Talk:Film speed

Hurter & Driffield system bug

It says "In their system, speed numbers were inversely proportional to the exposure required. For example, an emulsion rated at 25 H&D would require ten times the exposure of an emulsion rated at 2500 H&D." Does is mean the inverse square of the exposure required? Or should the 2500 be 250? Can someone check the Sowerby ref? Dicklyon (talk) 06:53, 1 August 2011 (UTC)

Since I added that information, that'll be me. Unfortunately my scanner is currently broken and I can't scan the referenced 'Fig.3' at this time. The entire paragraph (H & D Speeds, p.583) sayeth:

The first scientific method of measuring emulsion speed was that of Hurter and Driffield. They ignored any effect of the printing paper and laid down that the minimum correct exposure was the least that would give accurate tone-reproduction in the negative. They based their film-speed, fundamentally at least, on the point A of Fig. 3. (My note: Fig. 3 is a graph showing negative density over exposure time). But partly owing to the difficulty in determining the exact point at which the 'toe' of the curve began, and partly to find a speed-value that did not greatly depend on gamma, they continued the straight line part of the curve backwards as shown dotted in Fig. 3, and took as a measure of the H & D speed the the exposure indicated at the point where this extension cut the line indicating fog-level. This exposure they called the "inertia"; the speed -number was found by dividing the value of the inertia by 34. Thus the H & D speed numbers are inversely proportional to the exposure required; a plate of speed 250 H & D requires ten times the exposure needed by a plate of 2500 H & D.

I hope that's useful. Baffle gab1978 (talk) 18:15, 1 August 2011 (UTC)

Yes, thanks. I have made the indicated correction (changed 25 to 250). Dicklyon (talk) 18:22, 1 August 2011 (UTC)

250 is correct - my typo on the article! :-) Baffle gab1978 (talk) 18:36, 1 August 2011 (UTC)

• Hi from Russia! I'm editing this page in Russia Wiki and I can't find some information. Does anybody know what else countries except SU used H&D system? Does the USA used it? Unfortunately, our domestic sources of that historical period are not credible for obvious reasons. And, if you interested, I have a link for Soviet photo book with the table of GOST into H&D units accordance. Runner1616 (talk) 08:17, 10 November 2012 (UTC)

H&D speed accordance is in the table 32 (Таблица 32) Runner1616 (talk) 08:32, 10 November 2012 (UTC)

Photons/area at saturation?

It seems that film speed, at least for an ideal linear sensor, is a measure of how many photons (of a given wavelength) per sensor area it takes to saturate the sensor. Is that right? That seems like the most natural units in which to reason about film speed and noise given that an ideal sensor will still have shot noise, and so the number of photons matters. Beyond that you do have quantum efficiency and pixel size (or grain size) which will affect how noisy an image appears, but isn't photons/area at a wavelength at saturation really the fundamental measure of speed for an ideal sensor? If so, it would be nice to see that on this page. It seems like an approachable mental model. As it is, the article goes on and on about arcane systems without giving a straightforward definition. —Ben FrantzDale (talk) 11:57, 12 August 2011 (UTC)

Film (and sensor) speed ratings are for the use of the photographer, to obtain desired exposures. ISO 12232:1998 contains a saturation-based method of determining ISO for digital cameras, and that option was carried over to ISO 12232:2006. However, it's rarely used because it doesn't mean much to the photographer, and it's based on the camera-generated sRGB TIFF file, not the behavior of the sensor itself. Early on, Kodak did attempt to adapt the older ISO specifications for use on the sensor in isolation: http://www.kodak.com/global/plugins/acrobat/en/business/ISS/supportdocs/ISOMeasurements.pdf ... I should note that the current SOS technique is essentially saturation-based, giving an ISO that's 0.704 times the ISO obtained by the older saturation-based technique; other than the multiplier, the big difference is that SOS can be applied to JPEG output files as well as TIFF.

As for a straightforward definition of "digital ISO," I'd suggest the section entitled "The ISO 12232:2006 standard". The reality is that today, digital camera ISO values are almost universally determined using the REI technique, which is basically "whatever the manufacturer thinks gives good results." Doug Pardee (talk) 15:38, 9 October 2011 (UTC)

Mostly true, but the SOS method is not saturation based. It is used to define how the output image brightness relates to the ISO EI setting on the camera; it's not a sensor or a camera spec, since the sensor and camera have many different ISO settings, and the SOS definition does not constrain their range. Dicklyon (talk) 00:05, 10 October 2011 (UTC)

Interesting. So looking at the formula on page 2 of the Kodak PDF,

$\mathrm {ISO} ={\frac {15.4N^{2}}{L\cdot t}}$,

where N is the f number and L is the luminance of an 18% reflector in the scene, and I think t is the exposure time to achieve 18/106% saturation. Then $L\cdot t$ is a measure of luminous energy per area per solid angle. The numerator is dimensionless so the units of ISO in this expression are area times solid angle per luminous energy, I think. If it weren't 8 in the morning and I used photometric units more, I could make sense of that. But it seems like that plus the photometry curves should be able to work out to at least a ballpark value for the conversion between ISO and photons/area at saturation. This also seems to give a conversion for using a camera as a crude light meter (assuming the manufacturer was honest):

$L={\frac {1.54N^{2}}{t\cdot \mathrm {ISO} }}$

Does that sound right? —Ben FrantzDale (talk) 12:47, 12 October 2011 (UTC)

Alternately, it seems like the sunny 16 rule could be inverted to compute an ISO:

$\mathrm {ISO} \approx {\frac {1\,\mathrm {s} }{\text{proper exposure time of sunny scene at f/16}}}$

With a precise definition of "proper exposure time", this should be pretty precise. That is, use a prescribed gray card and say that it should be exposed to some percent of saturation. This could also be pushed through to find photons/area/second at saturation. Thoughts? —Ben FrantzDale (talk) 13:33, 7 May 2013 (UTC)

Yes, when I was an image sensor designer we explicitly used photons per area in our sensor noise models and sensitivity calculations. But the ISO rating is usually defined based on noise, not on saturation. So you can actually calculate the maximum ISO rating for a given pixel size with 100% quantum efficiency and no readout noise (you have to assume a light spectrum, too, which may be affected by the filters used, if any). Probably some of the books I showed in PhotoTechEDU number 5 cover this well. Dicklyon (talk) 22:16, 7 May 2013 (UTC)

Hi again, Dicklyon. Wait, are you saying that by measuring ISO in terms of noise that if I have a sensor that has an extra-noisy ADC you would measure it as having a higher ISO? That doesn't seem right. (Incidentally, my interest in this at the moment is in comparing CCD/CMOS image sensors to my DSLR, just to get a ballpark sense of a sensor's performance compared to a high-production high-quality sensor. For many sensor chips I know QE*FF and can compute noise and I am interested in comparing that to an SLR at, e.g., ISO 100.) Thanks. —Ben FrantzDale (talk) 13:10, 9 May 2013 (UTC)

A noisier sensor needs more signal to get up to the criterion SNR, so it has a lower ISO rating (not higher). When you set the ISO setting on your DSLR, that bears no relation to either the noise-based or saturation-based ISO speed definitions; rather, it controls something called "standard output sensitivity" iirc. Dicklyon (talk) 00:41, 10 May 2013 (UTC)

Can we have an article without film?

Film is dead from a perspective of most people. People want to find out what ISO means from a digital perspective and don't care about film. Any suggestions how to rearrange things so people don't have to go halfway down the article to find the information they really wanted? Daniel.Cardenas (talk) 20:13, 17 February 2008 (UTC)

Film is far from dead, Daniel. — Preceding unsigned comment added by 67.253.117.64 (talk) 18:18, 12 June 2011 (UTC)

So, let me get this straight. You're complaining that the "film speed" article has information about... film speed? And that this is before an explanation of how digital cameras emulate the film-based standard? Just mind boggling. --Imroy (talk) 17:46, 13 June 2011 (UTC)

Can we get an article that doesn't start out about film? I want to read about ISO on digital cameras as do most people. They don't want to be bothered about film that has been dead to most people for 10 years. Daniel.Cardenas (talk) 19:33, 7 December 2013 (UTC)

Daniel, two years ago I thought about splitting the article to separate film speed and digital sensitivity articles; consensus came up against this idea—see the discussion here. Strange you should say that film is "dead"; that's certainly not my perception. I wonder how you know what "most people" want to read about. Cheers, Baffle gab1978 (talk) 17:59, 8 December 2013 (UTC)

Dead to most people. Do think most people consider film? If you mention film to a young adult they will think you are talking about some slimy substance. :)

If splitting the article is less than ideal than perhaps putting the digital aspect more prominently would be better solution. Rename the article, ISO / film speed?, and put the digital stuff in the intro. Daniel.Cardenas (talk) 18:20, 8 December 2013 (UTC)

Yes I agree there should be some digital info in the header; which should be a summary of the main points of the article, per WP:LEAD. I may get around to doing a full c/e one day (but not soon), but in the meantime you can always fix it yourself. I find your sweeping assertion about the ignorance of young people rather amusing. Cheers, Baffle gab1978 (talk) 23:06, 8 December 2013 (UTC)

The description of historical systems is mostly "dead", even to those of us that still use film. I would have no problem with moving most of that section to a separate "historical film speed systems" article. Just leave DIN, ASA, ISO, and perhaps GOST. And the conversion table is large while serving little purpose.

I like the idea of moving most of the digital stuff into a separate article too and have ISO (disambiguation) point to it so people can find out what this "ISO" thing is. And have the two articles link to each other for historical completeness. --Imroy (talk) 06:33, 9 December 2013 (UTC)

I don't agree with splitting out any of the film speed related stuff (in fact, there's quite a bit more, that could and should be added).

If anything, the contents of chapter 4 ("digital camera ISO speed and exposure index") is the only stuff that does not directly belong into an article about "film speed". So, if this article would become too long somewhen in the future (it isn't too long at present), this digital camera related stuff is the info, that could be moved elsewhere, not the film-related stuff. --Matthiaspaul (talk) 16:49, 3 June 2015 (UTC)

ISO Examples

Just now, experimenting with the same shot with a digital camera at different ISO settings, I have a series of five pictures with the ISO settings captioned in the corner.

Would it be a good idea for me to add them to the bottom for a visual example?

PS: Here's what they look like http://youngwilliamthe.tumblr.com/post/83167569926/realizing-i-had-no-clue-what-iso-settings-do Youngwilliam (talk) 04:22, 19 April 2014 (UTC)

Please don't. Get a clue first. If you just change the ISO setting, but leave the exposure time and aperture fixed, which is what it appears you did, then it will be pretty much like adjusting the white level in Photoshop: just a gain variation. But that's not how the ISO setting is normally used. Usually, you'd set the exposure to make the image as bright or dark as you want, for whatever ISO setting you're using; or let auto-exposure set it for you. Then the images would all look nearly alike, except that high ISO ones would be made from many fewer photons, so would have more noise, which you would not notice in small thumbnails like those anyway. Dicklyon (talk) 04:48, 19 April 2014 (UTC)

ps. There's no such thing as ISO 159. Dicklyon (talk) 04:49, 19 April 2014 (UTC)

Okeydoke! I was basing the 159 off of what what Flickr was reporting post-upload. I was guessing it wasn't such an odd number, either, but was instead their setting detection trying to make the best of the default setting. Youngwilliam (talk) 04:52, 19 April 2014 (UTC)

How odd. Also odd is the fact the ISO is only defined for a discrete set of values; 160 is one and 159 is not. Dicklyon (talk) 04:54, 19 April 2014 (UTC)

As for "what ISO does" on a digital camera, the section you want is Film speed#Standard output sensitivity (SOS). It tells you (inversely) how many lux seconds of light the sensor needs to make a medium gray, and it adjusts the camera's processing gain to make it so. Dicklyon (talk) 04:54, 19 April 2014 (UTC)

Simpler examples/language

You know what this page needs? Some basic, simpler examples in plain language. I am a Super8 hobbyist (speaking of the conversation you had concerning "Film is dead," there are a LOT of us out there), and it seems like every time I sit down to buy more film from my favorite supplier, http://www.pro8mm.com, I have to try to remind myself what sort of ISO I need, and what it means. Is a small number more sensitive to light, or less? What looks best in bright light? What works fastest in low light? These are real and pressing questions for those of us shooting on film, and frankly, (talk)'s lo-tech example of the same picture in low light at different settings was exactly what I was looking for. It told me that my ISO 200 Reversal film was going to be OK in low light before I even pulled the trigger, which is why I came by, worried about wasting yet another roll. Having basic stuff like this alongside all the charts and graphs and gobbledygook would be swell. Playerpage (talk) 04:47, 7 December 2015 (UTC)