Which exposure length is more favourable

[Wailin]

I am just wondering what other folks opinions are on exposure lengths for certain objects. For example, which results in the best quality image (if any), 12 300sec exposures or 6 600sec exposures? I took 24 300sec exposures on the Rosette nebula recently over 2 nights. Would I have gotten more detail had I done 12 600sec shots? My autoguiding seems to be rock solid and I have tested 600sec exposures which were perfectly fine with nice round stars. I would hate to have to discard 10min over 5min though!

[anthony]

I think you'd get a better signal to noise ratio on the longer subs - you're collecting more photons and so you can get more detail. This could be offset though by the camera you are using as a dedicated, cooled CCD will perform better in this respect than a DSLR.

If your skies are awash with light pollution the longer your subs are the more the background will intrude. The longer your sub the more chance of tracking error or aircraft lights spoiling the show.

Have a bit of fun and try it and see!

[IanL]

The simplistic answer is that for a given total exposure time, fewer, longer subs are better because you will get a better signal to noise ratio.

If you had the best dark skies possible, perfect tracking and you camera could stay non-linear you'd make one exposure for as long as possible. In the real world, you have to think about sensor linearity, which will place an upper limit on how long you can expose for,how long depends on the camera, sky and target brightness. in heavy light pollution there is a definite advantage to shooting more shorter subs rather than long ones. Here's a useful link where the author tries to determine the best combination of number and length of subs:

http://www.cloudynights.com/item.php?item_id=1622

Risk is another issue; whilst you can use various rejection techniques when stacking to eliminate satellite and aircraft trails from individual subs, if your tracking is marginal or other issues play a part, you'd be better taking shorter subs in the hope that some percentage of them are good, rather than longer and finding that you have few or none that are usable at the end of the night.

[Wailin]

Thanks guys, I guess it makes perfect sense when you think about it. I'll up the ante to 600sec exposures on the next session to see how it goes.

[dph1nm]

If you had the best dark skies possible, perfect tracking and you camera could stay non-linear you'd make one exposure for as long as possible.

If you did your image would probably be trashed by cosmic rays hits!

NigelM

[IanL]

If you did your image would probably be trashed by cosmic rays hits!

Shhh, there are people on here who have tried to convince me that cosmic rays are a government conspiracy and you couldn't possibly (indirectly) detect them using something as mundane as camera

[ollypenrice]

Always try to expose for as long as your skyglow will allow, though a very small number of targets may need short exposures to combine with long where there is something ultra bright in the picture. This does not happen often. I mean, basically, that apart from M42 it doesn't happen!!!

Olly

[laser_jock99]

Following on from what Olly said, it is always worth studying your image histograms as these give the best indication of whether parts of an image are irrecoverably over exposed and whether you might need exposures of different lengths to build up a balanced image.

In the screen shot below for instance, the histogram (top right) is 'flat topped' in the centre. This indicates to me that parts of the image (the stars & centre of the nebula) are overexposed in this case and will always be 'white'. However- this amount of exposure was necessary to reveal the faint parts of the nebula.

Super long exposures are not always possible even in the darkest of skies. There are physical limits due to natural 'sky glow' or air glow which has been reported as increased lately by some imagers during the current maximum of the solar cycle.

[IanL]

That isn't how you read a histogram. The histograms in many imaging applications are pretty small and hard to read, so I've mocked up and example here:

A histogram plots the Pixel ADU (brightness) against the number of pixels in the image that have a given ADU value.

By convention, the Pixel ADUs are plotted along the X (horizontal) axis, so in my example you can see that we have values ranging from 0 (black) to 65,355 (white) which you might find in a 16 bit monocrome image. The count of pixels for each ADU value is plotted on the Y (vertical) axis. For a 1,024 x 768 image there would be 786,432 pixels in total.

(N.B. In my simple example I've mocked up the curve using 10 data points at intervals between 0 and 65,535. In a real histogram there would be a data point at each ADU value, so 65,535 data points in total).

The curve is a fairly typical one that you would see for a deep-sky image. On the left hand side there are no data points with values of or near to zero. This is what you want, either dark current or (ideally) sky/glow or light pollution would mean every 'black' pixel has some value more than zero.

The next think you see is a big peak. The left half of the peak and a fair bit of the right half of the peak represents the background sky in the image. The majority of pixels in a DSO image are going to be dark sky and this is what the peak represents.

The right shoulder of the peak represents nebulosity, galaxy arms and the other faint objects in the image. This is where most of the interesting detail usually lies. Some of the detail may be close to the middle of the peak as well, which is what makes it hard to tease out the fine detail from the sky background.

Finally we have a long, flat tail running off towards the right hand of the graph. This is the stars in the image. If your image is over exposed and saturated, then you will see that the value at the right of the graph (65,535 in this case) is not zero; that means that some pixels have reached maximum exposure and you have lost some detail. You may need to zoom in on the histogram to check this end. If you see a peak at the right, then you really have over-exposed.

The flat top in the peak in the example above is not over-exposure, it just means the software hasn't properly scaled the histogram for some reason and the chart looks screwy. This is not unusual in capture software, and the best way to check histograms is to use a proper processing application that will usually have more options and a better/zoomable histogram plot.

If you are imaging in RGB, you may see three curves overlaid on one chart, one for each colour. This just separates the colours but means much the same thing.

[IanL]

Had a closer look at the histogram in the overexposed example above, and I can see that:

1. The top of the peak has been sliced off in the plot, either the software has automatically zoomed in on the lower part of the graph or the user has done it. That is actually quite useful in this case as the important information is the stuff close to the X-axis at the right of the histogram. If the whole of the peak was shown, the useful information would be plotted at too small a scale to be visible.

2. The red, green, blue and (I think) luminance channels are plotted on top of each other. You can just about see some of the R, G and B curves peeking out around the edges of the white L curve in places. Using a solid/area plot for a multi-channel histogram isn't that helpful, and a simple line plot like my mock-up is more use since you can see more of the underlying curves.

3. If you squint at the bottom right of the histogram, you can see the L (white) curve bumping along all the way to the right hand end. You can also see a small blue peak at the far right. This is what shows the over-exposure, and given the blue nature of the subject not surprising that there is a saturated blue peak (there are probably smaller red and green peaks at the right hidden behind the blue/luminance plots).

Hope that is useful information for you guys. As I said, the main problem with most caputure software seems to be the rather inflexible/tiny histogram plots that you get. Given it is probably the most critical piece of information for analysing your captured images and adjusting your exposure time/gain/iso settings, I wish that authors would spend a bit of effort developing this tool into something more comprehensive rather than treating it as an afterthought.

You can load your images into a separeate processing package, but it is a pain and not ideal to be firing up a big application like Photoshop or PixInsight on your imaging laptop in the middle of a run.

[Gina]

Very useful information there Ian I usually go by the ADU count of the brightest part of the image in Artemis Capture after taking a sub and keep that below 65000 but the histogram method may be getter. Next time I think I'll try transferring a sub image file to my main PC via TeamViewer and examine it in PhotoShop, though the histogrem view is quite small and I always wish it was bigger.

[sloz1664]

Hi Gina

can't you view your image in Fits Liberator. That has a far better Histogram.

cheers

Steve

[dph1nm]

As I said, the main problem with most caputure software seems to be the rather inflexible/tiny histogram plots that you get.

It can be even worse - my Canon 1000D only shows an 8-bit range for the histogram on the back of the camera even if you exposure in RAW (it's a 12-bit camera). And it is not the top 8-bits! So the histogram can show you as saturated when your RAW data are actually a factor 4 away (it 'loses' the top 2 bits of the range)! I have no reason to suspect that other Canon cameras are different.

NigelM

[Gina]

Hi Gina

can't you view your image in Fits Liberator. That has a far better Histogram.

cheers

Steve

Could do - hadn't thought of that - thank you

[valefor]

how are people exposing for so long? i live in an area where the street lights are turned off at midnight. got clear skies. and this is the photo i took

its a 120second shot with a canon 1000d. i used to take hour long shots with a film camera with no problems.

[astrosathya]

I guess I can answer that. Light pollution creates a lot more noise and hence the S/N ratio is far lower than from darker areas. Therefore, after a certain exposure length, say 5 minutes, you will not have any advantage from the longer subs because light pollution simply comes in the way. I have the kit below and use it as much as possible and have noticed that exposure length depends on the light pollution factor also.

Regards,