Looking for an explanation for this...

[brianchapman]

These are 100% crops from two 15 minute exposures taken about 15 minutes apart (one in between).  They are both with the Canon 5D Mark 2, ISO 400, 15 minutes, f5.6, Canon 17-40L, same DPP recipe for processing, yet there is a dramatic difference in resolved detail.  The different color of light can be excluded because I have several others showing the same lack of sharpness under lighting more similar to the first (this one was used because the objects were of similar size and shot at a similar distance).  The image on the left was the first in a series of four long exposures and is clearly the sharpest.  The image on the right was the third.

I'm thinking there are four possible explanations:
1.  A minute change to focus (although I'm sure both were focused at infinity) really made a huge difference in resolved detail (if this is the case I've got to figure out a consistent way to get the focus exact in the dark!!)
2.  The first image is the only image that wasn't taken while the camera was working on a long exposure noise reduction for a different image in the background.  I don't think that should matter but I guess maybe it could.  
3.  The sensor was beginning to heat up after nearly an hour of exposure and a similar amount of dark frame subtraction going on in the background.
4.  Out of the four images the first is the only one where the tripod didn't shake at all (I have no way to prove this).

Anyway, what do you think?  I have not had a chance to test this yet but when I do I will post the results.  If I were guessing I'd say #1 or #4 or a combo of both...

[Panopeeper]

1.  A minute change to focus (although I'm sure both were focused at infinity) really made a huge difference in resolved detail

I gather you focused manually. Can't you turn the 17-40mm past infinity? All of my lenses can be turned past infinity; I guess this is so with all Canon lenses.

2.  The first image is the only image that wasn't taken while the camera was working on a long exposure noise reduction for a different image in the background

I don't understand this. Long exposure NR means, that a second shot will be taken with the same shutter time, but with closed shutter. You can not shoot during that time.

[brianchapman]

Thanks.  Yes you can turn the focus ring past infinity (the mark for infinity at 17mm is beyond the normal infinity mark as it is).  They also provide a "range" of infinity to compensate for different conditions but I have never dealt with it before.  So perhaps incorrect focus is the culprit here.  That wouldn't surprise me because with a 1.6 crop sensor (my old camera) it was really easy to get everything in nearly perfect focus...full frame is a different animal in that regard.

I don't understand this. Long exposure NR means, that a second shot will be taken with the same shutter time, but with closed shutter. You can not shoot during that time.

You are correct, however, both the 5D and the 5DII (and no other models to my knowledge) allow you to start another exposure while the dark frame is being taken for the original exposure (it's a fantastic feature imho).  I'm not sure how it works but it does and it's in the manual so I assume they don't think there is an issue with it affecting the dark frame or the next exposure.

So for the samples posted I started the first 15 minute image, when that ended I waited about a minute then started the second 15 minute exposure while the dark frame was still being collected, then the same for the third image, etc, until I ended with a 23 minute exposure four shots after the first.

[agavephoto]

Thanks.  Yes you can turn the focus ring past infinity (the mark for infinity at 17mm is beyond the normal infinity mark as it is).  They also provide a "range" of infinity to compensate for different conditions but I have never dealt with it before.  So perhaps incorrect focus is the culprit here.  That wouldn't surprise me because with a 1.6 crop sensor (my old camera) it was really easy to get everything in nearly perfect focus...full frame is a different animal in that regard.



You are correct, however, both the 5D and the 5DII (and no other models to my knowledge) allow you to start another exposure while the dark frame is being taken for the original exposure (it's a fantastic feature imho).  I'm not sure how it works but it does and it's in the manual so I assume they don't think there is an issue with it affecting the dark frame or the next exposure.

So for the samples posted I started the first 15 minute image, when that ended I waited about a minute then started the second 15 minute exposure while the dark frame was still being collected, then the same for the third image, etc, until I ended with a 23 minute exposure four shots after the first.

brian,

focus can shift due to temperature change: how much did the temperature drop while you were out? I'm not sure i can conclusively say that's the culprit, though. I have experienced this myself, which is why I bring it up here. In some of the astrophotography I do with telescopes, I utilize a temperature sensor that is tied to a small motor that adjusts focus to compensate for the focus shift due to temperature changes. One way to show this might be if by chance each image shows a slow degradation of focus/detail in the sequential order you took them in.

Also, how much was the camera moved between shots? Was it tilted between portrait and landscape mode at all . . . this may result in small shifting of an/the internal element/s of the lens (hopefully not too much!) that could result in this effect.
Just some quick thoughts: if I come up with something else i'll try to remember to post it.

Also, the 5D and 5D2 (assuming since I have not tested this part yet with my 5D2) will allow continued shooting after a long exposure. However, the dark frame reduction cannot happen at the same time. By definition, a dark frame is with the imaging sensor, and here it will be an exposure of the same length. I think both the 5D models just wait until you are finished exposing to finalize the dark frames needed for earlier shots if you interrupt it from taking them right after the exposure.

[dwdallam]

One of the images, the left, sems to be a larger crop of the trees, which will look more out of focus because you are zooming in more.

Also, the infinity mark is an approximation. I've experienced that if you turn the focus dial all the way over and then back it off just a tad to the left of the infinity mark, that is the sweet spot.

If you want to auto focus at night, bring a huge flashlight and aim as far away as you can, or at the subject you want in focus given that it is miles in the distance. Turn the light on, auto focus, or manual, shut it off, then start the exposure. I use a Cyclops 15 million candle power light. On a clear night, you can see it well enough to focus at 1/3 mile.
http://www.cyclopssolutions.com/HTML/spotlights.html

I got the first iteration from Costco for 30 bucks. The newer one is better for several reasons, one, the charge shuts off automatically. They use motocycle sealed lead acid batteries and yuo can charge them from your lighter outlet, or use them directly plugged into it also.

[brianchapman]

Also, how much was the camera moved between shots? Was it tilted between portrait and landscape mode at all . . . this may result in small shifting of an/the internal element/s of the lens (hopefully not too much!) that could result in this effect.

I used a level so it should have been nearly the same (all in portrait mode as well).

Also, the 5D and 5D2 (assuming since I have not tested this part yet with my 5D2) will allow continued shooting after a long exposure. However, the dark frame reduction cannot happen at the same time. By definition, a dark frame is with the imaging sensor, and here it will be an exposure of the same length. I think both the 5D models just wait until you are finished exposing to finalize the dark frames needed for earlier shots if you interrupt it from taking them right after the exposure.

I suspect you're right but the manual says "...after the picture is taken, the noise reduction process may take the same amount of time as the exposure.  During noise reduction, shooting is still possible as long as the maximum burst indicator in the viewfinder shows 1 or higher" which makes me think it might be trying to do both (or something different) at the same time.  This is consistent with the fact that after my last exposure the camera was finished processing everything 25 minutes later which means to me that either it took the dark frames for the three 15 minute exposures at the 15 minute time mark and then took the 23 minute one at the 23 minute mark (meaning it captured all the dark frames in one "exposure") or it was able to do multiple things at once (which to your point means it wasn't doing true a dark frame capture).  

In any case, that's why I posted that as a potential reason because I'm not sure exactly how it works!

[brianchapman]

One of the images, the left, sems to be a larger crop of the trees, which will look more out of focus because you are zooming in more.

They are actually pretty close in distance and zoom level - it's a 2mm difference in the lens.  But I still don't think that should make as big of a difference as the images show.  So I'm still leaning toward a focusing error...

If you want to auto focus at night, bring a huge flashlight and aim as far away as you can, or at the subject you want in focus given that it is miles in the distance. Turn the light on, auto focus, or manual, shut it off, then start the exposure. I use a Cyclops 15 million candle power light. On a clear night, you can see it well enough to focus at 1/3 mile.

Thanks, I'm looking at these today - it'd be nice to have something to check focus (either with autofocus or with live view).  How is your battery life with that?  And how often do you charge it/top it off?

[dwdallam]

They are actually pretty close in distance and zoom level - it's a 2mm difference in the lens.  But I still don't think that should make as big of a difference as the images show.  So I'm still leaning toward a focusing error...



Thanks, I'm looking at these today - it'd be nice to have something to check focus (either with autofocus or with live view).  How is your battery life with that?  And how often do you charge it/top it off?

Why to the trees look so much bigger in the right image then?

Batteries last forever in those lights. I think you get 40+ minutes continuous at full power, but you never just leave it on. They also have a high and low setting.

[BernardLanguillier]

Anyway, what do you think?  I have not had a chance to test this yet but when I do I will post the results.  If I were guessing I'd say #1 or #4 or a combo of both...

My bet would be wind induced camera shake.

Another possibility is that the light level dropped between the exposures (or the color temperature changed), which results in more noise being present in the image. DPP would be trying to compensate this with more noise reduction eating up detail? Have you tried with a different raw converter?

Cheers,
Bernard

[brianchapman]

Why to the trees look so much bigger in the right image then?

Batteries last forever in those lights. I think you get 40+ minutes continuous at full power, but you never just leave it on. They also have a high and low setting.

I think the trees look much bigger because of the rocks - they create an illusion of size when it really isn't there.  I looked again at the full size versions and the trees are really very close.  

Thanks for the light info - I'm definitely going to get one.  I was trying out live view focusing tonight in the SNOW!!! in Seattle and it helps quite a bit if there is enough light...so that 3/4 of a mile light beam won't hurt :-P

[brianchapman]

My bet would be wind induced camera shake.

Another possibility is that the light level dropped between the exposures (or the color temperature changed), which results in more noise being present in the image. DPP would be trying to compensate this with more noise reduction eating up detail? Have you tried with a different raw converter?

Bernard,

I agree with the camera shake idea - it's just odd that the first shot of the night is significantly better (sharpness) than the rest.  I've tried this in ACR, Raw Therapee, and DPP, all produce very similar results (which also supports the camera shake or focus error theory.)

Brian

[dwdallam]

Bernard,

I agree with the camera shake idea - it's just odd that the first shot of the night is significantly better (sharpness) than the rest.  I've tried this in ACR, Raw Therapee, and DPP, all produce very similar results (which also supports the camera shake or focus error theory.)

Brian

Can you post a crop of the exact same area Brian, say the snag tree (dead)  in the left photo?

[brianchapman]

Can you post a crop of the exact same area Brian, say the snag tree (dead)  in the left photo?

The photo in the original sample on the right was actually pointing the other direction so I don't have the same crop (or I would have posted it for sure!)  I do have one that is nearly the same but the exposure was longer (23 vs 15 minutes) and the focal length was shorter (28 vs 36mm) which I have posted side by side below.  Initially I thought this one was as "bad" as the other but on second look I think both of these photos are of similar sharpness considering the longer exposure and particularly the shorter focal length.  There is definitely less separation in the trees and less detail but considering the different focal length I would think it's not unexpected.  I originally posted the sample I did because the exposure was the same and the focal length was similar.



After looking at this comparison I'm now inclined to think that it really is a focus issue or camera shake...

On another note, I tried using live view to tune focus last night and as long as there is some type of detail (which you could definitely use that spotlight to enhance) it's really easy to check and tune the focus.  Another spotlight question - do you have a smaller one?  Like 2 million candle power?  I'm wondering how they compare...the size of the larger ones is considerable so I'm not sure I'd want to lug it around.

[dwdallam]

The photo in the original sample on the right was actually pointing the other direction so I don't have the same crop (or I would have posted it for sure!)  I do have one that is nearly the same but the exposure was longer (23 vs 15 minutes) and the focal length was shorter (28 vs 36mm) which I have posted side by side below.  Initially I thought this one was as "bad" as the other but on second look I think both of these photos are of similar sharpness considering the longer exposure and particularly the shorter focal length.  There is definitely less separation in the trees and less detail but considering the different focal length I would think it's not unexpected.  I originally posted the sample I did because the exposure was the same and the focal length was similar.



After looking at this comparison I'm now inclined to think that it really is a focus issue or camera shake...

On another note, I tried using live view to tune focus last night and as long as there is some type of detail (which you could definitely use that spotlight to enhance) it's really easy to check and tune the focus.  Another spotlight question - do you have a smaller one?  Like 2 million candle power?  I'm wondering how they compare...the size of the larger ones is considerable so I'm not sure I'd want to lug it around.

Naw I just have a small mag-light I carry in my camera case for short distance low light focusing. But you don't need to throw light all the way to the subject, but only 1/3 or less of the distance, or as far as you can focus using manual focus and the "near" infinity mark.

[Rob Reiter]

With my 5D I CAN take another picture while the camera is processing the first for NR. You just cant review it on the LCD until the first one is finished (and the second one is processed.)

I don't understand this. Long exposure NR means, that a second shot will be taken with the same shutter time, but with closed shutter. You can not shoot during that time.