New(ish) to astro imaging

[broadsword]

As the title suggests, I've been attempting inaging for around 6 months, but in reality the awful summer and North West skies has limited me to 5 or 6 intensive nights' activity. However, there has been plenty of time ffor practicing processing!

As I'm starting with modest equipment (large aperture but slow scope, low-end unmodded DSLR, no autoguiding, 8 bit processing software (GIMP)) I'm not expecting miracles but I want to know if I'm getting as much as I could expect to. I've compared my stuff with googled dslr images and some doesn't look too bad, particularly open and globular clusters, but I'm struggling with galaxies possible due to local light pollution (more of which later).

I've got quite a few pics to ask for feedback on, but to begin with my first question relates to open clusters. I wasn't sure whether it's necessary to produce multi-image stacks or just use single exposures. Below are examples of both for M103: a single frame (120s, ISO800), tidied up in the basic EOS sofware package and then a bit in GIMP, and then a DSS stack of 10 or so images (same exposure with darks and bias frames) also processed in GIMP. Both seem to have their merits (smaller stars for single frame, more background stars for the stack) but was wondering what the norm is for OC's? I may have overdone the colours a bit.

Any other comments much appreciated. If all goes well I'll post some different questions...

[Atlas]

I think your images came out fantastic! I'm short on knowledge when it comes to DSO imaging... but I would say your spot on!

[ollypenrice]

You are in focus there and getting fairly round stars without guiding. Very good going.

To progress;

Take multiple subs. These totally transform your signal to noise ratio. Not for nothing have I just taken 12X15 minute subs per colour filter to get our next image to about the half way stage...

Don't remove LP by black clipping the image. Your skies are jet black because they have been clipped, so you have also clipped real signal and on galaxies or nebulae this would be throwing the baby out with the bathwater.

Check that the top left of the histogram peak lines up in all three colour channels. The image on the right has too much green, I think, so clipping that realtive to the others would help.

Take flats. Even clipped your images show vignetting, a brightening towards the centre. Flats offer another way of getting more faint signal to appear on galaxy and nebular images.

Lots to learn but you are well on the way.

Olly

[broadsword]

Thanks for the kind words Atlas, and for the great tips Olly.

To your point on black clipping Olly, as I'll expand in in a future thread, I seem to have horrendous LP and with galaxies I'm sure I'm losing loads of detail, as you say. I'll try and follow your tip on aligning the colour histograms, but what exactly do you mean by the top left of the histogram peak (any chance of a rough pic)? I think I have been tending more to align the peak's right hand slope.....

I know that flats are a hill I will have to climb. What puts me off is that many of the methods to take them seem to involve taking images during the day, or taking the scope indoors to shoot a light box. Given that the camera orientation and focus cannot be touched between taking the lights and the flats, that effectively means imaging only one object per session! I guess I'm being greedy with those rare dark skies, but I try and get 3 or 4 DSO's in!!

[broadsword]

Looking at the posted images again, the colour isn't quite as I remember! I had to do a kind of double jpeg conversion to get the file size down to allow uploading, not sure if that hasn't changed them a bit. Anyone know what the best file format is to allow a decent image quality upload?

Just to check Olly, although I'd always use stacked multiple images for nebuale and galaxies, do you think its's the norm for open clusters as well?

Sorry for the persistent questioning, very keen to learn...

[ollypenrice]

Noise is noise so, yes, always use multiple stacks. A digital image is made up of a series of measurements. One measurement is never as accurate as the average of many similar measurements so the image will become more and more like the truth as you stack. It flattens out around 40 images. To approach the resolution of your system you need to combine multiple images. Besides, you never know what you might find if, like Fabian Neyer, you spend twenty hours instead of the usual 90 minutes on something like the Double Cluster...

http://www.starpointing.com/ccd/ngc869_ngc884.html

Here are the RGB combined histograms at the top in Ps and the three colour channels below that. I've put a pointer on the red one to show the top left which should align in all three channels. To quote Dennis Isaacs, though, remember that it isn't the histogram which you hang on the wall. This is just a good rough estimating method. Refine to taste.

Olly

[broadsword]

Excellent, that's very clear now, many thanks Olly.

[broadsword]

Sorry Olly, another question - when you say to avoid black clipping when removing LP, how would I have caused it? Would it be from levels or curves stretching?

[ollypenrice]

Sorry Olly, another question - when you say to avoid black clipping when removing LP, how would I have caused it? Would it be from levels or curves stretching?

From Levels, when you move the black point slider to the right after each iteration of Curves. At least I'm assuming this is so. I suppose it might just be the single short sub but it doesn't look noisy so much as clipped to my eye.

The first task in losing LP is to get the colour balance right but I really don't know about dealing with LP because I don't have any to speak of. Just a question on the imaging processing tips and techniques board, maybe?

Olly

Olly

[broadsword]

Hi Olly,

good call, I feel confident enough now to move to the Image Processing board.

ThanksG

[ollypenrice]

Hi Olly,

good call, I feel confident enough now to move to the Image Processing board.

ThanksG

Here's a 'visual aid' I produced for helping in tutorials;

Olly

[broadsword]

Awesome Olly, really awesome. So to avoid black clipping, is it genreally a case of not moving the black (left hand) slider in Levels all the way over to the start of the Histogram?

I sometimes get some wierd extra 'mounds' in the historam to the left of the main peak, all the way down to zero...never sure what to do about these. If I knew how to, I'd attach an example!

[poogle]

When stacking in DSS it adds a one pixel "frame" that is not completely black around the image, which might give that effect in the histogram (to the left of the main peak). Try cropping the image to remove those pixels and then have a look to see if the histogram changes.

/Patrik

[broadsword]

Thanks Patrik, this looks worht a try.

[ollypenrice]

Awesome Olly, really awesome. So to avoid black clipping, is it genreally a case of not moving the black (left hand) slider in Levels all the way over to the start of the Histogram?

I sometimes get some wierd extra 'mounds' in the historam to the left of the main peak, all the way down to zero...never sure what to do about these. If I knew how to, I'd attach an example!

Question one, yes, exactly that. Don't be tempted to cut back early in order to get a a nice looking picture in the early stages of processing. You need lots of room to manoeuvre so leave a flat strip on the left of the peak until literally the last operation in processing. Trust me, this is good advice. It took me three years to learn it!!

Question two, the mounds on the left; That's normal and the mounds are from edge of field artefacts. You could crop the image to avoid them but I prefer not to (a tip picked up a few years ago from my fine friend Mr Tom O'Donoghue.) All you do is use the Marquee selection tool to make a rectilinear selection just within the boundary of the image and process that. With this slightly reduced selection you'll get a clean and meaningful histo when you look at levels. No mounds!! However, at any point you can go back to the linear stack and do a different stretch for certain parts of the image and, because you haven't done any cropping, it will fit onto the one you have as a Layer for selective application if you wish. This is way better than trying to mix cropped and uncropped images...

Olly

[poogle]

Thanks alot for that tip Olly!

I went from this: http://www.tware.se/images/andromeda3.png

to this: http://www.tware.se/images/andromeda4.png

after your tip about leaving room in the histogram

I'm amazed at the difference it made. I have previously just made one radical change in curves and then the levels. Iteration seems to be the key

/Patrik

[ollypenrice]

Thanks alot for that tip Olly!

I went from this: http://www.tware.se/.../andromeda3.png

to this: http://www.tware.se/.../andromeda4.png

after your tip about leaving room in the histogram

I'm amazed at the difference it made. I have previously just made one radical change in curves and then the levels. Iteration seems to be the key

/Patrik

At a rough guess an image of mine has had about five hundred iterations of this or that so, yes, iterate like crazy!

Olly

[broadsword]

Hi Olly, these are great looking tips. I can see I was a bit mislead by a couple of tutorial videos on YouTube which I saw early on - they always started with bringing the LH slider up to the histogram in levels.

A lot of the terminology leaves me a bit baffled as a relative beginner. Do you have any recommendedguides (articles, websites, videos etc) covering a step up from the basics in image processing? Or is it a case of looking at brochures for SE France!