2

I often find myself taking still pictures of some gizmo using a tripod mounted DLSR. One decision that always vexes me is what ISO to set my camera to and so minimise sensor noise. The important aspect of tripod mounted photography is that there is no need to consider any form of camera shake (given a decent firmly erected tripod).

I expect to find an absolute quantitative scientific answer, such as set ISO = 755.2, either from a formula or a graph. I've reviewed this forum and specifically this, this and this question. Disappointingly I find words and pictures of books. All very qualitative. I don't find any formulae or graphs. I don't accept that noise level is an artistic or subjective choice. It's a Gaussian distribution around a mean per each sensor pixel.

For my make of model of camera at a fixed temperature, sensor noise is a function so:-

Noise amplitude = f(ISO setting, exposure duration, sensor incident light)

and you might develop sensor incident light as:-

f(aperture, ambient light)

I'm holding image exposure level as a constant as it's got to fit into about 4 - 5 stops, otherwise you can't see it. Noise is a stochastic predictable process. The distribution's size and shape is entirely deterministic via experimentation. So it's a function of maybe four variables. These variables can be related together in a formula or series of graphs that could be published by a camera manufacturer. Where's that noise formula as I can't find it? Or is it as simple as use lowest ISO irrespective of exposure duration?

mattdm
  • 143,140
  • 52
  • 417
  • 741
Paul Uszak
  • 1,012
  • 7
  • 14
  • 7
    Why do you believe the answer might be anything other than "use the lowest ISO?" – Philip Kendall May 03 '17 at 23:01
  • @PhilipKendall 'cause there might be non linear relationships involved. See comments to answer below as to why you might be wrong... – Paul Uszak May 04 '17 at 01:27
  • 1
    @PhilipKendall this is not as clear cut as it seems. If you convince your camera to shoot at a lower ISO than the native one (e.g. Canon 6D allows shooting on ISO 50, native is ISO 100) sensor noise will not improve; provided you do not change your aperture you need to increase exposure time. This needlessly prolongs exposure and thus increases risk of camera shake. – Jindra Lacko May 04 '17 at 10:44
  • @JindraLacko I actually know that, it was just a bit too much to put in a comment. The risk of camera shake on a tripod is pretty much irrelevant though, but there is the issue of heat dissipation as Michael Clark has pointed out. – Philip Kendall May 04 '17 at 10:51
  • I did not mean to offend, just to clarify :) You and Michael Clark are right about the sensor heat. I like the story how some astrophotographers for this reason choose to shoot film in a fully mechanical camera. – Jindra Lacko May 04 '17 at 11:03

3 Answers3

5

Generally, the best performance is on your camera's native ISO. This is usually ISO 100, but not always - some Fuji cameras use ISO 200 (go figure..). Check your manual.

For decently lit subjects (gizmos on a table, not the Milky Way) and exposures in low single digit seconds sensor heat should not be an issue.

I would worry more about camera shake - even an expensive tripod alone is no guarantee of shake elimination. Remote release is safer than pressing camera button. Shooting with live view - or even better tethered from computer - eliminates mirror shake. Table mounted camera - an old enlarger stand is an option - is more stable than a tripod. I have even heard of shots ruined by subway passing under the building.

Jindra Lacko
  • 5,990
  • 1
  • 11
  • 39
  • If you dont have neither remote control neither PC connection, you can try 10s countdown. A man walking by can ruint the image also. – Crowley May 04 '17 at 08:34
  • 1
    @Crowley Ruint? Hmmm – Janardan S May 04 '17 at 09:25
  • @JanardanS It was a typo. Luckilly, google says ruint means "to fall" in latin. – Crowley May 04 '17 at 10:56
  • @Crowley Latin? Wow. The quality of the membership has really improved :-) – Paul Uszak May 04 '17 at 11:38
  • The problem with using Live View instead of mirror lock up is that the sensor remains energized constantly and thus heats up more. That's the last thing you want when doing astrophotography. Remote release and mirror lockup is the best solution with a DSLR. – Michael C May 04 '17 at 22:17
  • Lots of early Nikon DSLRs and some Canon DSLRs also had a native ISO of 200. – Michael C May 04 '17 at 22:22
2

The best ISO for minimum sensor noise depends on the camera. This is often listed as the "native" ISO in the camera manual. This value is 200, for example, in my Nikon D3s.

However, there is also noise that accumulates with long exposures. If the native ISO requires a long exposure, it's quite possible that higher ISO and shorter exposure results in overall less noise.

I'm not sure how long "long" is in this context, but I expect it would need to be more than at least a second or two before this effect is significant. Look at the maximum exposure time your camera allows. The manufacturer doesn't want you taking longer exposures than that because the noise would accumulate to where you wouldn't be happy. That is often around 30 seconds for sensors of current technology. Clearly you don't want to be near that.

There are digital cameras that are meant for much longer exposures, but those actively cool the sensor. Scientific telescope cameras do that, for example. I'll assume you're asking about a more normal consumer or professional photographer camera at most, so this doesn't apply.

In the end, the best way to answer this question is to run some experiments with your particular camera. Try taking de-focused pictures of a evenly lit piece of paper or something, at different ISO and exposure times. You can then use a computer program to find the high frequency noise content and give each picture a quantitative noise score.

Olin Lathrop
  • 17,402
  • 1
  • 32
  • 68
0

The lowest possible ISO gives the lowest noise.

Exposure time has no influence; duration may help a bit because the noise averages away a bit, but not as much as the higher ISO you need to compensate.
Opening the aperture all the way also reduces noise, but it has other effects on the composition and picture quality, so you should only do that if you really need to.

Aganju
  • 2,022
  • 1
  • 11
  • 17
  • Not exactly. If using the lowest ISO necessitates a shutter time so long that the sensor heats significantly more than at a higher ISO, the increased noise due to the warmer sensor may more than offset the increased noise from higher amplification using a higher ISO and a shorter exposure. – Michael C May 04 '17 at 00:35
  • Keep in mind that the amount of heat produced by the sensor being energized will be fairly constant. But the amount of heat dissipated by the camera per unit of time could mean the sensor temperature does not increase in a linear fashion if the camera can't dissipate it as fast as it is being produced. It's much like the memory buffer when shooting continuous frames: if the frame rate is low enough the camera can shoot at the same rate as long as the shutter button is pressed. If the frame rate produces images faster than the data can written to the card, then the camera must slow down. – Michael C May 04 '17 at 00:40
  • 1
    @MichaelClark Are you able to quantify a shutter time so long that the sensor heats significantly? Seconds? Minutes? Hours? – Paul Uszak May 04 '17 at 11:42
  • @PaulUszak That would totally depend upon a number of variables: The camera and how its design moves heat away from the sensor, the ambient temperature, the temperature of the sensor at the beginning of the exposure, etc. If you want a specific answer then you need to ask a more specific question. What camera are you using? In what environmental conditions are you using it? How may such exposures do you wish to take consecutively without allowing the camera to cool to the ambient temperature between each one? – Michael C May 04 '17 at 21:38
  • In most normal use cases for mass produced DSLRs, an exposure longer than one second is considered long enough for heat to affect the resulting image. Just how much the effect is depends on the previously mentioned variables. That's one of the disadvantages of cameras that require the use of the sensor to provide a preview image via an LCD or EVF. – Michael C May 04 '17 at 21:40
  • @MichaelClark I guess that's the underlying point behind my question. Why does the manufacturer not supply this information? This not an artistic point, it's entirely stochastic based on the camera physics. They clearly use some algorithm for AUTO-ISO, why don't they (or 3rd party) publish tripod data? – Paul Uszak May 04 '17 at 22:00
  • @PaulUszak Probably because the many variables mean the different ways different users employ the cameras would mean an immense amount of data would need to be measured and published. That, and from a marketing standpoint, if such data were released the absolute worst case possible that gives the worst result would be proclaimed far and wide by fanboys of other cameras as "THE noise level of that useless piece of junk." – Michael C May 04 '17 at 22:21
  • @PaulUszak Please see the answer I just posted to At what exposure times do fixed pattern noise become apparent? ISO and Tv are intimately related with regard to S/N ratio and the effects of sensor heat. – Michael C May 04 '17 at 22:25