View Single Post
  #7  
Old 21-03-2017, 07:58 AM
Placidus (Mike and Trish)
Narrowing the band

Placidus is offline
 
Join Date: Mar 2011
Location: Euchareena, NSW
Posts: 3,719
Quote:
Originally Posted by MLParkinson View Post
Thank you for your positive feedback Mike.

You would like to see a “straight” Hubble rendition? Is there any such thing?

To me, the Hubble palette is a challenging exercise in colour manipulation. In the process, much of the useful H alpha depth and contrast is suppressed in favour of the low depth and contrast of the sulphur channel. Andy01 has a knack for building exceptional Hubble palette images. Not me. Perhaps it is easy when you have the right tools, and you know how! I’ve still got a lot to learn!

The H alpha channel has tremendous bit depth, and the sulphur channel little, even when I expose on the sulphur channel for 40 minutes. If I construct a straight Hubble palette image the result will be garish green because the H alpha green channel dominates the palette, with the sulphur red channel far too weak to prevail in the image.

Perhaps one of the leading photographers in this forum might write and post an easy to understand tutorial on how to achieve a pleasing Hubble palette result using PI.

Best wishes, Murray

Hi, Murray,

We wouldn't call ourselves narrowband experts, but we think we're on our green P-plates. Our approach: Take enough SII so that after background subtraction you have roughly equal total photon counts in all channels. Then the problem completely goes away, and the choice of palette is decoupled from the relative intensities.

The universal problem as you elegantly point out is that sulphur emission for most objects is typically say five, ten or twenty times weaker than the H-alpha emission. That means that if one only collects enough data to get the H-alpha channel looking good, then one hasn't really done a three channel image. It's only a 2.2, 2.1 or 2.05 channel image. You can't get round that shortage of information by any choice of colour, only by getting more information!

We tend to throw a complete night - say 6 or 8 hours - just at the sulphur channel, sometimes a couple nights say 14 hours - in order to get an adequate amount of information for a true three-channel image. Another approach, if you are normally over-sampled, is to use 2x2 binning (with dithering) on the sulphur channel, in which case you get four times as many photons per pixel in the SII channel, and you can get enough info in the SII channel in a couple hours or so.

So once you have enough SII data - either by throwing hours at it or by 2x2 binning, the next step in our mind, regardless of what mapping you have chosen, is to carefully set the zero point for each channel to the foothill of the histogram, so that black is black, but there is no information loss. The final step is to set the white point so that the average value for each channel is the same. No channel outweighs the other. For example, if (after subtracting moonlight and sky glow) there is twenty times as much H-alpha up there as SII, you make the SII channel twenty times brighter, and now they are balanced. If there is 5 times as much H-alpha up there as OIII, you make the OIII channel 5 times brighter, and it now balances both the other two. All three channels will have the same average brightness.

It is very important to notice that if you are using 32 bit processing (which you should) you have lost no information in the H-alpha channel by making the SII channel brighter. A corollary is that no harm comes from having even more information than you need in the H-alpha channel. You can never have too much.

The approach I've outlined means the image isn't ghastly green (like our early stuff used to be), but it doesn't have to be riotous red, or billious blue, or magenta or cyan either, but shows the differences between regions. There is even some advantage to using a modified version of this 'colour agnostic' approach on straight RGB galaxies, because it means you can up the saturation quite strongly, to show the differences between regions - core and spiral arms for example - without the galaxy as a whole becoming all the one colour.

Confession: For the Lagoon end of Bigfoot, the approach I've outlined works really well. For the Big Toe end, there is so little SII that we gave up and did a two channel image.

Very best,
Mike

Last edited by Placidus; 21-03-2017 at 08:13 AM.
Reply With Quote