2
\$\begingroup\$

How come I can't tell the difference in quality of these two photos, even though one is shot at ISO 160, and the other at 1600 ?

Note: EXIF data has been stripped from images below by imgur saving process.
Originals are here and here. Images must be saved to disk then opened to see EXIF data. Both photos at 39mm focal length (35mm equivalent).

ISO 160 f/22 8 seconds:

ISO 160 sample

ISO 1600 f/4 1/40 second

ISO 1600 sample

What difference should I see between 160 and 1600 ISO?

These were both taken with a Nikon D7000.

Improve this question
\$\endgroup\$
15
  • \$\begingroup\$ I know you've already accepted - but...Were these shot RAW or jpgs out of camera? Also, the D7000 has pretty good handling at 1600 - its not that high for the D7000. \$\endgroup\$
    – rfusca
    Commented Jun 4, 2012 at 4:02
  • 5
    \$\begingroup\$ @Sonic Soul: You might want to wait longer before selecting an answer as accepted. You only posted your question less than an hour ago, somewhat late at night for the US...there could be many very good answers just waiting to be offered tomorrow. Since you have already selected an "accepted" answer, its very possible no one else will stop by to offer those answers, though. Just a tip. \$\endgroup\$
    – jrista
    Commented Jun 4, 2012 at 4:10
  • \$\begingroup\$ Consider the points in my answer re technical differences in how the photos were taken. The two sets of settings are so different as to make comparison unlikely to be meaningful. Comment on why would be very interesting. Also details differ in the photos suggesting something arcane was at work in the processing. Do you know why or how this happened? \$\endgroup\$
    – Russell McMahon
    Commented Jun 4, 2012 at 10:59
  • \$\begingroup\$ actually those were "fine" jpegs. i thought it was raw, but it was dark and i set it incorrectly. would raw make much difference in respect of iso noise?? \$\endgroup\$
    – Sonic Soul
    Commented Jun 4, 2012 at 12:18
  • \$\begingroup\$ jrista, good point. i just felt a little bit silly that i posted my question before actually zooming to 100%, because at that level the difference was apparent. so i thought my question was not that good which is why accepted so quickly. \$\endgroup\$
    – Sonic Soul
    Commented Jun 4, 2012 at 12:19

7 Answers 7

Reset to default
2
\$\begingroup\$

In the dark area in the top 1/3 of the photos, you can see more noise in the ISO 1600 version. It shows up as the uneven graininess. I think the noise isn't as obvious in the areas of the wall because the colors are very bright and also very distinct--there are no gradual fades from one color to another.

Improve this answer
\$\endgroup\$
2
  • \$\begingroup\$ you're right.. i zoomed in to 100%, and focused on upper area. i get it now :) \$\endgroup\$
    – Sonic Soul
    Commented Jun 4, 2012 at 3:58
  • \$\begingroup\$ after zooming, even the lit areas are grainy at 1600 \$\endgroup\$
    – Sonic Soul
    Commented Jun 4, 2012 at 3:58
5
\$\begingroup\$

I'm adding this as a second answer as it is probably the fundamental issue involved.

Apples <> Pears

  • The two pictures have fundamentally different conditions and cannot be fairly compared.

  • The ISO 160 picture contains 30 times more sensor noise than it should have for a fair comparison.

What I and others have said in the other answers all has relevance, but -

1600 ISO settings were f/4, 1/40s
160 ISO settings were f/22, 8 seconds.

Assuming starting at ISO 1600 (either way works)
If exposure time had been decreased when ISO was decreased then the change would have been:

  • f/4, 1/40s to f/4, 1/4s

ie ISO ratio = 1600/160 = 10
exposure ratio = (1/40) / (1/4) = 10

BUT at the same time, aperture was changed from f/4 to f/22
This necessitated an increase in exposure time by a factor of apertures-squared (as f number is a measure of diameter but light input is related to area which is proportional to diameter squared.
So exposure time had to be increased by a furher ratio of (22/4) squared
= a further 30.25 times !!!

So the ISO 160 picture contains 30 times more sensor noise than it should have for a fair comparison.

This still does not explain the actual content differences in the images.

Improve this answer
\$\endgroup\$
2
  • \$\begingroup\$ interesting, so you are saying that the longer the exposure is, (the smaller the aperture) the more noise gets in? \$\endgroup\$
    – Sonic Soul
    Commented Jun 4, 2012 at 15:26
  • 1
    \$\begingroup\$ @SonicSoul - Yes - sensor noise is integrated with time. Longer exposures = more noise. There are a number of noise sources and not all add linearly but a major portion does. The above is the best answer given so far, but nobody has noticed, and they never may :-). \$\endgroup\$
    – Russell McMahon
    Commented Jun 4, 2012 at 18:44
1
\$\begingroup\$

There is a visible difference between the two shots...there is more noise in the ISO 1600 shot, albeit minimally. Part of the reason you are seeing such a minor difference is due to the fact that the Nikon D7000 has a truly stellar Sony Exmor sensor in it, with unparalleled noise characteristics and dynamic range. If you use an older Nikon camera or even cameras from most other brands, you would probably see more significant image degradation at ISO 1600.

Additionally, these days, ISO 1600 is not the maximum ISO setting...it is often several stops below the maximum. That usually means cameras with higher ISO settings, such as 12800, 25600, and even 51200, look better...sometimes MUCH better...at ISO 1600 than cameras from even just a few years ago. My first camera was a Canon 450D, which had a maximum ISO setting of 1600. It obviously looked quite terrible at that setting. The newly released Canon 5D Mark III supports ISO all the way up to 25600, and its ISO 1600 looks as good as my 450D's ISO 400. The same would be true of the Canon 1D X or Nikon D800.

I find it a little ironic that you have such a powerful camera with one of the best sensors on planet earth...and you don't know about the benefits it has to offer. ;) But you can pretty much thank your camera for having stupendous image quality as the reason you don't see much of a difference in those two shots.

Improve this answer
\$\endgroup\$
5
  • \$\begingroup\$ yes, i am no hardware guru, but actually image quality and low noise at high ISO were major selling points in this camera for me. The reason for the question above is just me trying to understand the sensor, and exactly what to expect at different settings. thanks for taking the time to answer \$\endgroup\$
    – Sonic Soul
    Commented Jun 4, 2012 at 12:24
  • \$\begingroup\$ If that is the question you are asking, it definitely warrants a very different answer. If you are interested in the detailed inner workings of how a sensor receives light, amplifies it, and converts it into a digital image, I could certainly provide that. It would be a much larger and more complicated answer, and I think a new question better formulated to that specific request would be a better place to offer such an answer. \$\endgroup\$
    – jrista
    Commented Jun 4, 2012 at 17:19
  • \$\begingroup\$ jrista, i appreciate your willingness to help. the question i was asking was caused by my initial thoughts that lower ISO should not only decrease the amount of noise, but also improve image quality including things like quality of colors. basically i was looking for a more visible difference between 160 and 1600. but i get it now that the iso range is much larger and the difference is narrowing in newer cameras (mine being brand new). i also learned about "exposure time noise".. very helpful.. i'll make another comparison on more even grounds. \$\endgroup\$
    – Sonic Soul
    Commented Jun 4, 2012 at 18:34
  • \$\begingroup\$ Also note, since they were in camera JPGs they probably also had NR applied to further minimize the difference in noise. \$\endgroup\$
    – rfusca
    Commented Jun 5, 2012 at 0:25
  • \$\begingroup\$ @SonicSoul: I wouldn't really say there is such a thing as "exposure time noise". Its more complicated than that, and there are specific types of noise that can (but don't necessarily do) increase as exposure length increases. A proper answer could be lengthy, so I might actually write a blog entry about it on the PhotoSE blog. I'll post a not here if/when I get that done (hopefully soon, if I find the time.) \$\endgroup\$
    – jrista
    Commented Jun 5, 2012 at 0:56
1
\$\begingroup\$

ISO does not cause noise in photos.

Low levels of light causes noise in photos. High ISO settings alter camera metering in certain modes which causes the camera to choose shutter speed / aperture settings which don't let enough light in, resulting in increased noise.

But

The camera's metering can get it wrong, and not let enough light in even with a low ISO setting, so using a low ISO is not a guarantee of getting less noise.

See the following example:

Here both images received the same amount of light, but the ISO100 shot looks a lot worse because it was underexposed (I used auto levels in Photoshop to normalize brightness to facilitate a visual comparison of signal to noiseratio).

Improve this answer
\$\endgroup\$
5
  • 1
    \$\begingroup\$ Without knowing the light level in the photo(s) it is not possible to be sure what effects are occurring but it seems likely that the 100 ISO image was underexposed, and so amplified by the camera based on the information provided such that available low level bits of ADC output were discarded, whereas if they had been retained the result would have been better to far better. ie if this is the case, then what Matt is saying is true BUT effectively the user is lying to the camera. This setting in this light level says that the user values higher .. It is very likely that processing \$\endgroup\$
    – Russell McMahon
    Commented Jun 4, 2012 at 10:49
  • 1
    \$\begingroup\$ ...lighting levels in some other part of the picture. eg in a real world example this may be a corner bookshelf in an otherwise brightly lit scene with the main focus on eg a character by a window. In such a case the proper choice is to let the shadows "noise up" so that the subject highlights are not blown. ie you can certainly push the camera into situations where low ISO is worse in some parts of the picture but the supplied example is unlikely for the main subject in a photo. A "live view" screen in the top case would have been 4 stops too dark,almost unviewable and not liable to be used. \$\endgroup\$
    – Russell McMahon
    Commented Jun 4, 2012 at 10:56
  • 1
    \$\begingroup\$ @Russell McMahon The light level was the same, the ISO100 shot was underexposed (as I state in the answer). I think you're going in to too much detail here - the point I was trying to make is this: just because an image was shot at a lower ISO value doesn't automatically mean noise will be lower - there are other factors involved. \$\endgroup\$
    – Matt Grum
    Commented Jun 4, 2012 at 17:24
  • 2
    \$\begingroup\$ Lol - everytime I see these two shots, I know the point Matt will make. \$\endgroup\$
    – rfusca
    Commented Jun 5, 2012 at 0:26
  • 1
    \$\begingroup\$ @rfusca I don't mean to keep going on about this point but the very existence of questions like this indicate there is still widespread misunderstandings about the origins of noise. I've seen several experienced photographers claim they never go higher than ISO 1600, because high ISOs 'create too much noise', and they then go on to underexpose all their ISO 1600 shots, thinking the lower ISO value makes some sort of guarantee about noise... \$\endgroup\$
    – Matt Grum
    Commented Jun 5, 2012 at 12:44
0
\$\begingroup\$

When noticeable a higher ISO will show a graininess like if the image was printed onto paper with a cheap printer and then you took a photograph if that paper. As sensors improve cameras can take pictures with a higher ISO before this is noticeable. See this review of your camera, with test shots at the bottom, for a more pronounced effect. http://digital-photography-school.com/nikon-d7000-review As you can see, if you bump up your ISO even more you will be able to notice, but the difference in ISO for your two shots isn't significant enough to be immediately visible.

Improve this answer
\$\endgroup\$
2
  • \$\begingroup\$ ok, but do you see any difference between those 2 shots ? \$\endgroup\$
    – Sonic Soul
    Commented Jun 4, 2012 at 3:29
  • \$\begingroup\$ thanks, i went back and looked at zoomed in versions. that finally revealed a more clear difference in the graininess. \$\endgroup\$
    – Sonic Soul
    Commented Jun 4, 2012 at 12:35
0
\$\begingroup\$

There appears to be something fundamentally wrong happening here with the image content, and the camera settings appear to have been chosen oppositely to what is intuitive. Substantially more information is needed to make full comment - camera settings (noise reduction, any anti shake, , tripod?, through glass?, path from camera to each image.

It seems that to a reasonable extent changes in ISO performance may have been masked by camera settings. The image content change is hard to understand. It's conceivable that whatever " program developed" these jpgs from the raw image (whether in camera or externally) did so "creatively".

Pixel peeping suggests that the 1600 ISO image is both sharper and has more fine detail in places.

Aim of the groups of finer "runs" of paint are different between photos. Not just more detail or less detail relatively bur different detail - eg a loop of 'paint' on one run appears to have appeared on a different run.

In addition, the low ISO picture has a darker band at the bottom (about 20% of height) and has reflections in it. The illumination changes at top left and top right (brighter on 160 ISO image). Both photos have highlights blown in areas around the lights (higher on 160 ISO) mostly in the red channel.

Somewhat unexpectedly, the EXIFs report 160 ISO = f/22 and 8 seconds (NOT 1/8) and the 1600 ISO reports f/4 and 1/40th second. These correspond to exposures of withing 6% of each other indicating that the figures are probably correct. For meaningful comparisons I'd have expected constant aperture and shutter speeds scaled the other way eg
f/4, 1/40th. ISO 1600 and
f/4, 1/400th, ISO 160.
Better still, as long as a tripod was used, would be to reduce aperture to get better depth of field and use longer exposure times - say f/11 and about 1/4s and 1/40s.

Improve this answer
\$\endgroup\$
4
  • \$\begingroup\$ thanks Russel. i did realize that aperture difference does not help a direct comparison (did not think of it at the time of shooting). Does depth of field matter that much when shooting an object that is equal distance all throughout the photo (such as a wall) ? also, why f/11 and not f/5.6 ? \$\endgroup\$
    – Sonic Soul
    Commented Jun 4, 2012 at 12:31
  • \$\begingroup\$ there was a tripod, no glass, also in one shot a car went by which is why you see the some reflection in lower left corner... i'd like to hear more about why you state that the camera settings were counter intuitive? at first i shot with low ISO to get best wall detail.. following that, i increased it to decrease apperture, and capture subjects crossing in front of the wall \$\endgroup\$
    – Sonic Soul
    Commented Jun 4, 2012 at 12:33
  • \$\begingroup\$ @SonicSoul If you increased ISO and decreased exposure time in proportion aperture will remain constant (or increased both). This would be most usual when making comparisons. This is a "fair" tradeoff between ISO and exposure time. BUT when going from 1600 to 160 you not only made exposure time longer by the ISO change ratio = 1600/160 = 10:1 BUT you chaged from f/4 to f/22 which increased the exposure time needed by a FURTHER (22/4) squared = 30 times !!! So you have 30 times as much exposure time noise in the ISO 160 photos than if you has stayed at f/4. \$\endgroup\$
    – Russell McMahon
    Commented Jun 4, 2012 at 13:11
  • \$\begingroup\$ thank you sir. "exposure time noise" was a concept i was not aware of. \$\endgroup\$
    – Sonic Soul
    Commented Jun 4, 2012 at 15:29
0
\$\begingroup\$

There is significant difference between the two pictures. Here is the same small section of each blown up so each pixel is replicated 2x2:

More noise is visible in the second picture, which is to be expected since it is taken at higher ISO near the edge of what the camera can do.

Both cases also have compression artifacts. These are actually much easier to see in the first picture because it is smoother so that the JPG compression algorithm applied a higher compression ratio. This is a good example of why JPG compression should only be used for the final post processed image, not for intermediate images that you might want to do other things to. It seems silly to worry about noise in the second picture but ignore compression artifacts added artificially by the camera in the first.

Improve this answer
\$\endgroup\$
4
  • \$\begingroup\$ If you got the imgur versions, note that I got those from tinypic, and both of those services probably recompressed the image from the original, significantly exacerbating the effect. \$\endgroup\$
    – mattdm
    Commented Jun 5, 2012 at 0:17
  • \$\begingroup\$ @Olin Lathrop - It seems to be foggy here at present. Nobody seems to have noticed that instead of just offsetting ISO change with speed change he also changed aperture from f/4 to f/22 thereby arbitrarily adding 30 x as much noise in the 160 ISO shot. I don't seem to be able to yell any louder :-). \$\endgroup\$
    – Russell McMahon
    Commented Jun 5, 2012 at 7:03
  • \$\begingroup\$ @Russell: It would have been good to keep the number of changed parameters down, but in this case small aperture isn't the issue. The noise it would add would be to soften edges due to diffraction. That may explain why the first picture appears less sharp, but it has nothing to do with the random noise looking artifacts between the two lights in the bottom picture. \$\endgroup\$
    – Olin Lathrop
    Commented Jun 5, 2012 at 23:40
  • \$\begingroup\$ @OlinLathrop - it's not that the parameters have changed - its that one has been changed quite unnecessarily in a manner which increases exposure times by ~3,000%. The 160 ISO shot had to contend with vastly more potential sensor noise so noise must be vastly lower for it to not show excessively (as would be expected.) The fair comparison is f/whatever and either (ISO 1600, 1/40th) and (ISO 160, 1/4) OR Constant exposure time and (ISO 1600, f/X) and (ISO 160, f/X/(sqrt(10) ) ) ... \$\endgroup\$
    – Russell McMahon
    Commented Jun 6, 2012 at 12:21

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.