm not doing.

[PeterW]

Hmm, just played with some of my solar and lunar images. Seems they all need subtly different treatment. I tend to find that i get some peppery noise when i use too much of the top few as well as artefacts. I am using the linear, default wavelets in Reg6. I have to agree in some situations when I had some of the top ones set, the bottom one did do a quite good job. In other circumstances the bottom one mushed everything up. I was thinking that there is a reasonable set of settings for images, but I am not too sure now. There is a fine line between under and over processing and it is hard to explain where I or other people draw the line. Sometimes it also depends on how well a job Registax has done, a bit of fluff from the seeing can make it impossible to get a really crisp result.

Looks like more experimentation and data is needed.

Cheers

PEterW

[markt]

i'm looking at doing wavelet sharpening on my solar images. i had downloaded the trials for astra & topaz, but both have now run out so i can't remember which was the better. i've also tried gimp with the wavelet sharpening plugin & also had a pop at registax.

which of these (or something different) would people recommend? i've struggled with registax (i'm using an slr at the mo) and the gimp plugin is very simplified. now it might be that i'll never achieve what i'm seeing on the forum whilst using an SLR, but i feel i need something to get more detail from my images. i'd prefer the outlay to be low / zero

any suggestions greatly appreciated!

Are you stacking the images from your DSLR prior to applying wavelets? What file format are the images you are trying to apply wavelets to? Are your solar images WL / Ha / other narrowband wavelength?

As others have intimated here the exact settings for wavelets is a very individual thing - depends on system used, aperture, f ratio, imaging system, whether the image is under sampled / over sampled, seeing conditions - the list goes on. Personally I don't think it possible to have a 'one size fits all' setting that works for everyone, trial and error and perseverence are the only way to find what works best...

[devilgas]

thanks guys, this is all very useful. just need to drag myself away from MW3 for a few hours to try it all

Are you stacking the images from your DSLR prior to applying wavelets? What file format are the images you are trying to apply wavelets to? Are your solar images WL / Ha / other narrowband wavelength?

As others have intimated here the exact settings for wavelets is a very individual thing - depends on system used, aperture, f ratio, imaging system, whether the image is under sampled / over sampled, seeing conditions - the list goes on. Personally I don't think it possible to have a 'one size fits all' setting that works for everyone, trial and error and perseverence are the only way to find what works best...

hi mark, shooting in Ha with a PST. shoot in RAW, output from lightroom as a 16 bit pro-photo colourspaced PSD file, then process in a very old version of photoshop. i was hoping to do the wavelets in photoshop, but have tried (using 16 bit TIFF) wavelets in registax. have also tried 16 bit TIFF and stacking in registax, but something goes wrong somewhere (that'll be me) in the process and i just end up with mush. same happens when i try to process webcam AVI files - the output looks like a motion filter has been applied. it's most definitely down to my lack of understanding on how to use, or get the best from, registax.

[Merlin66]

For my Ha solar imaging with the DMK, I generally only use the #6 to #4 wavelets...

Start with the bottom #6 and use the slider back and forth until you find the sweet spot. (This setting generally smooths the poor seeing )

Then move upwards; when you get to #4 the settings are generally pretty low ( slide them all the way up and see what happens to the image!!)

#2 and#1 I find are sometimes...not always ....useful to get that last bit of detail out of the proms.

Don't go for excessive contrast....it's not there most of the time, and the results always look "overcooked"

HTH

[Photosbykev]

Are you separating out the R G and B channels and then throwing away the G and B channels as they contain no useful data and just introduce noise.? then process the Red channel as a grey scale image before re-colouring it in PS if you want to add colour

My basic DSLR processing routine is here PST Solar Imaging PhotosbyKev

[devilgas]

@merlin - thanks. i'm presuming wavelet sharpening is done as the last step before colouring? this is the norm with general photos.

@kev - the process i use is very similar to what's on your site, so yep, separating channels. using the pro-photo colour space (i initially did this by accident) gives 'useable detail' in the green channel. for the surface detail, i compare the red & green to see which gives the better detail. generally it's the red channel. i then build the proms & surface layers from the greyscales, doing a final colouring with curves.

half the problem seems to be the vague focussing on the PST. is it normal for it to feel rubbery?

when i turn the focus knob (i have the larger diameter add-on knob), there is slight springy resistance as if the rod passes through an o-ring which needs a bit of lube on it. as you turn, there is initial resistance, then it gives. this makes fine tuning of the focus laborious. also turning the focus knob moves the internals of the scope slightly (i use live view on the dslr, zoomed in to 10x, turning the focus wheel visibly moves the image around on the screen). is this normal? it's been like this since it was bought new in may '11

i've been resisting the urge to strip it down, but increasingly feel it's inevitable....

so, "which wavelets" has turned into..... "which wavelets and is my PST broken"

[Merlin66]

Yes,

wavelets are the last thing to be done before....final processing....

There may be a little sharpening/ gamma curve adjustments then onto the colorising.

The focuser on the PST is very average, I think we all find the same issues as you.

(If you do open the side panel and have a look at the working of the pentaprism and the "slide" mechanism....you'll the problems...but not the solutions....)

[markt]

Wavelets are the FIRST thing to do after stacking images - you are stacking more than one image? If not then you will be left with a worse signal to noise ratio than if you don't stack. Like Kev says remove the green and blue channels else this will introduce more noise... All this noise just contributes to 'mush' in post processing if you don't minimise it.

I have to be honest here, using a DSLR for Ha imaging is far from the ideal solution...

[Photosbykev]

.....

I have to be honest here, using a DSLR for Ha imaging is far from the ideal solution...

The change from a DSLR to CCD camera is staggering

[devilgas]

kev / mark - yep, i appreciate swapping to CCD is the way forward, but i can't justify the outlay at the mo on what is essentialy a 3rd hobby. the dslr's get used extensively, whereas a CCD purely for astro work....

that said, is there anything worthwhile at the lower end of the CCD price scale?

thanks merlin. at least i now know that the PST isn't broken and that's just the way it is

re stacking - i'm presuming that i should emulate a CCD and stack the isolated red channel rather than the entire red image? e.g. image 1, extract red channel, save 16bit grey image. image 2, extract red channel, save 16bit grey image etc etc. then stack greys from 1-n

after stacking, do wavelet sharpening.

back into PS, build surface and proms layers, then colourise?

i got a load of images from the big prom 10 days ago, but have only ever worked with 1 image at a time.

sorry for all the questions!

[markt]

re stacking - i'm presuming that i should emulate a CCD and stack the isolated red channel rather than the entire red image? e.g. image 1, extract red channel, save 16bit grey image. image 2, extract red channel, save 16bit grey image etc etc. then stack greys from 1-n

after stacking, do wavelet sharpening.

back into PS, build surface and proms layers, then colourise?

That's the way

[devilgas]

That's the way

thanks. i'll have a play later, although we've got clear skies at the mo and i want to have a pop at the orion neb later

[devilgas]

If you do open the side panel and have a look at the working of the pentaprism and the "slide" mechanism....you'll the problems...but not the solutions....

things came to a head with the PST yesterday, so off came the panel. the focus mech is shocking! until i can devise some way of improving the 'design' the best i could do was to apply several layers of 2" wide sellotape to the back of the panel. this in turn presses tighter against the felt pad on the pentaprism, making the entire thing less floaty....for the time being.

even on the grounds of cost, i'm stunned that what was inside the PST was considered a solution. or am i being too picky because i'm trying to use the PST for imaging where focus is much more critical than eyeball mk1?

[Merlin66]

The original design of the PST was for visual use not imaging...

IMHO the ultimate solution is a Stage 2 mod. This fixes ALL the problems of the black box and focusing!

Details available on request.