My First Imaging Attempt

[roundycat]

Here is a quick go at the second of the two with a view of the History palette to see what has been done.

The reason I had a moan about the Info palette is that hardly anyone uses it. Seemingly everyone would prefer to use the Histogram which tells you next to nothing. The screen grab that Rik showed in post 22 is, of course, the right one. There is only one! You should, however, set the thing up properly.

Everyman and his dog has the first quadrant of the display set to RGB, no problem with that, those are the colours you are working in. The second quadrant is almost always left to CMYK which is no use at all unless you are in magazine pre-press. I suggest setting it to HSB, Hue, Saturation and Brightness. All three of those tell you useful things from time to time especially Brightness. 2 1/2% = one level. Aim for backgrounds of around 10 -20% depending on the picture.

Set the bottom left to Pixels, it helps to remind you not to post images more than 1000 pix wide!

Dennis

[terrym2002uk]

I think your stretched one is lovely.

I had quick tweak of your first version in between your posts and I think your second one is better than mine. It is smoother and still has loads of detail. I have increased the noise quite a bit as well and don't have time to fix it as I am at work. So I think I should stop offering advice on these threads.

Well done indeed.

Thanks Rik.

I didnt run any deconvolution of either of these, just stretched the detail out. Hopefully my data will be much better next time, I have learnt so much with just a one nighter. The main one being pictures through a dewy mirror are naff

[terrym2002uk]

Here is a quick go at the second of the two with a view of the History palette to see what has been done.

The reason I had a moan about the Info palette is that hardly anyone uses it. Seemingly everyone would prefer to use the Histogram which tells you next to nothing. The screen grab that Rik showed in post 22 is, of course, the right one. There is only one! You should, however, set the thing up properly.

Everyman and his dog has the first quadrant of the display set to RGB, no problem with that, those are the colours you are working in. The second quadrant is almost always left to CMYK which is no use at all unless you are in magazine pre-press. I suggest setting it to HSB, Hue, Saturation and Brightness. All three of those tell you useful things from time to time especially Brightness. 2 1/2% = one level. Aim for backgrounds of around 10 -20% depending on the picture.

Set the bottom left to Pixels, it helps to remind you not to post images more than 1000 pix wide!

Dennis

Thanks Dennis.

You really eeked out the detail in the nebula, rather nicely. I just couldnt seem to get the wisps like that, but i never deconvoluted the image before stretching it.

Few questions, which I suspect have many different answers depending on who you ask

1. Is the aim to bring out as much detail as posible? In that I mean every possible background star is made clearly visible?

2. How do you know when you have introduced artifacts, ie something that is not really there?

3. Your picture is cast green on the right and red on the left. Is this real or introduced processing noise?

[roundycat]

1. In general yes but don't exaggerate detail at the cost of showing more background noise. Compromise.

2. For the most part any artefacts you introduce will look more like godzilla than an astro object. They will be obvious but compare your work to the best in the class, ie Gendler, Croman and Schedler.

3. The casts look the other way round to me but they are there because I made no attempt to get rid of the background gradients.

Just out of interest, I never use de-convolution.

Dennis

[jager945]

My $0.02 (GBP 0.0134 )

Not all that much of an improvement over what already has been posted.

I did want to say something else about the whole background level thing. If you start to do more advanced processing, it is a bit more complicated than just 'keep 5% or 10% (or whatever your preference) and you'll be fine'.

Keeping a background level (when done at all times) during processing eats into your available dynamic range. Human eyes like contrast. Human eyes are also fallible. If you combine these two nuggets of information you can cheat your way to a perceived higher dynamic range and a 'punchier' image.

For example, this is exactly what Unsharp Mask does. It artificially inflates very local contrast by exaggerating blackness around the star and increasing brightness in the star. We have all seen (I presume) the black ringing/panda eyes (officially called the 'Gibbs phenomenon' effect around stars with Unsharp mask applied if these stars reside on a non-black background (such as bright background level or nebula).

While in the latter case of star sharpening this is an undesirable effect, we can also use this dynamic range tinkering to our advantage. Let's consider an image of a nebula.

Depending on your preference, you'd have a background level around it (where dark space is), but for pixels in the nebula itself adding this background level is just a waste of dynamic range - it just makes things harder to see; the nebula itself is typically brighter than the background level anyway and even if it is not, optimizing the dynamic range for the nebula itself between pure black (e.g. without background level) and pure white will do wonders for the 'punch' of your image. No one will ever see/know you didn't add the background levels to the nebula; welcome to the obscure art of local contrast optimization!

So, long story short, this is why you would want to refrain from *****-nilly maintaining that background level at all costs.

Lastly, Terry, to answer some questions;

1. Is the aim to bring out as much detail as posible? In that I mean every possible background star is made clearly visible?

No, not at all. The greats like Ken Crawford even intentionally leave some details blurry, just to emphasize others. A growing number of people even remove stars from their image altogether, just to show an underlying nebula in all its glory (there are some magnificent examples of this by Fred Vanderhaven and Mike Sidionio). It totally depends on how you want to present your subject and what you feel is the important subject in your particular image. There are many different scales (stars, dust lanes, big billowing clouds) in an image, all vying for attention and emphasis. Emphasize them all and an image looks busy (not a bad thing, just a choice). Emphasize just a few and the eye is drawn to them.

2. How do you know when you have introduced artifacts, ie something that is not really there?

Like Dennis said, compare with other images. Tying in with your 3rd question, beware of green tints. Very few objects in space have a predominantly green appearance in any part of their complex (a couple of nebs excepted, such as M42's core, the Tarantula neb's core, and probably some more OIII rich nebs).

There is a great free PS plug-in called HastaLaVistaGreen (HLVG), which gets rid of any green in your image. The reasoning is that if a pixel is green and your color balance is correct, then that green pixel must be there because of noise. PixInsight (SCNR) and StarTools (Cap Green) have a similar tool.

Hope this helps!

Ivo