Too much data

[Robny]

A question that has been bugging me for a while, and that is.

Can you have to much data in an image?

Meaning, do you get to a point on certain subjects where gathering more data will not gain ay extra detail, is there a cut off point, do you hit a wall at some point. Or is it constantly the more the better?

I've seen some GREAT images at 10hrs, 16hrs etc, sometimes more some times less, but lets say if time, weather, commitments and will permits would the images be so uch better if for the 16hr image you carried on and got 32hrs or even 40hrs?

Thanks

Rob

Sent from my GT-I9505 using Tapatalk

[todd8137]

The more the better

Pat

[Naemeth]

I don't have much experience of imaging, but from what I've seen and read I've learned that you can't really get too much data. I haven't seen many images with more than 3 days worth though (72 hours), but perhaps that because the people who have gathered that data want to try a different target.

HTH

[ollypenrice]

You can't have too much but you can have more than is really going to make much difference. Opinion varies and the question needs to be considered in two ways.

If you are just dealing in the number of subs then the statistics suggest that somewhere around forty (some say less) is about the upper useful limit.

More useful, if sky and guiding allow, would be longer subs. These will take you deeper. When I went after the faint outer halo of M31 I found half a dozen 30 minute subs were more useful than many hours of 10 and 15 minute ones.

It also depends on your target. Faint data just does take time. Dusty features are punishing of inadequate exposure time. Emmission targets benefit from narrowband input, some of which can be very slow. But there would be no point in spending twenty hours on the Double Cluster in RGB. Spending 20 hours on the Ha would be a different matter, as Fabian Neyer demonstrated. 

http://spacefellowship.com/news/art36932/double-cluster-in-perseus.html

Olly

http://ollypenrice.smugmug.com/Other/Best-of-Les-Granges/22435624_WLMPTM#!i=2266922474&k=Sc3kgzc

[Steve 1962]

How does the saying go...?

Too much of a good thing...can be absolutely wonderful!

[michael.h.f.wilkinson]

In theory you can reach a point when the signal to noise ratio is so good that quantization errors due to the numbers of bits per pixel in your image start to dominate. I am not sure whether it is possible to reach that point in practice.

Another theoretical limit would be that the proper motions of stars start to become visible over the integration time. Again, I doubt any of us will live to gather enough detail to get to that point

[Robny]

Tb be honest, that is kind of what I thought, it will be a better image figures wise and a lot less noisy with fainter details showing?

But... depending what you would do with your images, either view them on screen or print them, then the output interface you choose (screen or print) has its own resolution/detail restrictions so you would not gain any benefit from going above the maximum guidelines of your  interface? 

Although I see the benefit for on screen display as you can zoom into the image with little loss of image quality to take a closer look.

Not trying to be controversial, just trying to get my own thoughts clear and understand it better, although I suppose it also comes down to personal satisfaction knowing that it can be better and wanting to improve on what you have, which is the category I think I would fall under.

I suppose the reason I ask is because I ask myself a similar question with terrestrial Photography, cameras keep gaining in MP's but does it get to a point where you cannot notice the difference between 32mp and 45mp? Or is it just numbers, just something that has had me wondering for a while

Thanks again

[Robny]

In theory you can reach a point when the signal to noise ratio is so good that quantization errors due to the numbers of bits per pixel in your image start to dominate. I am not sure whether it is possible to reach that point in practice.

Another theoretical limit would be that the proper motions of stars start to become visible over the integration time. Again, I doubt any of us will live to gather enough detail to get to that point

But we wwould be talking ALOT of data and hours before that occurs I guess

In which case - the more the better

   

[michael.h.f.wilkinson]

But we wwould be talking ALOT of data and hours before that occurs I guess

In which case - the more the better

   

Years, or decades, more likely

[Kirscovitch]

Interesting post. Has any imagers here taken years of data? 

Whats the most that any of you know about?

[Ptarmigan]

Another theoretical limit would be that the proper motions of stars start to become visible over the integration time. Again, I doubt any of us will live to gather enough detail to get to that point

!! Ah, that reminds me, someone (on this forum?) has been imaging Barnard's star (? or one of the run-away stars ) since a long time ago and posted a composite image ! , , now how to find that post again , , ,

Edit

Here we go :

http://stargazerslounge.com/topic/155332-barnards-star-re-visited/#entry1569162

and http://www.eagleseye.me.uk/Sky/Wordpress/wp-content/uploads/2012/07/BarnardsStarNegative2.jpg

My memory was a bit faulty, it is a partial (annotated for 1960) composite.

[dph1nm]

The improvement in s/n goes up as the square-root of the total exposure time. So although  more exposure will always improve things, it is a bit of a law of diminishing returns.

NigelM

[Robny]

Interesting, so in layman's terms (simple - me) S/N will always keep improving but the more data you add the improvements will start to slow down, not be as noticeable but will still be there?

That is sounding familiar, maybe from a book I read on the subject or an internet article.

Sent from my GT-I9505 using Tapatalk

[glappkaeft]

Interesting post. Has any imagers here taken years of data? 

Whats the most that any of you know about?

This group effort has 125 hours of exposure time - http://www.astrobin.com/55058/

For my own imaging the question I ask is not if longer is better but if one 32 hour exposure is better than two 16 hour or four 8 hour or eight 4 hour ....

[Zakalwe]

This group effort has 125 hours of exposure time - http://www.astrobin.com/55058/

For my own imaging the question I ask is not if longer is better but if one 32 hour exposure is better than two 16 hour or four 8 hour or eight 4 hour ....

The signal-to-noise ratio will be the same. The brightness of the fainter stuff will be greater in the long exposure

[glappkaeft]

The signal-to-noise ratio will be the same. The brightness of the fainter stuff will be greater in the long exposure

I'm not talking about sub lengths but if one image of a particular target with an total exposure length of 32 hours is more satisfying than four images of different objects each with an exposure time of 8 hours etc..

[frugal]

I'm not talking about sub lengths but if one image of a particular target with an total exposure length of 32 hours is more satisfying than four images of different objects each with an exposure time of 8 hours etc..

Only until the point where you saturate pixels, at which point you start to lose data due to white clipping.

[glappkaeft]

I'm not an native English speaker so I might have phrased something wrong but how could that have been misunderstood? I'm talking about when the improvement you get from spending more time imaging (a limited resource due to work, sleep, energy, daytime, weather, the moon, etc) one target for a long time (say one image of M31 - many subs - total exposure time 32 hours) is no longer worth it and you'd be happier with spending less time imaging a object and end up with more images in total (say M31, M45, M13, M42, each many subs, each target gets 8 hours) using the same total amount of exposure.

Saying that more exposure time is always better is true but not very interesting. The limiting factor is your imaging time and the number (OR quality) of images you want to take. My preference after seven years of observing and imaging have turned into wanting the occasional long exposure DSO image (one or two 10 hours exposures using the club RC every year), a bunch of 1-4 hours exposures of DSOs (using my own equipment) and short (minuters and even seconds) exposures of interesting targets (NLC, aurora, Iridium flares, light pillars, and similar). The rest of the time I spend observing visually or doing photometry.

[ollypenrice]

I'm only happy when I've nailed an image, got it as good as I can get it, or close. I share resources with two other imagers who feel exactly the same (well, one of them is actually worse than me for this!!  ) so we get along fine. But it's quite natural that those who come here for a week don't want to do just one image and that's perfectly fine as well. In fact it's a good discipline for me because my dog headed approach is usually just to spend more time at the capture stage rather than exploiting the post processing to best effect.

My own record was a collaborative M31 which has now topped 50 hours. Very nice data to work on...

Olly

[Earl]

Your environment dictates the answer for you.

If LP washes out your subs at 300s then you have to work within that parameter.

If you can get 30 min subs and your gear is also happy doing this then the more will overall be better.

Local conditions are the biggest factor.