22
\$\begingroup\$

I have been doing a lot of research lately in planning an upgrade to my old D70. One thing that has really jumped out at me is that I've seen the same general phrase repeated on three different discussion sites, from three different users, from time periods spanning several years. The phrase is always some variation of:

only photographers care about noise; real people don't even notice

My first foray into photography was in my teenage years with a 1996 P&S digital camera. I learned to hate photography because every image I took was full of noise, and I certainly wasn't a photographer at a time. Ever since, I've been on a crusade to reduce noise at all costs. That said, I've also learned over the years to be careful in extrapolating my own life experiences to others.

I'd like a answer that is as objective as possible. As such, I'd like to ask for some form of support for any claim. Some examples, in descending order of preference:

  1. A peer reviewed study
  2. A well designed and applied formal survey
  3. An informal study
  4. A casual public survey
  5. Anecdotal evidence

I'm sure noise becomes a problem for everyone at some point. So it may be helpful to also address at what level noise becomes an issue for photographers vs 'regular viewers'.

\$\endgroup\$
9
  • 20
    \$\begingroup\$ "I've been on a crusade to reduce noise at all costs" That's how you end up with skin that looks like plastic in photos ! \$\endgroup\$ Jun 11, 2019 at 14:52
  • 2
    \$\begingroup\$ It depends on the output medium. I am far less likely to see noise in photos on my phone unless I pinch to zoom in really really close. Likewise a portrait of someone at 100x100 px will undoubtedly always have noise and most people will chalk it up to being a low-res image. If you're printing a 10x10 foot banner then noise might make a difference based on your viewing distance. At the end of the day, life has noise and you see it because you choose to see it. \$\endgroup\$
    – MonkeyZeus
    Jun 11, 2019 at 15:44
  • 2
    \$\begingroup\$ @MikeBrockington Define "noise". I consider jpeg pixilation or artifacting as noise which is quite common in smaller images found on the web. \$\endgroup\$
    – MonkeyZeus
    Jun 11, 2019 at 15:49
  • 3
    \$\begingroup\$ No, certainly not. In this context, noise is where an otherwise smooth image has speckles that differ. Pixellation and artifacting are two very different things, though both can also be distracting. \$\endgroup\$
    – MikeB
    Jun 11, 2019 at 15:52
  • 4
    \$\begingroup\$ Some of the most iconic images are noisy. \$\endgroup\$ Jun 12, 2019 at 11:23

7 Answers 7

33
\$\begingroup\$

I do expect work has been done on noise perception to build perceptual models to compress images and compare image quality. However, I am unaware of any studies that compare photographer vs non-photographer perception of noise in digital images. I also did not see any in the first several pages of results of a Google Scholar search.

only photographers care about noise;

Photographers likely have a lower noise threshold because of factors, like pixel peeping. They are also more likely to edit images, which can enhance the appearance of noise, so it's worthwhile to minimize noise in the first place. They can also differentiate types of noise because of exposure and training.

How much one cares about noise depends on the photo and type of noise. For instance, I have low tolerance for chroma noise, but more luma noise isn't as objectionable. Since my current camera has well-controlled chroma noise, I don't mind pushing ISO to 12800. This has enabled more low light shots than I would have gotten with a DSLR that produces less overall noise, with more of it being chroma.

real people don't even notice

Non-photographers do notice noise, but may use different words to describe it. They may complain about color, sharpness, specks, etc. Some may recall the film days and call it "grain". Others may just think the image looks strange, but be unable to explain why.

There is a saying, The eyes do not see what the mind does not know. Point out the appropriate details and terminology to a layperson, and they will be able to see and describe it too. It's a learnable skill.

I'm sure noise becomes a problem for everyone at some point.

noise

\$\endgroup\$
12
  • 25
    \$\begingroup\$ There's something wrong with your Magic Eye autostereogram... it doesn't resolve to a 3D image, and it seems to move for some reason... ;-) \$\endgroup\$
    – scottbb
    Jun 10, 2019 at 19:03
  • 16
    \$\begingroup\$ There's a moth flapping its wings somewhere in there. \$\endgroup\$
    – xiota
    Jun 10, 2019 at 19:06
  • 8
    \$\begingroup\$ Photographers are exposed to frequent "bad image artefacts" and people get better at recognising patterns, good or bad. My experience of studying image processing and encoding technologies at university resulted in immediately noticing the encoding artefacts in broadcasts that I had not paid much attention to before. Photographers will notice the noise far more vividly than the general population. There may be studies on trained vs. untrained perception which could be explored? \$\endgroup\$
    – TafT
    Jun 11, 2019 at 7:57
  • 2
    \$\begingroup\$ It's in the context of rendered animation, but Yee, Pattanaik and Greenberg have done some work on perceptual differences between images. You might find some interesting things around there. \$\endgroup\$ Jun 11, 2019 at 10:26
  • 2
    \$\begingroup\$ All I can think now is TV static sound. \$\endgroup\$ Jun 11, 2019 at 13:27
18
\$\begingroup\$

Based on my informal study of my customer preferences and anecdotal evidences, I found that some laypersons do notice noise.

'Noise' is not a familiar term to most non-photographers but I heard my customers say words like, 'dots', 'roughness', 'pixellation' etc. Those who noticed it disliked it and told me that they hoped that I will ensure that the the photos I am going to give them won't have such problems.

There are other symptoms of bad photography like incorrect white balance, blur/sharpness etc. which some of my customers notice and that subconsciously influenced whether or not they liked a given photograph. It is likely that they are having the photos taken on their smartphones as reference like @Hueco said.

\$\endgroup\$
4
  • \$\begingroup\$ Anecdote: I'm not a photographer, but I know a fair amount about image quality from playing with video compression, e.g. finding x264 / x265 settings that give the acceptable quality at the lowest bit-rate with a tradeoff of CPU time for different videos. (And most video that's not pure CGI has some noise or film grain, so that's a well-known thing for me). Video noise-reduction filters to reduce artifacting and/or noise are usually worse than the original problem, making everything look plastic, unless the original video was horrible. :/ Although nlmeans is interesting. \$\endgroup\$ Jun 12, 2019 at 1:56
  • \$\begingroup\$ I'm also a physicist, so I understand why sensor noise is a thing. \$\endgroup\$ Jun 12, 2019 at 1:57
  • 2
    \$\begingroup\$ @PeterCordes Noise and plasticky looks are the two extremes of the Noise spectrum. I'd rather not be on that spectrum if I can help it but if I can't, I tend towards noise. Plasticky photos most definitely look amateurish whereas you can pretend noise is an intentional, artistic effect. Haha! \$\endgroup\$
    – raviputcha
    Jun 12, 2019 at 2:12
  • \$\begingroup\$ Yup, same. But unfortunately temporal noise (changing every frame) compresses very poorly in videos, and size/bitrate still matters for video files. So there's some reason to accept more flattening of noise (and textures / details) to get reasonable file sizes; psycho-visual tuning settings like favouring higher-frequency energy (even if it doesn't match the source), i.e. psy-rd can only do so much. As can AQ (spending more bits on textures instead of just edges). \$\endgroup\$ Jun 12, 2019 at 2:23
14
\$\begingroup\$

I don't think you'll find that this topic has been studied to the degree that you're looking.

You may have some luck in finding a study on perception based on some tangible knowledge or background - but what exactly that background/perception mix is...well, who knows? My wife is in school for her PsyD and has access to more reports than I could ever hope to read...if either of us find one along these lines, I'll update this answer with it.

In the meantime - my anecdote is along the lines of the comments. What people perceive is based on their existing awareness. Fact is, most consumers shot disposable film back in the day - which had its own problems in low light. When digital came out, well, I remember rocking a Coolpix 4100 (4MP CCD) and it was the greatest thing ever - because it was one of the first digital cameras out there and it was replacing a disposable film camera.

Now, people mostly shoot with their phones. Phones take nice, clean images in good lighting and absolutely noisy, terrible images in bad. If, in the same bad lighting, your DSLR shot not only has less noise than an iPhone but the subject is frozen (no motion or subject blur) then you have far outdone the iPhone. The person used to seeing iPhone images alone will see yours as "cleaner" (less noise) and outstandingly sharp (no blur).

The noise that you see is simply because your mental bar is higher and harder to clear.

Other reasons may include: you pixel peep while most people don't; You look at images with an artistic eye and most people look at subjects with a boolean attitude (did you get the subject or not?); some people (lookin' at you, mom) still end up shooting their finger, even with an iPhone...noise is the least of their concerns in their own photos, let alone yours.

\$\endgroup\$
9
  • \$\begingroup\$ "some people... still end up shooting their finger, even with an iPhone" – I do that. That's part of why I can't use phone cameras. \$\endgroup\$
    – xiota
    Jun 10, 2019 at 18:58
  • 7
    \$\begingroup\$ Re. phone performance in dark environments - this used to be a real sticking point, but mobile cameras nowadays are much better in the dark - including longer exposures with better sensors. Some phones also utilize advanced software trickery (similar to astrophotography's stacking techniques) to result in excellent shots in dark environments - see Google's Night Sight feature on the new Pixel phones. \$\endgroup\$
    – Tyzoid
    Jun 10, 2019 at 21:30
  • 4
    \$\begingroup\$ Quality Control: This is a better picture than that one. See, this one has her head and feet and that one you don't show her feet. \$\endgroup\$
    – Stan
    Jun 10, 2019 at 21:30
  • \$\begingroup\$ @xiota Try flipping the camera 180° so the lens is on the other end from the offending finger. \$\endgroup\$
    – Michael C
    Jun 12, 2019 at 7:36
  • \$\begingroup\$ @MichaelC That's not helpful, there are fingers on the other side too. \$\endgroup\$
    – xiota
    Jun 12, 2019 at 7:42
13
\$\begingroup\$

Part of why noise is a problem in digital photography is ironically due to noise reduction. Grass "dissolving" into the distance is one of the things that is affected worst for me: noise reduction tends to construct areas of average color under the theory that different colors may be due to noise, and the patterns created by those constructed areas do not scale along with perspective. That makes the fine structure of denoised grass in the background reject perspective which I find very distracting. The impressive high ISO performance of newer cameras is to a good degree due to improved noise reduction algorithms which do a pretty good job on actual connected or regular surfaces. On chaotic but scale-specific input (like grass) the results interfere with human vision.

Analyzing, recognizing and describing problems require skills, but that doesn't mean that people's perception is not affected by things they cannot pinpoint.

For that reason, surveys such as the one you imagine stand the danger of understating the relevance of image artifacts to a typical viewer's reception and appreciation.

\$\endgroup\$
1
  • 6
    \$\begingroup\$ I concur. My tolerance for noise is way higher than my tolerance for excessive, detail-killing, noise reduction. \$\endgroup\$ Jun 11, 2019 at 8:53
11
\$\begingroup\$

Many non-photographers will appreciate a sharp, lifelike image, with a high degree of detail visible.

Non-photographers may not always be able to distinguish all the different reasons why a photograph lacks detail (e.g. poor focus or limited depth of field, or camera shake, or lens distortion, or noise, or over-saturation, or limited pixel count). However, they may well enjoy a picture that does not exhibit these issues. In that sense, non-photographers do care about noise.

(As a counterpoint to this, I also agree with Stan's point that film grain can be aesthetically pleasing - even to those who don't recognise the technical reason in the final photograph. Whether digital sensor noise is quite as pleasing may be another question!)

\$\endgroup\$
1
  • 5
    \$\begingroup\$ THIS. Not only photographers will notice reduced contrast, reduced saturation and reduced detail. Which are direct effects of noise. \$\endgroup\$ Jun 11, 2019 at 12:16
9
\$\begingroup\$

Beauty is in the eyes of the beholder.

It used to be popular to shoot Kodak Royal Pan X film rated at 'ASA' 1200 (That was fast back in the day.) and "soup" it in Dektol (Kodak D-72 paper developer) to get "popcorn" sized grain with practically no enlargement.

The noisy grain-pattern was what we were after as aesthetic expression.

We'd try the same thing with whatever we could shoot and soup to get that "look" of urgency——even better with colour.

It wasn't noise then. It was the effect we were striving for.

\$\endgroup\$
4
\$\begingroup\$

I think what matters is not who is looking at the picture, but what the picture is about. If we are talking about photojournalism, or about a photo of a special instant - something hard to shoot - then noise is not going to be important. That doesn't mean that it is not going to be noticed, everyone can see the grains. It's just that the subject is so strong that noise - along with other parameters of your composition - will have less weight.

If, on the other extreme, you have one of those landscape picture where everything is still, light is perfect and so on, people will notice - and by that I mean really notice - everything. Not only noise but also sharpness, if the camera is leveled, if your camera was still, if the day was clear, etc etc etc.

So it is more about what you want to show with your photo. If your subject is something that may happen only once, or in just a quick instant, then your subject is more important than a perfect setting. In this case, you go for a high ISO and take your shot of the rare moment. But if you want a picture that people will keep looking at it, entering the photo and exploring every corner of it, then you should invest in composition and light.

For example, you can easily see the noise in the picture bellow, but you are not going to notice it if you are not looking for it:enter image description here

\$\endgroup\$

Your Answer

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

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