Venice

[bambleweeney]

Venice, chock a bloc with tourists so ... you take photos at night or before 6am in the morning when few people are around. Uploaded from Flickr now that Tinypic is dead and just as I was getting used to it

4am just a few pigeons





and one during the daytime



and one at midnight, little chance of people in this one ... it just looked nicer at night

[Tom Barry]

A nice series, for sure.

[billtils]

Nice story, good set of pics.

[DanK]

very nice series. The final one is a particularly good use of night-time lighting. I might darken the right side and brighten the left a tad.

[bambleweeney]

Thanks Dan, I might give that a try

[Manfred M]

+1 to Dan's thoughts. A very nice series and does bring back memories of my visit to Venice.

As is typical of night time shots, the light distribution in the scene can be quite interesting and fairly stark. I find I do a lot more dodging and burning in my nighttime shots than I do with images taken at other times of day. Getting the tonal range closer together would take these very nice shots and make them even nicer.

I found that opening up the shadow detail can introduce more noticeable noise, so noise reduction may enter into this step as well.

[DanK]

I found that opening up the shadow detail can introduce more noticeable noise, so noise reduction may enter into this step as well.

Indeed. This leads to one of the suggestions I make to people starting night photography, which is to expose quite far to the right. It seems counterintuitive, as it leads to thumbnails that often don't look like night time, but you are better off darkening in post than lightening.

Another difficulty with urban night photography is that the artificial light often creates scenes that have too much dynamic range. If one exposes the shadows sufficiently, the lights blow out. I don't think there is one answer to this. Sometimes it's fine (in terms of the photographer's preferences) to let the lights blow out; sometimes it's better to bracket; and sometimes it's best to combine both, letting some areas blow out but less than one would have without bracketing. When in doubt, I bracket, usually with a total range of 4 stops. I end up throwing most of the extra shots out, but other than the time it takes to sort and cull, that isn't costly.

For example, in this shot, which I have posted before, I let the lights blow out and used a small aperture to create starbursts from them:



I hope you keep doing some night photography. I find it fascinating, and it would be fun to have more of it to look at.

[pschlute]

I like the daytime shot. Subjects are a little soft but I don’t think that detracts.

[billtils]

I have seen results from another way to approach night time shots and deal with the dark area by firing off a burst all at the same optimum exposure (not ETTR) and process as a stack. The noise was reduced to where it was not an issue.

[DanK]

I have seen results from another way to approach night time shots and deal with the dark area by firing off a burst all at the same optimum exposure (not ETTR) and process as a stack. The noise was reduced to where it was not an issue.

Interesting. This is done a lot in astrophotography, where shutter speeds have to be short if you don’t want trails. Is there an advantage to doing this instead of a longer exposure in urban night photography?


Sent from my iPad using Tapatalk

[Shadowman]

Nice series.

[billtils]

... Is there an advantage to doing this instead of a longer exposure in urban night photography?
Sent from my iPad using Tapatalk

Dan, I think that the intent is to give acceptable shadow detail but not at the cost of losing highlight detail. However, it could be argued that if you are shooting from a tripod why not bracket the exposure values and process using HDR or exposure blending.

And none of these suggestions is of any use if you don't have a tripod with you at the time .

[bambleweeney]

Bill, I've taken lots of HDR images without a tripod ... some of my most notable failures

[bambleweeney]

Dan, tried this once and the general idea is that as noise is random you sort of cancel it out by stacking images. It works if I remember rightly but I never found any practical use for it (apart from astrophotography). Anything moving in the image and I imagine it would be a nightmare.

[DanK]

Dan, tried this once and the general idea is that as noise is random you sort of cancel it out by stacking images. It works if I remember rightly but I never found any practical use for it (apart from astrophotography). Anything moving in the image and I imagine it would be a nightmare.

Very interesting. Re things moving: that would be a problem with long exposures too. This raises a question that I can't answer. The problem of noise in the shadows isn't that there is more noise there; it's that there is less signal, and when you boost the brightness to bring out the signal, you bring out the noise. So, if you expose dark and stack to remove noise, there should be less information in the shadows than there would be if you expose longer and then darken in post. However, I don't know how extreme this would have to be before you would notice it, particularly in a low-res image like ones on screen.

[Manfred M]

Dan, tried this once and the general idea is that as noise is random you sort of cancel it out by stacking images. It works if I remember rightly but I never found any practical use for it (apart from astrophotography). Anything moving in the image and I imagine it would be a nightmare.

It's a technique I've heard is also used in getting long exposures in other areas of photography. I remember watching a YouTube video on using it to emulate using a high density opaque filter for smoothing water (something a number of members object to), but I've never tried the technique myself.

The only multi-exposure technique I try from time to time is using it to get rid of people when shooting heavily visited sites and monuments. If one takes a sufficient number of shots, there is a Photoshop function that will automatically remove anything that has moved between the images.

As for the noise reduction claims, I would like to see some solid work on this to quantify this assumption as the underlying assumptions regarding random noise are weak (i.e. random noise will show up in either way of shooting, so the impact would be similar). The only possible difference that I see that could impact noise is that we know that temperature of the sensor and other electronics increases as these component heat up. If the sensor is turned off between exposures and cools down, then there could be some validity to the premise...

[DanK]

The only possible difference that I see that could impact noise is that we know that temperature of the sensor and other electronics increases as these component heat up. If the sensor is turned off between exposures and cools down, then there could be some validity to the premise...

I have some practical experience with this because of the amount of night photography I do. I have had temperature-related noise problems using very long exposures (I think it was 20 or 30 minutes) on a hot, humid night using a 50D, which is a noisy camera. I haven't had significant noise issues with my 5D III, even with exposures up to 10 minutes on hot nights. For example, this is a 7-minute exposure at ISO 100 on a summer night, with no noise reduction in post. I probably did use subtractive noise reduction in camera (subtracting a black exposure of the same length) in camera, but I can't be certain, and in any event, subtractive noise reduction addresses precisely the noise that stacking multiple exposures would not address: problematic pixels in a fixed location.



So I think that under most circumstances, one can ignore the impact of longer exposures on noise of the sort that stacking would remove.

[Manfred M]

Thanks Dan - I'm familiar with the in-camera noise reduction. It's technically referred to as Dark Frame Noise Reduction and the way it works is that the camera takes a picture that is the same exposure time as the original image but without opening the shutter. The dark frame exposure captures exposed pixels due to noise and this is subtracted from the regular exposure and this eliminates a lot of noise from the image.

The main downside is that it doubles the exposure time and if I recall correctly, it does not impact the raw data.

[DanK]

Yes, although it is also referred to as Long Exposure Noise Reduction in some Canon materials, and you will find people using the acronym LENR. Unlike other forms of in-camera noise reduction, it does affect raw data, at least in Canon cameras.

The only reason I pointed it out is that I didn't want to be inaccurate in saying that there was no noise reduction in the image I posted. There was no noise reduction that would address the random noise that stacking addresses.

Canon cameras have a setting that allows the camera to decide when LENR is warranted. I usually leave my camera set that way for urban night photography, but with a 5D III, the fixed noise apparently is never sufficient for exposures of that length (most often, less than 30 seconds) to trigger it, and I haven't seen any need for it in the results. However, when I am doing shots like the one I posted, I usually turn it on manually to play it safe, even though it means leaving the camera for 10-20 minutes instead of 5-10.

Dan

[xpatUSA]

<>

As for the noise reduction claims, I would like to see some solid work on this to quantify this assumption as the underlying assumptions regarding random noise are weak (i.e. random noise will show up in either way of shooting, so the impact would be similar).

Searching the net for << "noise reduction" averaging factor >> gets us lots of links to "solid work". For example:

B. Signal averaging

Signal averaging exploits the fact that if one makes a measurement many times the signal part will tend to accumulate but the noise will be irregular and tend to cancel itself. More formally, the standard deviation of the mean of N measurements is smaller by a factor of N than the standard deviation of a single measurement. This implies that, if we compute the average of many samples of a noisy signal, we will reduce the fluctuations and leave the desired signal visible. There are, of course, a number of complications and limitations in practice.

From http://www.owlnet.rice.edu/~dodds/Fi...oise_notes.pdf

The generally-accepted improvement in SNR is the square root of the number of averaged images. That is to say that the mean of e.g. 9 images will probably have an SNR 3x "better" than any one of those nine. The quote above mentions "complications". One complication would be the calculation of the SNR for the long exposure versus the SNR for the shorter exposure. Another is plain statistics. That is to say that any SNR improvement will virtually never be EXACTLY 3x; there's confidence levels, margins, sample size, population, etc. ad nauseam. In other words, I could post an image showing an enormous improvement in SNR and it would tell us next to nothing ... it would be a fluke!

The only possible difference that I see that could impact noise is that we know that temperature of the sensor and other electronics increases as these component heat up. If the sensor is turned off between exposures and cools down, then there could be some validity to the premise...

The wording of Manfred's post appears to imply serious doubt as to the extent of any benefits gained by averaging multiple exposures. On the other hand, people are wetting their pants elsewhere over the lack of noise in Sigma's recent "SFD" mode, which is only a simple in-camera bracket of 7 raw shots at intervals of 1 EV.

This whole discussion suffers from the common use of the single term "noise" without qualification and the lack of actual numbers.

Pardon my pedantry ...