Talk:Exposure (photography)

From Wikipedia, the free encyclopedia
Jump to: navigation, search
WikiProject Film (Rated Start-class)
WikiProject icon This article is within the scope of WikiProject Film. If you would like to participate, please visit the project page, where you can join the project and see lists of open tasks and regional and topical task forces. To use this banner, please refer to the documentation. To improve this article, please refer to the guidelines.
Start-Class article Start  This article has been rated as Start-Class on the project's quality scale.
Checklist icon
Taskforce icon
This article is supported by the Filmmaking task force.
WikiProject Photography  
WikiProject icon This article is within the scope of WikiProject Photography, a collaborative effort to improve the coverage of photography on Wikipedia. If you would like to participate, please visit the project page, where you can join the discussion and see a list of open tasks.
 ???  This article has not yet received a rating on the quality scale.

Total amount of light[edit]

What do you mean by "the total amount of light"? --- Miguel April 12th 2006.

The article on light value explains this a bit better. In a nutshell, you got an adjustable amount of light reaching the sensor while the shutter stays open for an adjustable period of time. "Total amount of light" refers to the exposure value, which is the product of the aperture value (which determines how much light is transmitted to the sensor) and exposure duration (except that everything is usually measured on a logarithmic scale, so EV is actually expressed as a sum). Depending on the application, exposure may be specified in foot-candle seconds. --MarkSweep (call me collect) 02:58, 13 April 2006 (UTC)
To get a total amount of light in the focal plane, you need more than EV. You also need to know the luminance of what you're pointing it at. The net result is usually measured as lux seconds (in SI units), the product of focal-plane illuminance times the exposure time. It is straigtforward to calculate a focal-plane illuminance from an f-number:
E = {\pi L \over 4 N^2}
where E is focal-plane illuminance in lux, L is scene luminance in candela per square metre (cd/m2), and N is the f-number (assuming unity transmittance, for which f-numbers are usually adjusted anyway).Eq. (6) The result is not really a "total amount", but an amount wavelength-weighted by the luminosity function to model how bright that amount appears to human eyes. Multiply E by exposure time to get total exposure in lux seconds. Dicklyon 22:38, 2 January 2007 (UTC)

Double exposure[edit]

double exposure is when u take one image make a image over it. but u can't over expose it.

Included multiple exposure bit into preamble/disambig. Hope that covers it. --Redbobblehat (talk) 02:21, 12 February 2009 (UTC)


Modified a number of areas, in particular info that was not correct on latitude (film/digital/slide etc), correct exposure (there is none! etc). Latitude seems to have been mistaken with "dynamic range".... More work needed here....any thoughts feel free. —Preceding unsigned comment added by Barryfitzgerald (talkcontribs)

I added a header and reformatted your note a bit; hope you don't mind. Dicklyon 21:37, 2 January 2007 (UTC)
I'm wondering what's behind some of your latitude statements. Do you have sources? In particular, digital at high ISO doesn't have less latitude, it just moves the latitude from the shadow end to the highlight end (at least in cameras that I'm familiar with that capture RAW data independent of the ISO setting). Dicklyon 22:14, 2 January 2007 (UTC)

My experience suggests that at say ISO 100 you can get over 2 stops worth of info back, at ISO 1600 that trashes the image quality...feel free to tart up the text...just put it down quickly. Dpreview supports this in their reviews also.. example:

Note reduction in dynamic range etc etc.

Yes, but it's too bad they didn't combine the RAW and high-ISO tests to see what the real highlight range is. These guys are pretty good at camera phenomenology, but don't really know enough about the science of it to design good tests and reports sometimes. Dicklyon 23:06, 2 January 2007 (UTC)

With latitude of film/slide/digital...well it is well known film has good highlight, poor is the many ways..and not at all like slide which has little lattitude either side in general.

Yes, but since digital can be underexposed, equivalent to shooting at higher ISO, you can get back a more symmetric range, more like negative film; you need a decent raw file and raw conversion program to take advantage of it, though; as shown by dpreview, Nikon tends to hard-clip their highlights, while Canon does a nicer film-like shoulder in their rendering.
Let's be careful about repeating impressions and folk wisdom, and go for verifiable facts instead. Dicklyon 23:06, 2 January 2007 (UTC)

I would like to help expand the Latitude + Overexposure + Underexposure section(s), perhaps they deserve their own article eg exposure latitude? IMO a "Latitude-lite" section should also try to mention exposure bracketing, dynamic range (qua toe-linear gamma-shoulder) as referenced in film speed and HD curve. --Redbobblehat (talk) 02:10, 12 February 2009 (UTC)

Wikipedia needs to address the oft-quoted-and-never-cited "expose for the shadows and develop for the highlights" rule as it applies to negative film, but the opposite for reversal film or digital/video ... here would seem to be a good place. Was it Anselm Adams ? --Redbobblehat (talk) 02:10, 12 February 2009 (UTC)

Exposing to the right[edit]

The section on Latitude currently states: "Digital should be considered to be the reverse of print film, with a good latitude in the shadow range, and a narrow one in the highlight area." However, this is not the whole truth. As I understand it a finished digital image (eg, tiff), where the tonal values are non-linear, has more or less equal latitude in the shadows and highlights, but a digital negative (raw), where the tonal values are linear, has far greater latitude in the highlights - provided that one is using a linear editor (raw converter).

The reason I bring this up is because I have recently re-written/expanded the article on Exposing to the right, which is predicated on the concept of digital negatives storing tonal value data in a linear fashion. ETTR is an oft misunderstood technique which I must admit to being not 100% certain of myself (mostly due to the sheer volume of naysayers!) I'd really like to invite comments from experts in the field, hence I have created this subsection. Alternatively, you may wish to leave comments on the ETTR talk page. Thanks. Regards, nagualdesign (talk) 23:49, 26 November 2011 (UTC)

That sounds like inappropriate original research to me. It really doesn't matter whether the values are stored linear or gamma-compressed; the problem in the highlight range is the hard clipping at the max value. As for ETTR, I've always thought it was too simplistic to be useful. Dicklyon (talk) 03:51, 27 November 2011 (UTC)

Hello Dick! I was hoping to have you onboard, but I'm a bit disappointed that you too have misunderstood ETTR, or are perhaps conflating the issues. For the remainder of this discussion let's ignore the hard clipping problem, as that is an entirely different matter. Obviously if the whites are clipped they are irredeemable, as are the blacks. What ETTR is about is maximizing the numerical range of the dataset which represents the image without clipping the highlights. And the article is by no means original research - ETTR has been used by digital photographers for many years.

Let's imagine a low-contrast scene in which the light ranges from Zone III to Zone VII (in Adams' system), and let's also imagine taking 2 shots, one at -1 EV and one at +1 EV. (Of course one would normally take a shot between these 2 values (as per the meter) but this is purely for demonstration purposes, right?) Let's also assume that both of these shots fit comfortably within the dynamic range of the camera.

For a 14-bit raw image the data values of the darker shot would range from about 16 to 1023 (~6% of those available) whereas the data values of the lighter shot would range from about 64 to 4095 (~24% of those available). If both shots were then normalized to produce the same looking image before being exported to Photoshop the lighter shot would have 4 times the tonal resolution to play with, and if both shots had their shadows lightened by the same amount you would see a marked difference in quality. This is also true of the lighter tones (if you were to darken them) but, as darkening tends to narrow the data range, it is far less of a problem. Hence my original assertion of highlights (or lighter tones) having more latitude than shadows (or darker tones).

Does this make sense to you? And if not, how might I best demonstrate the phenomena? As a (possible) 'ETTR skeptic' you may be the ideal test subject for me. :-) I think that much of the misunderstandings of ETTR stem from the fact that there isn't yet an accepted universal language for talking about some of these concepts, so we are left quoting numbers that ignore other issues, like noise, or the fact that my imaginary camera has way too many stops. Please let me know what you think and do try to keep an open mind. ETTR is no more or less simplistic and no more or less useful than the Zone System. Thanks for your time. Kind regards, nagualdesign (talk) 06:01, 27 November 2011 (UTC)

I hope that you don't find the above example patronising. I am aware of your expertise in this field, but ultimately I'm trying to write an article that non-experts can understand. Perhaps I'm going round the houses a little. How about this..
From Gamma correction:
"Gamma encoding of images is required to compensate for properties of human vision - to maximize the use of the bits or bandwidth relative to how humans perceive light and color. Human vision under common illumination conditions (not pitch black or blindingly bright) follows an approximate gamma or power function. If images are not gamma encoded, they allocate too many bits or too much bandwidth to highlights that humans cannot differentiate, and too few bits/bandwidth to shadow values that humans are sensitive to and would require more bits/bandwidth to maintain the same visual quality."
From Charles Poynton (2010). Frequently Questioned Answers about Gamma:
"If linear-light coding is used to represent image data, then from 12 to 14 bits are necessary in each component to achieve high-quality image reproduction. With nonlinear (gamma-corrected) coding, 8 or 10 bits suffice."
From Raw image format:
"Raw image formats are intended to reproduce as closely as possible (i.e. at the best of the specific sensor's performance) the sensitometry of the image, that is, physical information about the light intensity and color of the scene."
..So in a way ETTR is a method of 'pseudo-gamma compression' of raw images, which maximizes the bandwidth of digital negatives, giving greater latitude, especially in the shadow values. Given that a great deal of work can be done in Camera Raw (or DPP, etc.) to enhance an image before it is edited in Photoshop, especially with low-contrast scenes or ones with high- and low-key areas but no 'middle', ETTR is a phenomenal boon. But obviously like HDR or photostitching or stacking it is but a single tool. Still, let's not understate it's potential. nagualdesign (talk) 08:59, 27 November 2011 (UTC)
My point (the reason this discussion is taking place here and not at the ETTR article) is that the sentence "Digital should be considered to be the reverse of print film, with a good latitude in the shadow range, and a narrow one in the highlight area." does not apply to digital negatives. Sorry if I wasn't clear in all my waffling! Regards, nagualdesign (talk) 09:03, 27 November 2011 (UTC)

You're doing a lot of talking to yourself here. I find much of it to be "original research", as I said; not that ETTR is original research. The statement "Digital should be considered to be the reverse of print film, with a good latitude in the shadow range, and a narrow one in the highlight area" is fairly accurate, due to the hard clip at the high end and the better-than-film noise characteristics at the low end. Changing from 14-bit linear to 10-bit gamma compressed doesn't change that. The reason I call ETTR too simplistic is that it encourages risking of highlight detail to improve shadow noise; I find that highlight problems are much more common than shadow problems, in any good modern DSLR at least, so I end up more often at -1 EV exposure comp, exposing to the left to get good highlights. Noise is seldom an issue at all at ISO 100, and at any higher setting exposing to the right is just lying to yourself about the ISO. What's the point? Plus, exposing to the right means that you need a longer exposure time or larger aperture, either of which is likely to reduce your image quality by more blurring; total exposure needs to consider how these effects trade off; exposing less is often a way to get a better picture (and the high ISO settings are provided to give you an easy way to do that, taking advantage of digitals great dynamic range at the low end). Also you've misused the term "latitude" where you said "So in a way ETTR is a method of 'pseudo-gamma compression' of raw images, which maximizes the bandwidth of digital negatives, giving greater latitude, especially in the shadow values" – the latitude is a property of the medium, not a property of your choice of what part of that latitude to use for highlights and what part for shadows. And what you mean by "pseudo-gamma compression"; exposing to the right is just a gain change. As for "bandwidth" I can't see how it's very relevant; there are plenty of levels in the 12- or 14-bit linear space or corresponding gamma-compressed space that the data bandwidth is irrelevant. Yes, more exposure results is less noise, but at low ISO it's seldom noticable, and at high ISO it's just an ISO change. So ETTR just seems silly. On the other hand, it was defined almost 10 years ago, when dynamic range of even good cameras was more limited by shadow noise, and it did help some people cope with that. Dicklyon (talk) 14:33, 27 November 2011 (UTC)

I realise that I have been a little verbose, Dick, but with all due respect your response is more combative than constructive. Much of what I said was (digressing) about ETTR, which isn't original research. So how much of "[The sentence in question] does not apply to digital negatives." is OR? It is, after all, a singular statement!
Having had more experience of photo manipulation than taking photographs, I had always used the term 'latitude' to describe the degree by which one can increase/decrease the brightness, contrast or saturation of an image before degradation becomes an issue. This seems to be borne out by Wikipedia's working definition which states: "Latitude is the degree by which one can over, or under expose an image, and still recover an acceptable level of quality from an exposure. Typically negative film has a better ability to record a range of brightness than slide/transparency film or digital." Given that raw images record a much greater range in the lighter areas than darker areas I would state that there is far more latitude in the lighter tones and less in the shadows. And changing from 14-bit linear to 10-bit gamma compressed changes everything.
ETTR does not encourage risking of highlight detail to improve shadow noise at all. It's a way of avoiding posterization in the shadows, and if it is done at the expense of highlight detail you are simply doing it wrong. As for 'bandwidth', there are plenty of levels in the 12- or 14-bit linear space for a normal photograph, but if you intend to edit the image it can be limiting.
I'm not going to push the issue as I know how you like to get your teeth stuck in, ;-) but I would like to encourage you to consider that linear and gamma-compressed images respond very differently to post-processing, and that the sentence in question fails to differenciate between the two. Given that the section is comparing film negatives to 'digital' I think it only fair that it should stick to comparing film with digital negatives. nagualdesign (talk) 00:12, 28 November 2011 (UTC)
If anyone's interested: nagualdesign (talk) 19:31, 28 November 2011 (UTC)

Error in EV interpretation[edit]

I fixed the lead where it said higher EV meant MORE total exposure. Actually, if all you vary is EV, that's backwards, since higher EV means faster shutter or smaller aperture, which gives you LESS exposure. This is a common confusion, since a higher EV is apppropriate when shooting a brighter (higher luminance) scene. See exposure value.

How about a formula for total exposure from EV and scene luminance? Using the formula above for focal-plane illuminance and multiplying by exposure time t we have total exposure:

 E t = {\pi L t \over 4 N^2}

But since {t \over N^2} = 2^{-\mathrm{EV}}, we can write the total exposure as:

 E t = {2^{-\mathrm{EV}} \pi L \over 4}.

That's the total lux seconds, for L in candela per square meter. Dicklyon 22:58, 2 January 2007 (UTC)

Removed Link[edit]

Why was the link to this page removed: ? It nicely shows how exposure varies when ND filters are used.--Tiberius47 02:37, 13 January 2007 (UTC)

Better question might be why was this page added. The same guy added this page to several other articles recently. It's a whole page to help you divide by 2 a few times, and it's on a commericial selling site. Shouldn't we just remove all those? Dicklyon 05:54, 13 January 2007 (UTC)

Correct Exposure[edit]

Can people stop putting "correct exposure" there is no such thing, it's all down to taste Have removed references to correct exposure.. —Preceding unsigned comment added by Barryfitzgerald (talkcontribs) 12:07, 4 January 2009 (UTC)

Sorry, I had to do it! IMO it is worth making a distinction between the meter's indicated exposure and the popular conceit "correct exposure". I hope you see my point. Also, it is possible to get the exposure wrong, especially when dealing with white and black clipping ... --Redbobblehat (talk) 02:20, 12 February 2009 (UTC)
Can anyone think of a way to describe a photograph that is neither overexposed nor underexposed ... which is better than "correctly exposed" ? --Redbobblehat (talk) 13:37, 12 February 2009 (UTC)
  • "Optimum exposure criteria : Before considering methods of measuring subject luminance or illumination to estimate camera exposure, it is useful to consider the criteria by which the correctness of exposure is judged. For black-and white or colour negatives, judgement rests finally on printing quality. So a correctly exposed negative is defined as a negative that will give an excellent print with least difficulty." (Ray 2000 p.311) ... which means - if I've interpreted correctly - the exposure that records the most useful information (ie keeps as much of the subject as possible out of the toe/shoulder ranges), the implication being that 'lossy' artistic effects are done in the darkroom/photoshop.
  • In the context of clever TTL metering systems, he points out "the metering system may give an optimum exposure for the circumstances."(Ray 2000 p.159 - my italics; the context implies caution.) --Redbobblehat (talk) 15:51, 13 February 2009 (UTC)
  • "Conventionally, exposure latitude is expressed as tolerance in stops or exposure values (EV) relative to the optimum exposure. Colour reversal materials usually have a latitude of ± 0.5 EV or less, where ± 1 EV indicates a permissible error of double or half the optimum exposure. ... Ideally, reversal material should be exposed to within 0.3 EV of optimum." (Ray 2000 p.325) --Redbobblehat (talk) 15:51, 13 February 2009 (UTC)
So we could use - and define - "optimum exposure" and "optimally exposed" instead of "correct exposure" ? --Redbobblehat (talk) 15:51, 13 February 2009 (UTC)

Exposure settings[edit]

I've just added a new (extensive) section on exposure settings. My intention was to provide a brief whistle-stop tour highlighting the main issues and referencing other articles (for more expansive information) where I could. In many ways, it's like an inflated disambiguation page, hoping to provide a reasonably brief but helpful context to deeper stuff. It took me ages so please be nice :) --Redbobblehat (talk)

video gain[edit]

I realise that some bits may be contentious - the bit on video gain in particular. I hope you don't find it too objectionable. I worded it very carefully, so please read it carefully before you freak out ;) --Redbobblehat (talk) 02:11, 12 February 2009 (UTC)

Photometric exposure[edit]

IMO this article should offer at least 1 rock solid definition of exposure.

The photometric formula is not really up for debate, and is key to interpreting photographic sensitivity - eg minimum useful exposure (Hm) for ISO 100 = 0.8/100 lux seconds (Ray 2000 p.305). Perhaps it should be in a "photometric exposure" section, rather than the preamble ? I struggled to get the "layman's terms" definition worded carefully so that it is photometrically accurate and localised to photographic terminology ... perhaps in vain ;) ? Also: exposure (H) would include accidental as well as controlled exposure. --Redbobblehat (talk) 13:19, 12 February 2009 (UTC)

"Exposure : When a photograph is taken, light from the various areas of the subject falls on corresponding areas of the film for a set time. The effect produced on the emulsion is, within limits, proportional to the product of the illuminance E and the exposure time t. We express this by the equation
H = Et
Before international standardization of symbols, the equation was E = It (E was exposure, I was illuminance) and this usage is sometimes still found. The SI unit for illuminance is the lux (lx). Hence the exposure is measured in lux seconds (lx s). It should be noted that the lux is defined in terms of the human observer, who cannot see radiation in either the ultraviolet or infrared regions of the electromagnetic spectrum. The inclusion of either of these spectral bands in the desired imaging exposure may therefore yield erroneous results with some films or other imaging systems.
As the luminance of the subject varies from area to area, it follows that the illuminance on the emulsion varies similarly, so that the film receives not one exposure over the entire surface but a varying amount of light energy, i.e. a range of exposures. As a general rule the exposure duration is constant for all areas of the film, variation in exposure over the film being due solely to variation in the illumination that it receives.
It should be noted that the use of the word ‘exposure’ in the sense in which we are using it here is quite different from its everyday use in such phrases as, ‘I gave an exposure of 1/60 second at f/8’. We can avoid confusion by designating the latter camera exposure, as we have already been doing in previous chapters." (Ray 2000 p. 218-9, original italics) --Redbobblehat (talk)
This lot seems to offer a number of succinct definitions ... can we cannibalise it ?--Redbobblehat (talk) 18:31, 13 February 2009 (UTC)
These accord with my understanding. But I don't see how they square with your arguments. Dicklyon (talk) 03:57, 14 February 2009 (UTC)

Visible & invisible light[edit]

Because of the distinction between photometry and radiometry, both photometric exposure and luminous exposure quantify only human-visible light. If there is any debate about this use of photometric I suggest we ditch it and just use luminous exposure. To improve clarity and flow, and avoid getting bogged down in "infra-red herrings" ;-) I've tightened up the definition and added a little note with links to the most useful articles about "invisible" light in photography. I hope this is satisfactory. (incidentally, IR filter needs expanding ... ) --Redbobblehat (talk) 19:53, 14 February 2009 (UTC)

Seems to me that photometric is the more common and precise term; if luminous is mentioned, it should be given as a synonym. JeffConrad (talk) 02:35, 16 February 2009 (UTC)

Why the revert?[edit]

Robbie, you reverted, here, saying that I and ref were incorrect. Explain? Dicklyon (talk) 17:12, 12 February 2009 (UTC)

1. The Ray reference refers to the photometric definition, not your the lux-seconds sentence. The formula's legend makes it clear why H = lux seconds. repetition is redundant clutter.
BTW: AFAIK (H) itself is not an SI unit, so if E & t were given in footcandles and seconds, the value for H would be in footcandle-seconds ... See what I mean? (Jeff pounded that one into my mushy brain ;) --Redbobblehat (talk) 20:56, 12 February 2009 (UTC)
I'm not following your points here. What is lux-seconds if not a photometric measure of exposure? Footcandle is a candela per square foot; lumen is a candela per square meter; let's stick to the modern metric units (mention the alternative, too, if you think it's useful). But what is it that you were saying was incorrect? Dicklyon (talk) 17:09, 13 February 2009 (UTC)
2. "during the process of capturing a photographic image. " is not specific enough; it has very little information value: frankly it's waffle. Sorry to be so blunt :(
Find a better but still correct way to put it? Dicklyon (talk) 17:09, 13 February 2009 (UTC)
3. "Exposure can be computed from exposure value (EV) and scene luminance over a specified area." So what? irrrelevant here. The whole point of the EV system is to compute shutter/aperture settings from scene luminance and sensitivity, not the other way round! Exposure needs to be correct for the film's sensitivity (which ISO defines as 0.8/film speedISO) ... I recently improved a sentence in further down the article (which you inadvertently reverted): "The purpose of exposure adjustment is to control the amount of light reflected (or emitted) from the subject so that it conforms to the the film (or sensor's) sensitivity." Perhaps this would be better up top? I think its relevant only to "exposure settings" ...
That wasn't inadvertant; it seems quite wrong. Dicklyon (talk) 17:09, 13 February 2009 (UTC)
I have a bit of a problem with "(EV) and scene luminance"—it sounds as if they are to be used in combination, which isn't the case. Don't we obtain the former from the latter? Changing "and" to "or" would be a slight improvement, but further clarification would help. JeffConrad (talk) 03:18, 16 February 2009 (UTC)
The point here is that if you know the scene luminance and the f-number you can get the focal-plane illuminance and if you include the shutter speed you can get the photometric (luminous) exposure; the EV gives you the relevant combination of aperture and shutter speed so you don't have to know either explicitly. I can see how it might be confused with going the other way, or using the non-standard definitin of EV, to do something like choose your aperture and shutter speed settings; maybe we can make it more clear. Dicklyon (talk) 03:36, 16 February 2009 (UTC)
It may be true, but I still don't understand what relevance it has here ? Why would you want to know the image plane luminance ? --Redbobblehat (talk) 07:15, 16 February 2009 (UTC)
Indeed, image plane luminance would be pure nonsense; why do you bring it up? Dicklyon (talk) 07:18, 16 February 2009 (UTC)
4. "For HD curves, exposure is conventionally expressed as log10(H), and is plotted against optical density, another logarithmic unit that describes the darkening of the developed film. The exposure value (EV) of the APEX system, is based on "stops" or log2 units, is not directly related, as it measures a combination of exposure time and aperture, independent of the actual amount of light, but the stops can be related to log10 units (e.g. optical density units) via ratio of 3.32 stops per log10 unit.""
4a) The linked HD curve article provides definition & explaination: it is not relevant enough to repeat here.
4b. The linked exposure value article does likewise - it is not relevant here.
4c. The relationship between log2(H) and log10(H) is a mathematical fact. It has nothing to do with stops. I provided the info because it always comes up when interpreting HD-curves, however it is not part of HD-curve 'spec' nor EV 'spec'. It is to do how exposure (H) is commonly represented in 2 different contexts. An EV ≠ log2(H), but the EV system is based on a log2(H) scale. IMO that statement is sufficiently correct and concise for this context.
I was trying to take what you written about the different log units and make it correct. You had said "whereas the exposure value system is based on log2(H);" but there's no such system; and EV in the APEX system doesn't depend on H. Maybe you can work out a better correct way to get your point across (or maybe this stuff about different log units just doesn't belong here). Dicklyon (talk) 17:09, 13 February 2009 (UTC)
Please also note the talk section entitled "#photometric exposure". Perhaps it is unwise to try and combine photometric exposure with the article introduction? Perhaps p-e should be section no.1 ? Perhaps no "general introduction" is necessary, except perhaps disambig ?
Isn't this an article on photometric exposure? Or what is the distinction you are trying to draw? Dicklyon (talk) 17:09, 13 February 2009 (UTC)
D, please use talk page before overwriting my contributions; I do not want an edit war. --Redbobblehat (talk) 20:25, 12 February 2009 (UTC)
OK, but work with me when I try to repair errors. Dicklyon (talk) 17:09, 13 February 2009 (UTC)

The section Photometric and radiometric exposure seems a bit heavy on radiometric for a basic article such as this. It's obviously something of interest to a sensor designer, but does it matter to the average photographer? In any event, I find the current wording a bit hard to follow.
I agree with Dick that APEX and EV have nothing to do with H; I've made a slight revision to make this more clear. I've also cleaned up the formula for conversion; like Dick, I prefer 3.32 because it's more accurate and consistent with most sources I've seen, but I've left it at 3.3 for now. I don't even like just giving the number, because the reader must do some work to see that
\log_2 H = \frac {\log_{10} H} {\log_{10} 2} \,;
perhaps the exact relationship could be given in a note.JeffConrad (talk) 03:12, 16 February 2009 (UTC)
I'm happy with jeff's inline version. The math format formula looks like overkill to me : I only put it in originally as a little mnemonic tip for reading HD curves; referring to EV but not trying to define EV or APEX. If it needs reams of footnotes to make it stand up I would rather cut it altogether and keep the main article flowing. BTW: if EV has nothing to do with H, why are they both called 'exposure' values ;-? --Redbobblehat (talk) 07:04, 16 February 2009 (UTC)
Have you read the Exposure value article? The standard definition depends on just your camera settings, independent of scene luminance; but it is also used as a stand-in for scene luminance, non-standardly, by representing the camera settings that would give a nominally correct exposure at ISO 100. Dicklyon (talk) 07:14, 16 February 2009 (UTC)
I tend to agree with Redbobblehat about keeping the units out of the definitions of H and the other terms; the equation is valid for any consistent system of units (such as SI). I'll concede, however, that I can't recall seeing exposure given in units other than SI. JeffConrad (talk) 03:12, 16 February 2009 (UTC)
Actually I wasn't so sure after "(Ray 2000 p.218) also observes that before SI units the equation was E = It ( where E was exposure, I was illuminance)." ... kind of implies that if it wasn't in SI units, it wouldn't be called H :-( ... maybe this article isn't really the best place for trying to make The Ultimate and Transcendent Definition of Photometric Exposure" ? ;-) --Redbobblehat (talk) 06:45, 16 February 2009 (UTC)
That source wasn't talking about SI units, just about the terminology, H versus I; I don't think it's worth mentioning. Dicklyon (talk) 07:00, 16 February 2009 (UTC)
Yes perhaps too much on radiometric; but R introduced spectral sensitivity and I'll tried to make something correct of it. It's OK to omit units; in fact, the various weightings can be unified that way. Dicklyon (talk) 03:40, 16 February 2009 (UTC)
OK. We're all agreed then? So can I take out all the radiometric stuff and just leave this sentence as a pointer ? :
will that make us all happy ?--Redbobblehat (talk) 06:30, 16 February 2009 (UTC)
It's not clear that it should be entirely removed, or that the "invisible" stuff should be left. Whatever we do, it should accord with reliable sources. Jeff, what do you think? Dicklyon (talk) 06:34, 16 February 2009 (UTC)
  • "The section Photometric and radiometric exposure seems a bit heavy on radiometric for a basic article such as this. It's obviously something of interest to a sensor designer, but does it matter to the average photographer? In any event, I find the current wording a bit hard to follow. ... JeffConrad (talk) 03:12, 16 February 2009 (UTC)" ... "Yes perhaps too much on radiometric; but R introduced spectral sensitivity and I'll tried to make something correct of it. It's OK to omit units; in fact, the various weightings can be unified that way. Dicklyon (talk) 03:40, 16 February 2009 (UTC)"
Any chance of a decision on this ? --Redbobblehat (talk) 21:31, 26 February 2009 (UTC)
I think it's not bad; it would take some skill to shorten it without losing important sourced info. Dicklyon (talk) 04:34, 27 February 2009 (UTC)
IMO, the questions are about the level of detail required in this article:
1: Given the inline refs to photometry, lux and visible light, do we need to further emphasise that we're talking about a limited spectral range ... ?
2: If so, what's the best way to do it?
Personally I favour the minimalist approach. Like Jeff, I see this article as a fairly introductory overview of a complex subject. "Common" photography and photographic material data seem to assume the photographer wants to work with visible light, and "invisible" light photography would be considered a more specialist and esoteric variation on this theme. IMO, a section on "radiometric exposure units" would be more useful and intelligible in the context of a more specialist article such as full spectrum photography or infrared photography, or perhaps lux seconds deserves its own page?. To me it seems sufficient here to simply provide a nod in the direction of those articles. --Redbobblehat (talk) 14:26, 28 February 2009 (UTC)
The "invisbible" stuff is a red herring; who put that in there? The important issue here is that the spectral sensitivity of film is nowhere close to that of the eye, as represented by the luminous efficiency function, so the specification of film darkening in terms of photometric exposure is not really correct, or only correct given a fixed spectrum. Other point is that radiometric units are also sometimes used; same problem there, lacking a filter to model the film response. Dicklyon (talk) 22:56, 28 February 2009 (UTC)


And what's up with the Ray 2000? Looks like Attridge wrote that chapter; I fixed the ref in one place, but the "ibid" made less sense, so I took it out. Dicklyon (talk) 06:57, 16 February 2009 (UTC)

Well, "ibid" made sense before someone edited the preceding reference to that book ... I've reformatted and replaced the ref, and found a page where the term "log exposure" and discussion of log10(H) coincide. Now it should be foolproof? IMO the term "log exposure" is unusual and deserves a source ref. "log10(H)" can be found on any characteristic curve, although often abbreviated to "log(H)" which can seem ambiguous when thinking in base-2 log scales. --Redbobblehat (talk) 21:31, 26 February 2009 (UTC)
Ibid makes sense on paper, where things don't move; not in a wiki. Dicklyon (talk) 04:31, 27 February 2009 (UTC)
Good point, well made :-) BTW Nice cite book template too ... but copy & pasting it for every page reference seems a bit clumsy... could we justify a special template just for the Manual ... with just the page number as a parameter ? --Redbobblehat (talk) 14:26, 28 February 2009 (UTC)
You can name the ref, and put a page range for the whole chapter; or other ways to do this; Jeff will know. Dicklyon (talk) 22:57, 28 February 2009 (UTC)

Bad image and explanation[edit]

Robbie, your "HDR example" image is not a great one, as it is specifically an example of a scene for which no exposure can be considered correct. Find a more normal scene and make that kind of example. Dicklyon (talk) 17:15, 12 February 2009 (UTC)

Ah I just realised I'm Robbie! IMO The HDR image is eminently suited to illustrate the sentence "A photograph may be described as overexposed when the subject tones seem too light, or underexposed if the subject seems too dark. Underexposure may cause loss of shadow detail, also known as crushed blacks, and overexposure may cause loss of highlight detail, also known as blown-highlights." What the file happens to be called or the fact that it's a CGI is irrelevant. I really can't understand your objection to this image. --Redbobblehat (talk) 21:14, 12 February 2009 (UTC)
Sorry, I misrecalled your name. Anyway, several problems; if the image is CGI, it doesn't even have an exposure. But more to the point, the "correctly" exposed one also shows crushed shadows and saturated highlights. Dicklyon (talk) 06:36, 13 February 2009 (UTC)
np. It seems to me that the CGI has been carefully created to 'simulate' all the under/over exposure effects we're talking about. Think of it as a graphic illustration rather than a photographic example ;) I hadn't looked for crushed blacks and blown highlight in the central image, but its good they are there - we can use it to illustrate the effects of limited exposure range/latitude too (which of course is the raison d'etre for the whole HDR thing...). Ray uses the term "optimum exposure" instead of "correct" - would that help allay your concerns? ... I am assuming of course that you don't have in mind an even better image we could use for this ? --Redbobblehat (talk) 13:51, 13 February 2009 (UTC)
There's one in commons called Exposure.jpg (but another in en. that shadows it; how we specify commons?). Dicklyon (talk) 16:59, 13 February 2009 (UTC)

And you text "The purpose of exposure adjustment is to control the amount of light reflected (or emitted) from the subject so that it conforms to the the film (or sensor's) sensitivity" is misleading at best; the amount of light reflected or emitted is not affected; what affectes is the amount allowed to fall on the film, as already stated. So I'll back out this change, and the other revert (see above) until you explain it. Dicklyon (talk) 17:15, 12 February 2009 (UTC)

Not if you read the subsequent 2 sentences: "Often the most obvious methods, such as simply adding or avoiding local lighting, or using neutral density filters are overlooked. Choosing a film stock or camera sensor with a light sensitivity appropriate to the task can be a vital first step." ... ? --Redbobblehat (talk) 21:14, 12 February 2009 (UTC)
That seems to be confusing lighting and film selection with exposure; we should clean that up. Dicklyon (talk) 16:54, 13 February 2009 (UTC)
lol! no! :) The whole point is that "lighting and film selection" are inseparable from "exposure"! If you don't recognise that fundamental physical connection, you won't understand the mechanics of photographic exposure at all. I'm sorry mate but the confusion is yours. I'm going to revert your reverts, then I will create a separate photometric exposure section, and restore the 'old' article introduction. I hope that will satisfy all parties ? --Redbobblehat (talk) 17:19, 13 February 2009 (UTC)
Wouldn't it be more polite to ask for clarification if something doesn't read well to you, rather than reverting contribs and demanding an explanation ? I would be much happier if you reverted your reverts (sic!) so we can make improvements amicably. --Redbobblehat (talk) 21:32, 12 February 2009 (UTC)

I don't understand[edit]

R, your revert edit summary said "D you a really beginning to annoy me now. Please use talk if you don't understand something." I don't understand that. Are you sugggesting that my corrections were based on a misunderstanding? Please explain. Are you not aware that photometry is based on energy wavelength weighted by the luminosity function, and that the effective stimulus for films needs a non-photometric spectral sensitivity weighting? Dicklyon (talk) 23:28, 14 February 2009 (UTC)

Perhaps you should include that in the photometry or photographic film article ? --Redbobblehat (talk) 02:59, 15 February 2009 (UTC)
I'm not sure how you mean; the photometry article doesn't need to mention all the non-photometric ways to measure light, does it? The film article already talks about spectral sensitivity and notes that "Film has a number of disadvantages as a scientific detector: it is difficult to calibrate for photometry..." Dicklyon (talk) 04:02, 15 February 2009 (UTC)

I extended the section to explicitly include the alternative, as "Photometric and radiometric exposure", with sources and better explanations. Dicklyon (talk) 04:46, 15 February 2009 (UTC)

As I already pointed out under #visible & invisible light (above), I do not think that a whole lot of "whereas in radiometry ..." caveats is necessary or helpful in this article. IMO they clutter up the page with superfluous facticles, disrupting the flow of the article to the point of unintelligibility. Referenced articles such as photometric, lux already contain the information you are touting. Wouldn't your deductions be more usefully appended to more specialist articles such as infra red photography or full spectrum photography ? IMO you need to justify inclusion here. --Redbobblehat (talk) 17:21, 15 February 2009 (UTC)

"Exposure settings" section adjustments[edit]

R, in this edit, I attempt to make your recent commentary on exposure settings more correct as well as more concise. I think you'll see that each bit I took out is either sort of "advice" or "off topic". And the corrections and clarifications make it more correct; e.g. the bit about putting gain into the capture path, which really can help the sensitivity as defined via a threshold SNR criterion. Dicklyon (talk) 05:19, 15 February 2009 (UTC)

video gain vs other exposure settings[edit]

Original : Video and electronic still cameras provide the opportunity of applying gain to the electronic signal from the sensor. Technically this does not increase exposure or sensitivity; it is an image processing enhancement applied to the photograph. Nevertheless, there are practical advantages to applying gain during image capture rather than afterwards, and it has a genuine part to play in determining the exposure setting.
Your version: Video and electronic still cameras provide the opportunity of applying gain to the electronic signal from the sensor. Technically this does not increase exposure, but applying gain during image capture (before quantization) can increase sensitivity.
The main point of disagreement seems to be whether "gain" increases "sensitivity". My point is that applying gain (amplification) to an electronic signal does not alter the signal-to-noise ratio; it amplifies both signal and noise. The minimum useful exposure for a minimum SNR defines the sensitivity of a video sensor. This is sometimes specified as the sensor's "native sensitivity", precisely because it is unaffected by gain. An analog analogy: the quality of an underexposed film negative itself cannot improved by 'overexposing' the print; the print may be improved (hopefully), but the negative remains underexposed. ... If I have failed to explain it to your satisfaction, try googling "native sensitivity" or just read some of these : [1], [2], [3]. --Redbobblehat (talk) 16:58, 15 February 2009 (UTC)
The "advantages" I had in mind were 1) WYSIWIG user preview - so you know if you got the shot or need a retake (this reasoning applies to all "in-camera" image processing) and 2) the marginal advantage of applying basic image processing before compression (be it quantisation, chroma subsampling, or whatever).
However, the technical details of this subject belong properly in the image processing (or similar) article, not the "exposure (photography)" article. I specifically invited discussion under the #video gain section of this talk page to protect the article itself from the inevitable barrage of misinformed opinions. --Redbobblehat (talk) 16:58, 15 February 2009 (UTC)
And my point was that your point that "it amplifies both signal and noise" is in many cases not quite true, since some of the noise comes after the amplification; quantization noise in particular. This is why some cameras do high ISO by applying more gain early; it actually can improve the sensitivity. Not all cameras do it this way, since if you have a clean enough ADC with enough bits, there will be no signficant win to putting more gain before it. Dicklyon (talk) 18:06, 15 February 2009 (UTC)
The issue is sensitometric sensitivity as it relates to "exposure (photography)", not the efficiency of the entire image processing pipeline. --Redbobblehat (talk) 18:47, 15 February 2009 (UTC)
I'm unclear on exactly what you mean by that distinction or why you think it negates my point. Dicklyon (talk) 19:35, 15 February 2009 (UTC)
Probably it would be best to just remove the part about gain, unless we find something sourced about it with respect to exposure and/or sensitivity. Dicklyon (talk) 19:45, 15 February 2009 (UTC)
How about including this in the "Exposure settings" section ? :
  • Video and 'digital still' cameras provide the opportunity of applying gain to the electronic signal from the sensor. Technically this does not increase exposure or sensitivity; it is an image processing enhancement applied after the fact [4] [5] [6]. Nevertheless, there are practical advantages (eg. WYSIWYG) to applying gain during image capture rather than afterwards, and therefore it has a legitimate role in determining the exposure setting.
--Redbobblehat (talk) 20:31, 15 February 2009 (UTC)
Normal photography also has "gains" of sort, at various stages, involving film selection, development, printing, etc., to determine how the reproduction relates to the scene. In the film case, several of these (film selection and development) affect the sensitivity; similarly in digital or video capture, gains early in the system will afffect sensitivity, when gains later will not. Be careful of blanket statements that represent one simple model as true. Dicklyon (talk) 00:18, 16 February 2009 (UTC)
So I can go ahead and put it back then ? --Redbobblehat (talk) 06:17, 16 February 2009 (UTC)

Purpose of exposure adjustment[edit]

Following the discussion under #Bad image and explanation, your edit (diff[7]) still seems more confused than the original:

  • (original)The purpose of exposure adjustment is to control the amount of light reflected (or emitted) from the subject so that it conforms to the the film (or sensor's) exposure sensitivity. Often the most obvious methods, such as simply adding or avoiding local lighting, or using neutral density filters are overlooked. Choosing a film stock or camera sensor with a light sensitivity appropriate to the task can be a vital first step.
  • (your version)The purpose of exposure adjustment (in combination with lighting adjustment) is to control the amount of light from the subject that is allowed to fall on the film, so that it falls into an appropriate region of the film's characteristic curve.

I was considering changing "exposure sensitivity" to "exposure responsivity", but I'm not 100% sure that would be any more correct or helpful. However, I am quite certain that any light falling on the "characteristic curve" (which is a type of diagram), will not make any difference to the photograph (which is another type of diagram altogether). Your version omits the direct reference to the sensitivity (or responsivity) of the film being used, which I believe is an absolutely essential part of the exposure equation.

I find both versions of "The purpose of exposure adjustment ..." confusing. Is "exposure adjustment" the same as exposure compensation? If not, how does it differ? JeffConrad (talk) 02:47, 16 February 2009 (UTC)
Yes, it's very confusing, between the concept of adjusting the camera's exposure settings and adjusing things like lighting. We need to straighten it out. Dicklyon (talk) 03:43, 16 February 2009 (UTC)
What I'm trying to say is so simple & fundamental. Maybe it's so obvious that you guys are looking too hard to see it, maybe I've not expressed it well enough .... let's call your aperture and shutter your "exposure settings". Imagine you're setting up a studio shot and discover that your aperture and your shutter are jammed at f/2 and 1/2 sec, which will overexpose your ISO 800 film by 5 stops. You can't fix them, you don't have any spares, etc. What else can you change to get the right exposure ? answer: reduce the lighting levels and/or use a less sensitive film... At a really basic level, the task is to somehow balance the scene luminance with the film speed, and that task I called "exposure adjustment". Changing your "exposure settings" is not the only way of getting the right exposure. --Redbobblehat (talk) 06:07, 16 February 2009 (UTC)
But you're mixing up at least a couple of things. Changing your film speed isn't going to change the exposure, but I agree that changing the lighting (scene luminance) will. Dicklyon (talk) 06:15, 16 February 2009 (UTC)

Defining over and under exposure[edit]

I took out this definition: "A photograph may be described as overexposed when the subject tones seem too light, or underexposed if the subject seems too dark." It's really not at all clear what is meant by how bright the subject seems, but if it refers to the photographic reproduction, then this is a common misconception; it's very often the case that a too-light or too-dark photo was exposed just fine, but printed wrong (or rendered wrong to an image file). A common and more useful definition of over and under exposure is in terms of the available shadow and highlight detail (in the negative, or in the raw file, in particular). I added some sources about the relevant terminology. Dicklyon (talk) 06:34, 15 February 2009 (UTC)

"Correct" exposure may be defined as an exposure that achieves the effect the photographer intended*. A photograph may be described as overexposed when the subject tones seem too light, or underexposed if the subject seems too dark. Underexposure may cause loss of shadow detail, also known as crushed blacks, and overexposure may cause loss of highlight detail, also known as blown highlights.
The paragraph (above) attempted to summarise both the qualitative and quantative use of the terms underexposure and overexposure, without getting bogged down in the tired old subjective/objective POV debate. It was not a definition but an observation of how the terms are frequently used with significantly different meanings (See private language argument). I had hoped that a reasonably intelligent reader would accept this summary without debating which POV is the correct definition.--Redbobblehat (talk) 15:33, 15 February 2009 (UTC)
I found it to be misleading and incorrect as stated, but if you find a source that interpretation it would be OK to include it. Dicklyon (talk) 18:03, 15 February 2009 (UTC)
Please specify in what ways it is incorrect and misleading, so I can try to improve it. As far as I can see your changes to the original are entirely superficial, and improve nothing. --Redbobblehat (talk) 21:16, 15 February 2009 (UTC)
It may well be that under&overexposure deserve a fuller treatment than this, but under a different section of the article. I was considering how best to go about this because it is the same territory as the "latitude", "highlights" and "blacks" sections ...--Redbobblehat (talk) 21:16, 15 February 2009 (UTC)
Incorrect where you said "A photograph may be described as overexposed when the subject tones seem too light, or underexposed if the subject seems too dark." Too light and too dark ain't what it's about. Dicklyon (talk) 03:46, 16 February 2009 (UTC)


Several pictures used by Wikipedia give the exposure time as 0.016638935108153 seconds. This is what Nikon are claiming. I do not think that this is a realistic value. To measure to the nearest femtosecond. —Preceding unsigned comment added by (talk) 14:56, 23 April 2010 (UTC)

The cameras are shooting at 1/60. For some reason the mediawiki software is reading it as 10/601 and when converted to decimal format you get that long number. Nikon isn't doing anything wrong. gh5046 (talk) 16:14, 23 April 2010 (UTC)

File:Clifton Beach 5.jpg to appear as POTD soon[edit]

Hello! This is a note to let the editors of this article know that File:Clifton Beach 5.jpg will be appearing as picture of the day on June 13, 2011. You can view and edit the POTD blurb at Template:POTD/2011-06-13. If this article needs any attention or maintenance, it would be preferable if that could be done before its appearance on the Main Page so Wikipedia doesn't look bad. :) Thanks! howcheng {chat} 19:34, 12 June 2011 (UTC)

Picture of the day
Long-exposure photograph

A long-exposure seascape photograph of rocks at Clifton Beach, Tasmania. In photography, exposure is the total amount of light allowed to fall on the photographic medium (film or sensor): the longer the shutter speed, the more light is let in. This can be done for technical reasons, such as in low-light conditions, or to create an artistic effect as shown here, when the ocean waves appear to be fog.

Photo: JJ Harrison
ArchiveMore featured pictures...

A common misconception: Exposure is not equal to total amount of light[edit]

Exposure is not the total amount of light that falls on a photographic medium but the amount of light per area unit (lm*s per m2). Therefore the top entry on this talk page should be removed or changed accordingly. There also should be an addition to the article that there exists a common misconception that exposure is equal with the total amount of light. After all it took nine years for this basic mistake in the introductory sentence of the article to be noticed at all! The total amount of light is dependent on the area of the photographic medium. A larger photographic medium will have to be hit by a larger total amount of light to receive same exposure. At same aperture and shutter speed, FOV being the same, a larger photographic medium will have to receive a larger total amount of light for the same exposure. — Preceding unsigned comment added by (talk) 16:21, 26 July 2012 (UTC)

It doesn't say total; lux, lux seconds, illuminance, are all per area measures of light. But they're not constants over a scene, so you need to specify where you're measuring it. Dicklyon (talk) 16:51, 26 July 2012 (UTC)

1. Region: oops you are right. Proposal: [...] Exposure is luminous flux density averaged over a limited amount of time, measured in lux seconds. For each region in a photographic scene, it can be computed from exposure value chosen on the camera and luminance measured in that region.

2. On top of the talk page there is still a talk entry about total. (Maybe I am mistaken about the editing procedure for talk pages)

3. Since 2003 until today (2012) this Wikipedia article said total. On DPreview it says " The exposure is the amount of light received by the film or sensor and is determined by how wide you open the lens diaphragm (aperture) and by how long you keep the film or sensor exposed (shutterspeed). ", which is incorrect in the first part (amount not per area) and unclear in the second part. By aperture, do they mean relative aperture (f-stop) or absolute aperture (diameter)?. Most DPreview readers might assume F-stop. It should be easy to find more examples of such incorrect statements about exposure. Is this sufficient reason to include a clarificaton section in the WikiPedia article? (talk) 19:47, 26 July 2012 (UTC)

I just saw this page cited on DPReview here ( as 'evidence' that exposure is 'total light', not 'light per unit area'. Obviously the text as is is misleading some people, so I made an edit to the definition to make it clear what it actually is. It might be said that's clear from the lux seconds unit, but many readers can't get that deep - so best to be clear.Bobn2 (talk) 21:22, 13 March 2014 (UTC)


The usage of Overexposed is up for discussion, see Talk:Overexposed (album) -- (talk) 21:41, 4 December 2012 (UTC)

Exposure Triangle?[edit]

There is no mention of the misnamed 'Exposure Triangle' as an aid to quickly estimate what happens to output level (a proxy for brightness) relative to a given starting situation when changing Exposure and Sensitivity parameters in the field. The link at the bottom of the article is well done, but unfortunately confuses Exposure and Output Level. I understand that this would introduce Sensitivity to the article, which understandably appears to have been avoided like the plague, but I think explaining its indicative nature clearly here would help dispel a number of doubts. Jack Hogan (talk) 07:20, 22 April 2013 (UTC)

I removed the link as the Exposure triagle article is also wrong in content - for example it claims tha sensor sensitivity changes with ISO which is not the case. Additionally it creates confusion regarding the definition of exposure. -- (talk) 09:50, 21 September 2013 (UTC)