Night-time photography - image sharpness issues

[Manfred M]

I did a couple of night-time panos this week, and while I am generally satisfied with the results I achieved, the images are simply not as sharp as I would have hoped they would (viewed at 100%) and when compared to previous pano work. I thought I had "done everything right", so I am looking for some hints and suggestions on where I could improved.

Shots were taken with my D800 and ranged from around 2.5 sec to 4 sec exposures. The camera was on a large and very rigid (Feisol 3372 - three segments with the largest segment being 37mm / 1.5" diameter; rated capacity - 12kg / 26lb) tripod without a centre column using a RRS BH-55 ball head and Kirk long rail plate (a.k.a. nodal plate) and L-Bracket. I had about 6 kg (15 lb) of weight hanging down from the hook (my camera bag with other lenses and hardware). Shots were manually focused using the focus indicator and taken at f/4.5 at ISO 100 at the 24mm focal length (focused at ∞, things should have been sharp). I shot with a pro Nikkor f/2.8 24-70mm lens (no image stabilization), with autofocus disengaged. All shots were taken in "mirror up" mode and I waited about 4 sec between moving the mirror up and releasing the shutter using a cable release (and obviously not touching the camera or the tripod).

The surface the tripod was on was concrete with some hard packed snow (hard to avoid at this time of year). I couldn't feel any vibrations or other issues that could have affected the setup. The only thing that I was doing that was "out of spec" is that it was around -12°C / 10°F out and the camera is only rated down to 0°C / 32°F.

Any suggestions as to what else I might do to improve sharpness?

[Shadowman]

Have you tried focusing with the LCD with magnification or is your subject too dark?

[benm]

You say you manually focussed but also that you focussed at ∞. I assume you did the former and it just happened that the ∞ symbol lined up with the indicator line. Was there wind? Was the concrete on the ground or a bridge deck? Does it look like mis-focus or motion blur?

Could you post a photo? That would help.

[Manfred M]

Have you tried focusing with the LCD with magnification or is your subject too dark?

The D800 has a focus indicator that I find to be more accurate than the LCD magnification approach, but I could try that as well. Thanks.

[Manfred M]

You say you manually focussed but also that you focussed at ∞. I assume you did the former and it just happened that the ∞ symbol lined up with the indicator line. Was there wind? Was the concrete on the ground or a bridge deck? Does it look like mis-focus or motion blur?

Could you post a photo? That would help.

At 24mm on a full-frame body, anything near ∞ will be in focus, so no, I was not using the line, but the in-camera focus indicator which is very accurate when it picks up an edge. Wind, yes, there was some, but not enough to move things around even slightly; my setup is very rigid. The concrete was on the ground, so no issues with an unstable surface like a bridge.

I suspect that it is a combination of motion blur and misfocus. Just click on the images to see a larger view.

The full image is:




The 100% view is:

[PhotomanJohn]

Manfred - These are obviously guesses. Could there have been some slight fogging inside the lens or camera due to the extreme cold. The air in your house may have been moist enough to cause a little fogging when the camera got that cold. I doubt that differences in thermal expansion or thickening of the grease on the mirror mechanism could have caused any positions to change enough to be seen in the final images. It will be probably be difficult to determine the actual cause but good luck.

John

[Manfred M]

Manfred - These are obviously guesses. Could there have been some slight fogging inside the lens or camera due to the extreme cold. The air in your house may have been moist enough to cause a little fogging when the camera got that cold. I doubt that differences in thermal expansion or thickening of the grease on the mirror mechanism could have caused any positions to change enough to be seen in the final images. It will be probably be difficult to determine the actual cause but good luck.

John

Thanks John; I doubt it was fogging as I tempered the gear (it was outside for over an hour before these shots) and I did check for fogged surfaces with a flashlight. The lens itself responded remarkably well in the cold; I've done enough outdoor winter shooting with it and have had focused images in temperatures that were even lower than what I was shooting under here.

I'm looking at doing some more shooting tonight and am trying to figure out my shooting strategy to correct for this issue. The nice thing about the large pano is that one doesn't really see the softness until one looks at the base images at 100%.

[FrankMi]

I suspect that it is a combination of motion blur and misfocus.

Hi Manfred. Motion blur will usually show up in a pixel-peep as an elliptical but in looking at the lights on the trees, they are about as perfectly circular as you can get. That leaves miss-focus or something else as the likely culprit. I have found that in my D3100, manual focus is hit-or-miss. Focusing by the viewfinder isn't as accurate as you would hope as the mirror and viewfinder don't always rest at exactly the same plane as the sensor. Using Live-View with the mirror locked up may not be much better. Although you are now using the sensor directly, it is difficult, even with an eye loupe to see anywhere near as fine as the final image on a monitor so it is still a guessing game. The DoF for a small aperture should take care of the focus even if you are not dead-on infinity.

There are several tests you can do to try to isolate the source of the problem.

First I would test using identical setups in cold weather with the autofocus on and off to see if there is a difference.

Next, for night shots like this, compare using around a 20 second exposure as opposed to 2.5 to 4 seconds as 20 seconds is far more tolerant of even the slightest movement such as triggering the shutter (regardless of the trigger method used).

By going to 20 seconds you will also be closing down the aperture (assuming you don't raise the ISO) and this should also work in your favor.

Lastly, if possible, try to duplicate your setup in warmer weather. As John suggests, really cold temperatures may change the camera/lens behavior for a number of reasons, including uneven contraction of various metals and plastics and the viscosity of lubricants involved.

I would also closely examine the SOOC images to ensure that the issue is not creeping in during the pano and PP processing.

Hope this helps!

[Shadowman]

The D800 has a focus indicator that I find to be more accurate than the LCD magnification approach, but I could try that as well. Thanks.

I think the brilliance of the light and its direction is the leading cause. Areas not directly illuminated (the background buildings) by the decorative lights are reasonably sharp, the stairs are exceptionally sharp. Perhaps using a filter will tone down some of the brilliance of the lights.

[Glenn NK]

Were all three tripod legs on the concrete, or was there ice/snow between the point and the concrete?

[Manfred M]

Hi Manfred. Motion blur will usually show up in a pixel-peep as an elliptical but in looking at the lights on the trees, they are about as perfectly circular as you can get. That leaves miss-focus or something else as the likely culprit. I have found that in my D3100, manual focus is hit-or-miss. Focusing by the viewfinder isn't as accurate as you would hope as the mirror and viewfinder don't always rest at exactly the same plane as the sensor. Using Live-View with the mirror locked up may not be much better. Although you are now using the sensor directly, it is difficult, even with an eye loupe to see anywhere near as fine as the final image on a monitor so it is still a guessing game. The DoF for a small aperture should take care of the focus even if you are not dead-on infinity.

There are several tests you can do to try to isolate the source of the problem.

First I would test using identical setups in cold weather with the autofocus on and off to see if there is a difference.

Next, for night shots like this, compare using around a 20 second exposure as opposed to 2.5 to 4 seconds as 20 seconds is far more tolerant of even the slightest movement such as triggering the shutter (regardless of the trigger method used).

By going to 20 seconds you will also be closing down the aperture (assuming you don't raise the ISO) and this should also work in your favor.

Lastly, if possible, try to duplicate your setup in warmer weather. As John suggests, really cold temperatures may change the camera/lens behavior for a number of reasons, including uneven contraction of various metals and plastics and the viscosity of lubricants involved.

I would also closely examine the SOOC images to ensure that the issue is not creeping in during the pano and PP processing.

Hope this helps!

Thanks for the ideas Frank - The main reason I went for the relatively short shutter speed is to minimize the background noise in the dark areas of the image. I had to spend a fair bit of time squashing hot pixels and that will only get worse with longer shutter speeds. The other reason was that the sweet spot for the lens is around f/5.6, so I was trying to stay near that value.

The D800 has a significantly better focus mechanism than the non-pro models, like the D3100. There is an in-viewfinder display that indicates sharp focus, but as it is part of the AF mechanism, it does rely on edge detect / contrast. I've found it to be 100% accurate in previous testing, but strong suspect that the lighting might have thrown things off. Motion blur is exactly as you describe it if motion is unidirectional, but sinusoidal motion (set up by vibrations) can result in circular artifacts, so that is why I'm not 100% convinced that there isn't at least some component of motion happening here. There is some underground tunneling going on in the area for the LRT line, this is using a large boring machine and that might be a factor. A lot of the digging is going on 24/7, and rock transmits vibrations for a long distance.

Could the cold be a factor; perhaps but that is measured in microinches / inch / degree, which is far better than the manufacturing tolerances on the lens, so I would be rather surprised if that is a significant issue.

Regardless; I am going to try some of your suggestions to see where I get with them.

[Manfred M]

I think the brilliance of the light and its direction is the leading cause. Areas not directly illuminated (the background buildings) by the decorative lights are reasonably sharp, the stairs are exceptionally sharp. Perhaps using a filter will tone down some of the brilliance of the lights.

Not sure if a ND is going to do much as it will just slow down the overall level of light; not something one would normally want to do in night photography. Let me think about this one.

[Manfred M]

Were all three tripod legs on the concrete, or was there ice/snow between the point and the concrete?

Definitely both; two of the legs were solidly on concrete and the third was on hard packed snow. I should have ordered spikes for the tripod feet, but did not. That is something that I will be doing soon.

[Glenn NK]

Definitely both; two of the legs were solidly on concrete and the third was on hard packed snow. I should have ordered spikes for the tripod feet, but did not. That is something that I will be doing soon.

You know why I asked.

I use a Gitzo - they sell spikes but they are too short IMO, and hence useless. A 10 mm bolt with the head cut off and sharpened is cheaper and much better.

Is the bottom of the Feisol threaded with a female thread as is the Gitzo?

[Manfred M]

You know why I asked.

I use a Gitzo - they sell spikes but they are too short IMO, and hence useless. A 10 mm bolt with the head cut off and sharpened is cheaper and much better.

Is the bottom of the Feisol threaded with a female thread as is the Gitzo?

Yes it is. I have some Benro spikes and need to see if they fit first. Yes, I do know what you mean; not bad for lawns, but not good for any real amounts of snow.

[Adrian]

This is a long shot and directly refers to focus at infinity. One of the things I learned when shooting at low temperatures in the arctic, was to prep for night shots at infinity, by checking infinity focus on that lens in daylight. The difference between what one would expect from the lens scale, and what was really sharp at infinity, was slight but surprising. This was with Canon pro lenses. I find it hard to focus really accurately at night and so often try to set up in daylight and then copy the settings.

[Manfred M]

This is a long shot and directly refers to focus at infinity. One of the things I learned when shooting at low temperatures in the arctic, was to prep for night shots at infinity, by checking infinity focus on that lens in daylight. The difference between what one would expect from the lens scale, and what was really sharp at infinity, was slight but surprising. This was with Canon pro lenses. I find it hard to focus really accurately at night and so often try to set up in daylight and then copy the settings.

Did you mark your lens somehow so you could see this? Sounds interesting. I used a Petzl headlamp while setting up my shots so that I could see all the camera controls, but did find that fogged up my glasses on occasion, making this a tad more difficult to do.

[Colin Southern]

Hi Manfred,

Don't ever trust AF in situations like that; best way to achieve focus is to use live view on maximum magnification (10x in Canon world); adjust for smallest dot.

[Manfred M]

Hi Manfred,

Don't ever trust AF in situations like that; best way to achieve focus is to use live view on maximum magnification (10x in Canon world); adjust for smallest dot.

Thanks Colin - I will try that. I was looking to head out this evening, but changed my mind now that it has started to snow (again).

[DanK]

This is a bit puzzling. DOF is huge at that focal length, even when open to f/4.5. If the lens was actually focused anywhere close to infinity, it is hard to see how the buildings could be out of focus. From the first image, it looks like the very close areas are sharp, although it is hard to see without having that area blown up. If the close areas are sharp, then it is clearly a focusing problem. If there is no region, near to far, that is sharp, then it is not a focusing problem. So, my first step would be to blow the whole image up to 100% and then start with near detail and work back systematically.

Another possibility that occurred to me, if it is not a focusing problem, is that the weight actually hurt you--that the tripod leg on the hard packed snow actually sank a tiny bit while you were taking that part of the pano.

I do a little night photography, and I generally keep the aperture smaller because focus is so hard to manage in the dark.