deconvolution or denoising first ?

[glowingturnip]

So I'm working my way through my copy of the new Pixinsight book and in parallel am processing my latest effort.

Which do you do first whilst still in linear stage - do you deconvolve first or denoise first ?

I've always done my denoising first first since I don't want the deconvolution script to start sharpening my noise, but the book suggests the other way round.

 

Cracking book by the way, in my opinion.  I wouldn't recommend it for Pixinsight newbies, Harry's videos are still the best for that by a long way, but for setting up a good workflow, for getting more out of the tools, and using more of those parameter settings it's very good.  Fortunately I've still got a few secrets left

[MARS1960]

Mmmmm, well according to the book, if using OSC then you deconvolve first, i would be inclined to follow the book, but as a relative newbie to PI i had never come across deconvolution before.

[Filroden]

If you  apply a luminance mask using deconvolution you will protect the background and weaker signals from being sharpened. I use a fairly harsh mask so only the strongest signal is affected (by really raising the black point in HistogramTransformation).

Likewise, the inverse of that mask (probably one that is less harsh) stops you applying denoise to stronger areas.

I guess in theory the two masks are mutually exclusive so order of process might not be that important but I would still use deconvolution first to prevent the risk of trying to sharpen something you've just blurred! I'd rather blur something I'd sharpened where the masks don't quite line up.

[ollypenrice]

Denoise is blur by another name. You would only ever want to apply it to the faint signal where the noise is to be found. Deconvolution is a sharpening routine which you would only ever want to apply to the strong signal which will support it. So, assuming you have masked well in both cases, it shouldn't make much difference which way round you do it! Neither will be working on the same parts of the image as the other so both should be working on virgin data within the image. However, if I had to choose I'd deconvolve first because it would be absurd to blur then sharpen.

Olly

[glowingturnip]

hmm, good points - I've been careful about masking for deconvolution, but maybe less careful about masking for denoising - will experiment

[glowingturnip]

so here's a question...  diffraction spikes - sharpen or mask ?

I'm doing Alnitak and friends and I've got some stonking spikes

[ollypenrice]

17 hours ago, glowingturnip said:

so here's a question...  diffraction spikes - sharpen or mask ?

I'm doing Alnitak and friends and I've got some stonking spikes

There's no spike like an Alnitak spike! 

Matter of taste? I'd probably sharpen.

Olly