Stacking & processing software & Fourier transforms

[MishMich]

First.

Many years ago I did a Image Processing unit as part of my IT Masters, and one of the things I had to do was write a program for doing Fourier transforms. As I recall, it wasn't rocket science, and if I could be bothered I am sure I could dig out my notes and knock something up, although it has been ten years since I did any coding. From my recollection, applying Fourier transforms can be quite useful in processing astrophotographs, as one of the things we had to do was to enhance a lunar image. So, I found a couple of sites that do Fourier plugins - one for Gimp and one for PaintShop.

The Gimp one gives no instructions, and the PS one gives instructions that bear no relation to PSE7 (it describes assigning RGB channels - but I only get a layers tab, not a channels tab, as described in the instructions). When I did it in software, it was all command-line stuff, but in both plugins you end up with a blue image, but it is not obvious what you then do with that.

Does anybody know about Fourier transforms, and how you use the plugins - either in Gimp or PSE - to enhance digital images?

Second.

From what I can tell, both DSS and Registax operate by averaging, or applying median filtering. I find DSS works out for me with DSOs, Registax for planets & lunar. What I think happens with DSS is that if I stack a range of frames, say 5x1min, 5x2min, and 5x3min, I get a result roughly equivalent to a 2 minute image in intensity. Is there any way to get it to work in a way that is cumulative?

Third.

Buying a package - I'm not buying Maxim DL, I've decided, I downloaded the trial version and I didn't take to it. The next best thing I can make out is AstroArt. I was wondering if anybody has this, what they think of it, and qhat you can do with it. Could I do both planetary and DSO stacking, as well as post-processing and enhancement? Or would I still need to use Gimp &/or PSE? I can't justify shelling out on CS4, nor PixInsight (which I also tried out and found less than intuitive) - but want to get round the 8-bit colour limitation these two packaged impose (Gimp on everything, PSE7 on anything that is a bit more than tweaking levels and sharpening - such as Noel's actions).

Regards,

M.

[themos]

If complex interfaces don't scare you, check out Iris.

Iris software

I know a bit about FFTs but no real knowledge on how they apply to image processing, I'm afraid.

DSS works cumulatively, as far as I know. There's a yahoo group for asking questions.

DeepSkyStacker : DeepSkyStacker

[roundycat]

FFT's are on a par with the Emperor's clothes. Maxim has some and I have never found it necessary to use any of them. I think you should really try to like Maxim because it will grow on you and it does everything you need, even post-processing if you are up against the wall.

All the odd bits of software that control guiding, focus, image capture, calibration and combining and so on are contained in the one package called Maxim and you don't usually need to update it. It won't clash with itself either. One purchase will do you for a lifetime.

Someone on here found a way to buy Photoshop via e-bay or somewhere and they paid about £110 for a fully working, not stolen, registerable copy. Once again you omly need PS CS for 16 bit, there is hardly anything in the later versions worth buying.

Incidentally, a 5x3 minute stack should give you the exposure equivalent of about 12m.

Dennis

[Helen]

I bought CS2 via Amazon for less than £100.

Helen

[MishMich]

I can't see it.

In sequence (unprocessed JPEGs cropped & scaled from TIFF result of RAW stacks in DSS).

1x3 min

3x1 min

2+1 min

2x2 min

4x1 min

The 4x1 & 3x1 stacks appear much the same, they and the 2+1 min are 'darker' than both the 1x3 min and 2x2 min stack - although these appear to show more of the light pollution. The 'brightest' image is the single 3 minute (stacked with associated darks), and while the 2x2 stack is not as polluted, and appears to show more detail in the trapezium, is darker and shows less nebulosity.

Confused.

M.

[roundycat]

I can't help you here because I have never used DSS and most likely never will. Those that like it say there are a lot of controls/settings in the background that you need to be au fait with. It is easy to muck it up apparently. I know I could repeat what you have done and do the combining in Maxim and get virtually identical results with all five examples.

Dennis

[themos]

MishMich, the default brightness in JPEGs is not a measure of how much noise there is in the underlying 16bit image. To do a proper comparison you need to look at a small crop that contains no bright stars (make sure there are no saturated pixels) and calculate the mean level and the variance in the 16 bit images.

[MishMich]

I don't think you can post TIFFs, which is why I posted JPEGs. I tried adjusting the levels on the originals, and when re-balanced, the 3x1 min does appear about the same brightness as the 1x3 min (although it is subjective, as each image has to be adjusted slightly differently). The 3x1 min appears sharper and less noisy. I then ran a 7x1 min stack, and compared it with a (2x2)+(1x3) min stack, and adjusted the levels, and they do appear very similar - but I was quite surprised how little the detail increased above the others. It 'appears' that where taking more exposures, there are diminishing returns. Is there some kind of 'rule' about at what point the optimum length/number of exposures is reached?

M.

[themos]

I think you can attach 16-bit PNG and TIFF files.

Don't be confused with how bright images look, they can always be scaled up or down. What matters is the noise level relative to the average. IRIS has tools that let you compute these quantities. You can do an experiment with "bias" images. These are the very shortest exposures you can shoot with your camera. It's quite easy to shoot 100 of them and compare a single frame to a stack of 100. Have a look at

Interpret CCD Read Noise Frame FFTs

[Psychobilly]

Entropy Weighted Average (High Dynamic Range) Stacking in DSS

This method is based on the work of German, Jenkin and Lesperance (see Entropy-Based image merging - 2005) and is used to stack the picture while keeping for each pixel the best dynamic.

It is particularly useful when stacking pictures taken with different exposure times and ISO speeds, and it creates an averaged picture with the best possible dynamic. To put it simply it avoids burning galaxies and nebula centers.

Note: this method is very CPU and memory intensive.

if all else fails...

[MishMich]

Is there a manual for Iris anywhere?

From what I can make out, to load a sequence of RAW files, you have to load each file one by one. That is odd. I have tried selecting multiples - nothing happens. I have tried opening 'select files' option, but dragging a series of RAW files doesn't work. Decode RAW files is the same - I select a group of files, drag them, across, and nothing happens. I can drag JPEGs, but they aren't much use. Converting RAW to TIFF, BMP or PNG wouldn't be of much benefit, as that increases the messing about time. It is hard to see how you can hope to do much with a piece of a software if you can only open one file at a time.

M.

[callump]

I think first you need to press the Digital Camera button on the button bar, and select your model of digital camera, then it knows about the type of raw file format to expect.

On the Digital Photo menu, you should get a dialogue when you Decode Raw Files... that lets you drag and drop your raw image files from explorer into it. I think you can only drop files of expected format onto the window.

There are a bunch of tutorial pages on the IRIS website (scroll down from the download section), and I also like Jim Solomon's Cookbook, at:

Jim Solomon's Astrophotography Cookbook

HTH

/callump

[callump]

btw - i have used IRIS's fourier transform, and also LR deconvolution on some poor images (focus) i had, and they did work quite well.

/callump

[themos]

I've given you a link to IRIS documentation above.

[MishMich]

I found out the problem with loading RAW files, it defaulted to EOS350/Rebel - and having a 1000D I needed to set it to the 450D series (1000D is not listed in either).

I think the problem with Iris is that it is a bit like UNIX, counter-intuitive. You have to know exactly what you want to do and how to use it before you can use it. Not ideal for learning on. I used and administered all the main UNIX OS and RDBMS systems (Sybase, Oracle, Informix, HP-UX, SVR4, SCO, Solaris, Ultrix+OSF, as well as VMS) over a ten year period up to nearly ten years ago, but I can't say I'm keen on that way of doing things.

I had a look, but couldn't find a FFT - although it does appear that Iris uses FFT in it's alignment & stacking process.

I think I'll have to try Iris a bit further, though. Thanks. Just wish there was a comprehensive guide (Jim's is mostly command-line, assumes you know what you are doing, and pretty incomprehensible). Without knowing what Iris wants, it is pretty difficult to know how to use it. You type in what you think it wants, only to find that is not what it wants, and what it wants isn't there, but it isn't clear how you get what it wants to enable it to do what you want it to.

DSS you just select the files, select the darks, and if you don't have something it carries on regardless. The rest is all in the options.

M.

[MishMich]

Thanks.

Jim's site is very helpful - and the 'tutorial' is starting to make more sense now I have managed to get the RAW files loaded. Like you say, though, it is complex. I guess I'll have to capture some flats as well as darks in future sessions.

It looks like it is possible to autoguide with Iris as well? That'll be handy.

M.

[callump]

You'll have to delve down into the processing commands which are run from the command box to find out how to do fft's etc. But all the commands are there.

There is a big learning curve, but once you get going it becomes pretty straigthforward.

The commands list is quite helpful, and has links to the tutorials, etc. too.

IRIS Commands

but as they say, no pain, no gain...

/callump

[themos]

I think the problem with Iris is that it is a bit like UNIX, counter-intuitive.

I think that is its strength. In the visual way of selecting options you are never quite sure what the software just did or whether you can replicate it exactly (and then try variations in an organised fashion). In IRIS you just save the complete sequence of commands in a text file. It's fair to say that DSS and PixInsight allow you to use batch files (I believe) but many people just rely on their memory for mastering which options are relevant. I am old enough to have trouble with that.

[MishMich]

I'm so old I have trouble learning/remembering commands. My brain just refused to accept any new command-line arguments about ten years ago. That was a bit of a problem, as a systems admin. One day it just decided it was a brain and not a computer, and refused to respond to UNIX commands any more. Gone are the days when I would dream SED and AWK, and cram obscure instructions into a single line so dense few could unfathom.

M.