Log in

View Full Version here: : Processing Ha mono images


RoundStars
26-04-2020, 03:22 AM
I’m sometimes perfectly happy with an Ha image without color. But I yet haven’t found articles explaining how to make monochrome images “pop”. I mostly use Photoshop levels and curves, but would like to get better detail and crispness in my Ha images. Can anyone point me to a tutorial that'd stretch my skills? Thanks!

glend
26-04-2020, 04:00 AM
There are plenty of Utube tutorials on using Photoshop to process colour and mono images, layering, etc. It is more than using levels and curves, and never let auto functions do that for you, the artistry is in the manual control of Photoshop functions. Check out Utube.

Placidus
26-04-2020, 09:45 AM
The broad principle involved is to produce an increase in local contrast. By far the best tool for that is wavelet sharpening.


A general workflow:


- Set the zero point (black point) to the foothill of the histogram. The foothill is the point where the cumulative frequency histogram suddenly kicks up from zero frequency. That single step removes all the fog without losing any meaningful data.

- Apply cautious and gentle deconvolution, being careful to stop well before you get "worms" or a "glassy" look. I find about 5 rounds of Richardson-Lucy deconvolution, with "anti-panda" to prevent black ringing around the stars, is about all one can do, unless one has dozens of hours of exposure.

- Apply an arcsinh stretch, to get a sufficiently bright image.
- Apply wavelet noise reduction, just enough so that the background is smooth
- Apply wavelet sharpening. This is the key tool for producing increased local contrast. Anything else - unsharp masking for example, just about anything you can do in PhotoShop - is nowhere near as powerful.


Don't apply wavelet sharpening on the linear image. Only apply after the arcsinh stretch, or bright bits and black bits will burn out, and you'll get rings around stars.


If you still get burned out disc-like stars or rings around stars, you will need to make a soft-edged star mask so that the sharpening is applied to everything except the stars.


Personally, I don't actually use an explicit star mask. I work in a more intuitive way that is mathematically identical. I produce a starless, or nearly starless image. It doesn't have to be a publication quality starless image, and can be full of artifacts, just so long as it doesn't have any sudden transitions or really bright stars in it. I then perform the wavelet sharpening on the starless image. Then I produce a "stars-only" image: the difference between the original and the starless.


If you are doing SHO, this is a good time to manage star colour. There are many approaches: (a) desaturate the stars-only image to white; (b) toss the bloated OIII and SII, keeping only the H-alpha, and desaturate to white; (c) try to synthesise the actual star colours, by using the Ha for red, OIII for blue, and some mix of the two for green, or (d) drop in heavily exposed RGB stars instead. That makes the "stars and starless" approach very powerful and flexible. Then finally add the stars back into the sharpened starless image.



Wavelet sharpening works under the hood by dividing your image into say 6 levels of detail: detail at the single-pixel level, detail at the 2-pixel level, detail at the 4, 8, 16, 32 pixel levels, and a residual level (everything else). The contrast is then increased by as much as you can safely, in each of those 6 or so levels of detail excepting the residual. Generally, you don't increase the contrast at the single-pixel level very much at all, because in a properly over-sampled image, there should be no information at the single pixel level. You apply by far the greatest increase in contrast at the 2 pixel level (that is why it is called "local"contrast. I find that in our images, you can generally get away with about a 70% increase in contrast at the 2 pixel level. Then progressively less at the other levels, say 50% at the 4 pixel level, 35% at the 8 pixel level, and 25% at the 8 pixel level.



Those are the general principles.


PixInsight obviously provides all the tools, but you'll need someone to sit down with you, and take you through it. I found it easier to master the maths and write my own package. I had it knocked over in only a couple years, much quicker really.



Best,
Mike

Andy01
26-04-2020, 03:01 PM
You're using PS already, and NIK filters Silver FX pro plug in is an awesome and reasonably priced tool to enhance monochrome images.
Colour FX pro is a staple of mine for processing, as is Define and Output sharpener.
These are all bundled together - download them here. (https://nikcollection.dxo.com/)
Cheers

Andy

multiweb
26-04-2020, 09:40 PM
Good post Mike. Bookmarked for future reference. :thumbsup: