How can I rate a sub?

[James]

I've been accumulating 5 minute subs over the last few weeks and now that I have a pile of them it occurs to me that some will be better than others. For example I have 10-20 or so Ha subs taken at full moon (with the moon only about 30 degrees away), Whilst they haven't turned out too bad it crossed my mind that the ones coming out of my H9 right now are probably better.

Is there any way to evaluate subs for quality, for example amount of noise? I'm using Nebulosity and PixInsight in case there's a menu command I've missed!).

James

[harry page]

Hi

Under scripts in pixinsight there is a evaluate noise process

And if you use pixinsight to stack there is a option to measure the noise in each image and give it the appropriate weight when stacking

This is most excellent

Harry

[James]

I've started using the noise evaluation during stacking - we'll see how it turns out... I'll give the noise evaluation script a go as well, thanks Harry

James

[BlueAstra]

I have what I think may be an associated question. How do you judge the exposure time required, in a heavy LP environment, by looking at different time exposures of the same object. I've enclosed 3 subs taken recently of M45 at 32s, 60s, 210s, with a LPF in place. Is there a quality measure I can use to judge which (if any) is the correct exposure length to use?

(equipment: Meade 5000 ED80, Telvue 0.8 FF, Hutech LPF, unmodded Canon 1000D.

[dph1nm]

So I can think of three criteria:

(1) for faint objects e.g. nebulae, galaxies, expose until the noise from the sky background is >> read noise in all three colour channels (this is obviously easy to manage in R, but maybe not so in .

(2) do not exposure so long that your object of interest is saturated (i.e. =255 in R,G or B in jpeg numbers). If you are interested in the colours of bright stars this means short exposures. For nebulae and galaxies you can let the sky go right up as in your third example.

(3) how many subs are you prepared to store and process on your computer!?

NigelM

[tooth_dr]

On my 300d DSLR (which is modified) there is only a histogram of the combined channels, but on my 400d there are separate RGB histograms. Is it possible that I could be clipping the red channel on the 300d or would it be evident?

Thanks

[Tim]

Good question James. Here are some general pointers for DSS.

I use DSS for stacking at the moment, and that rates the sky background, number of stars, FWHM, along with giving each image a score.

I go through each pic with the sliders used to bring out the fainter stuff, and check for obvious problems, trailed stars, clouds etc, and delete those from the disk.

after that I check the subs that are left as regards the items above. Some will stand out as having a vastly different sky background, or much less stars. I boot those out too.

From any five or six hours worth of data on any one target, I usually have aircraft effect on them. This can be a problem with 20 min subs Sigma stacking can get rid of them if you have enough subs, but I have found that I get the best results using the Auto Adaptive Weighted Average stacking method in DSS. this sometimes includes artifacts from aircraft, so those that cut across or through the centre of the target are booted out as well.

One other tip for DSS, is to make sure the reference frame is the best frame, which is not necessarily the one chosen by the program.

As the targets rise or sink in the sky, generally the quality will go up or down a little. If only we had the luxury of enough clear nights to only image the target when it is a few degrees either side of the meridian.......

Tim

[James]

Thanks for the info Tim. I can't use DSS but Nebulosity does have a rating scheme for subs and PixInsight has some options - I just need to give them a go and try them out.

Luckily for me Taunton doesn't get many overflights as they all go south of us (except for one day when 4 American F15's flew over the town for 4 hours which nearly caused a war...).

I still have a niggling flexure problem so I'm having to can about 50% of all 10 min subs - on the 'plus' side trailing is quite easy to spot...

James

[MartinB]

James, one problem with answering questions like this is that we are all using different software. It wouldn't be so bad if the developers made some attempt to name their tools in plain english rather than trying to dazzle us with their sophistication. Many programmes have very similar tools but they all call them something different. I would like to devise a background adaptive weighted dynamic fourier transforming bull dodos extractor and apply it to all astro software making them come up with sensible terms!

...rant over. Pretty much what TJ has said. I use CCD inspector and CCDstack. Whatever you use the key thing is to understand the different quality aspects of an image and sort out those images that should be dumped and those, which although being quite different, can be incorparated. In other words i) problem images ii) different images. Most of what I do is very simalar to the things Tim said.

Problem images -

Trailing - the measurement of this is sometimes termed the aspect ratio, how wide is it compared to how high. CCD inspector gives a very precise measure. This helps flag up problem images but normally I just blink through magnified images in CCD stack. (this software is really good and so easy to use)

Bloating - normally measured as the width of the star light profile at half maximum intensity, FWHM. The FWHM is affected by poor seeing and poor focus (often affecting later subs in an imaging session as the temperature drops). You can have images which are very good in every respect other than the focus being a tad off. Again, I blink through them in CCDstack and make my choice. Sometimes I actually try different stacks with a different threshold and in geeral I've found that including poor images leads to a poor over all result despite the extra data.

Background brightness - the term often used is contrast ratio. You are looking at how much your target signal stands out from the background. This is an issue when high cloud affects the image, condensation or close to dawn. You can have great stars but the sky is just too bright. This is a different problem to that of differences in background brightness due to different exposure times. If you have subs which are significantly brighter than others due to changes in conditions they are best dumped unless you are desperate.

Aircraft and satellite trails - keep them and they should come out with stacking.

Different images -

Typically differences in image scale, rotation and so on. Many programmes can handle this for you with different amounts of automation or user control.

If you have used different set ups on different nights with different exposure times you will have images of varying signal to noise ratio. These images can be combined effectively. The key thing is to make sure the images with poorer s/n ratio dont degrade the higher quality images. The technique used is usually called "normalisation". Although there are different ways of doing this the software needs to be able to measure the brightness levels of matched images. In CCD stack you register the images first and I then use an automatic normalisation tool but manually select the part of the image for CCD stack to use for calculating. It's best to have some background and a mid range area of brightness. CCD stack then works out a scaling factor for each image. As Tim says, you should always use your highest quality image as your reference image (although this sub can vary depending on which quality you need - good stars for registering, good s/n for normalisation).

I'm sorry I can't speak for nebulosity or pixinsight but if you are clear what you are wanting the software to do that should help you find and make good use of the appropriate tools and break through the ridiculous jargon.

[harry page]

Hi

yes I know what you mean about the contest to find the most nutty name for a process, and Pixinsight is a front runner for this

strangely they call their noise evaluation , noise evaluation not right at all:icon_scratch:

Harry

[James]

That's great info, thank you Martin (and Tim earlier). There's a few options available to me - I think I'll be going over about 120 subs with them and see what happens and I'll be strict and only keep the best ones!

James

[James]

OK, I've been through 46x 10 minutes subs I have in SII for an object I've been imaging and looking at trailing, bloating and sky brightness/object brightness I've managed to get rid of those subs I'm not happy with. I used Nebulosity to grade them all an there was one sub thats a clear 'winner', no trailing, bright sharp etc. However comparing all the others to this one has led me to quietly put aside 43 of the 46 subs

Trailing is the biggest culprit (there's a little bit of flexure still to be dealt with ) and a few of the subs were particularly 'dim' in comparison to the rest. Oh well, I just need some more clear nights now!

James