Processing - theory in practice

[centroid]

I thought Id' share my recent image processing experiences with you, as while it will perhaps be 'old hat' to our more experienced imagers, it may be useful to those less experienced.

With this endless cloud preventing any imaging, and a new book on image processing to 'bury my head in', I've been busy reading (hours of it!!), and working through practical examples.

The book (Photoshop Astronomy), that I mentioned in an earlier 'post', is as the title suggests aimed firmly at image processing using Photoshop, and what for me is the final 'tweaking/polishing' of an image following the initial processing in Astroart 4 (de-bayering, Aligning, stacking, and possibly de-convolution and DDP if necessary).

I thought that I was reasonably competent at using Photoshop, but this book has shown me just how much I didn't know, and the theory behind what I'm trying to achieve.

Having worked my way through some of the sample images on the accompanying CD, I wanted to try these techniques on one of my own images, to get a feel for what difference, if any, it woud make.

A very important point, that the book's author makes, is that the most detail that you will ever have in an image, is within the original image. Every step we take in processing will degrade that original detail to some degree. In other words, process only by the amount needed to enhance the VISUAL appearance of the image. This is not the same as introducing any more information, as that of course is not possible.

So, the moral of the story is, to remember that you can't add any detail/info to the image, but you can easily destroy it.

For my 'test run', I took an image of NGC1977 (The Running Man), that I 'posted' a little while ago, and imported this into P/shop.

The image was from exactly the same combination of 12x300 sec subs, and no more data (subs) have been added.

While the image would clearly benefit from more added subs, to give a better Signal to Noise Ratio (SNR), this wasn't the object of the excercise, so don't judge the image on SNR.

I have attached both the original image, which was also processed in P/shop at the time, and new version, also processed in P/shop, but employing the new techniques that I had learned.

You can see that in the new version, a lot more fine detail has been revealed (revealed and not added remember), especially in the dark 'dust' areas, and the gaseous matter overlaying the 'running man'.

The improvement is subtle, but well worth the time and patience needed to obtain it.

The original image may look much smoother, and maybe prettier, a lot if fine detail is missing.

If and when I obtain some more subs of this object, then adding these to the original 'stack' will improve the SNR, and as such permit the extraction of yet more fine detail.

I hope that those of you just getting into astro image processing, will remember that:

If the info/detail doesn't exist in an image, then no amount of processing will create it. All we can do is use processing to make visible, as best we can, what detail lays hidden in the original image, without losing anymore data than is absolutely necessary. So, be gentle and don't 'push' the processing too hard.

Also, worth remembering is the fact that in this 'game', we never finish learning, no matter how long we've been at it. Perhaps that's what makes it fascinating.

Dave

Dave

[Greg]

Thanks Dave..A very interesting read..

Greg

[centroid]

The book can be a bit 'heavy going' in places, especially with the chapter on 'Layer Masking', but its nowhere as 'heavy going' as the 'handbook of astronomical image processing', which I also have (some 700 pages ), which does get very scientific.

For anyone with a basic 'working knowledge' of Photoshop, then 'Photoshop Astronomy', isn't difficult to follow, but trying to follow through the multitude of 'steps' involved in some processes, can make your brain hurt a bit. :scratch:

Dave

[jcm]

Hi Dave,

I have the same book. It contains a tremendous amount of info on nearly every aspect of astro image processing. I did find it quite an effort to follow the complete " working path " of some of the techniques described. Unfortunately I lacked the patience to follow right to the end some processes described.

John

[RichieJarvis]

That sounds like an interesting read. Q for you though, on the reprocessed image, how come the stars aren't as sharp as in the first pass, nor the colour differences? Just curious really.

Cheers,

Richie

[centroid]

I think the answer to that is that I was a bit more careful with the star 'shrinking' in the original image, whereas on the reprocessed image I concentrated on extracting more of the undrelaying detail of the nebula.

It was only a 'test run' of some of what I'd been reading, but a good example of why not to 'cut corners' when processing.

Another thing that this image demonstates, is if you are going to capture DSOs with an OSC camera, attached to an 80mm scope, you need to be prepared to spend the time capturing a lot of data. In this particular case, 12x300 sec subs just isn't enough.

Dave