« Photo Rumors (and Merry Christmas) | Main | Camera of the Year 2011: The Contenders »

Monday, 26 December 2011

Comments

Truly a great idea! But.. Likely too mathy, the logrithm part anyway.

"perceptually wopplejawed" = nice.

I suppose you also want(ed) to replace "megapixels" with "% of increase in pixel dimensions?" Marketing-types certainly won't cotton to your "non-wopply" logic (just sayin').

One could just call 51,200 and 204,800 50K and 200k...

DIN might be a more sensible way to keep track of the really sensitive ISOs.

The Dvorak keyboard layout is also a more efficient means when writing about photo topics. Wonder how many people actually use it, though? :)

And since you talk of standards, we should remember that there is (at least) one reason for manufacturers often use terms such as "Hi1" or "ISO equivalent" instead of just "ISO": because these numbers would *not* represent any performance appropriately described by... the ISO technical standard. They are not ISO at all, nor ASA, nor DIN...

204,800 is like a christmas tree with way too many flashing lights on it. To wrap my mind around those crazy numbers I started thinking of ISO 3200 as ISO 3k, 6400 as 6k etc. Once you pass 25k the numbers start to look very familiar again.

=:o) Gert

3k - 3200
6k - 6400
12k - 12800
25k - 25600
50k - 51200
100k - 102400
200k - 204800

Or perhaps we just scrap both of ASA and Din and move to "stops", starting at 100 = 1, with decimals allowed. Therefore, 200 = 2, 400 = 3, 800 = 4, etc..... Decimals could be used. This would make it even easier for everyone to translate it to stops because you would just need to subtract the two numbers. With DIN, each stop is represented by 3 numbers, not sure why, perhaps because at the time people only broke it down to 1/3 stop increments and they didn't like to use decimals? I'm sure Ctein could enlighten us.

Either way, I don't have a problem with the existing ASA scheme, but if we are going to change it, why not make it even easier.

Yes, because doing equations in your head with mutiple competing logarithmic scales is much easier.

To me, the biggest conceptual problem with the DIN scale is that I have trouble understanding why a doubling - or halving - of the sensitivity results in a change of 3 units. Mathematically I understand it (I think), but up front and in my face, it's a puzzle. ISO actually makes ore sense to me. yes, some of those very big numbers look odd at first sight, but I find the relationship between 51,200 and 204,800 makes more sense that between 38 (I think.....) and 44.

I think a more useful standard would be one that had the sensitivity on one side accompanied by the signal to noise ratio on the other.
This would help put these numbers into perspective.

Well said Mike.

Now to sober up the discussion: I don't think this would get very far in a world where the major market continues to hold out against rational units (i.e. the metric system).

Talk about resistance to change.

I hear you and agree that those super-high ISOs look silly. But the problem for me with the DIN scale is that its numerical compression destroys the obvious arithmetic relationship between two speeds. Those I can compare at a glance; not so with DIN's 1/3-of-a-stop steps.

Background: as a practical matter for me, who shoots film and digital (D40), all of my action is happening at ASA (ISO) 1600 or less. So for me, those sky-high ISOs are simply ridiculous.

I also wonder how many photographers actually care all that much about them. I can see why marketing departments and advertising companies will care. But picture takers?

If an ad tells me a camera's sensor is fast enough to shoot nighttime street scenes without flash using exposures of something like 1/60th second, then I'd say I'm covered. What more are you really going to need? I'm taking photos, not doing surveillance!

Cranking up the ISO strikes me as mainly a marketing gimmick — or, to be more precise, it's one of the selected technical details camera manufacturers are choosing to compete with. And frankly, like a great many of these advertised "features," the utility to picture-takers remains unclear.

Also — final thought — ridiculously high ISOs serve a marketing purpose only as far as people can intuitively see that X brand's sensor is clearly faster than Y brand. Somewhere around 6 or 7 digits is where most people's ability to mentally compare ISOs will stall out, I suspect.

And keep in mind that going DIN means a jump backward to low, low numbers. Try to write an ad that makes DIN 43 look better than ISO 12,800. Best of luck!

Jeesh Mike, DIN was even before my time. Besides exposure changes are in doubles or halves, exponential changes, not an arithmetic progression. Why would you want to use a system that doesn't reflect what's actually going on? As far as I'm concerned DIN is dead. May it rest in piece.

Woe is me. My head is reeling from thinking about this. Especially since I hardly ever shoot above DIN 30.

I propose a planetary scale. "Earth" would be ISO 100. "Venus" would be ISO 200, because of the dense atmosphere. "Mercury", ISO 50. The outer planets of course get very little light, so I propose we use something from the Kuiper Belt for the new high end of Canon's flagship, say, "Eris", only because Pluto was degraded to non-planet, and now has lost a little of its marketing charm.

Better yet, why don't we make a new scale? Since ASA numbers double to indicate a one-stop change and DIN numbers increase by three to make a one-stop change, why don't we create an easy to fathom scale like 10, 20, 30 etc? 00 can be the base of the scale and 10 would be a one-stop increase, 20 equals two-stops, etc. I'm just sayin'...

That (DIN) not only (seems) IS more proportionate but IS more clever too.
One number = 1/3 stop, three numbers = one stop and so on
To fine tune exposition is a far better metod.

Doesn't matter what system you use. Double the speed half the aperture.

With computer, never forget they are computers, taking over the camera who cares what system is used.

Then again I'm never used a higher ISO than 1600. If I had to go higher I'm probably in the wrong place and need to get out fast.

What I would like to see being used is a scale akin to the Exposure Value (EV). For example, at a family gathering yesterday, the ambient light must have been around EV 4 (I was shooting at around 1/125, f/1.4, ISO 1600).

If the "ISO" part could be replaced by an "ES" number (let's call it an "Exposure Sensitivity"), so that ES 0 was (say) ISO 100, then instead I could say I shot at ES 4 (ISO 1600).

ISO 204,800 would be ES 11, which is an easily manageable number, and understandable as 11 stops above ISO 100.

In my skeptic view, even in a Canon 1D X anything above ISO 25600 is pure gimmickry. I can't think of a serious photographer (the one allegedly targeted by such a camera) using anything above 12800, no matter what conditions. So, counting the ISO numbers is not a real problem, in my view. Obviously, some would take photos at ISO 204,800 just because they can, but in real life conditions, I very much doubt that a serious photographer would do that.
In my view, camera companies should focus their effort in getting lower noise in the ISO 800-6400 range instead of increasing absurdly the ISO range.

Mike

Working in the oil industry which uses a mix a of "American" and metric units the use of a scale in America which results in huge numbers is no great surprise. American engineers always give weights in lbs and European engineers use tonnes, 10,000lbs c/w 4.54 tonnes. The Americans could use short or long tons giving 5 short tons or 4.46 long tonnes but no it's 10000 lbs. I always assumed it's a marketing thing - look you're getting 10,000 whole pounds of stuff rather than 5 short tons. Which would you rather have? I imagine the ASA thing was similar ISO 800 just seems to sound better than ISO 30 except now it's gotten outta hand.
Hope you had a good Christmas
Gavin

Hmm, personally I think the current ISO scale makes more sense as it is consistent with the scales used for Aperture and Shutter Speed. If I halve the ISO, I double the shutter speed - easy. I think it would be difficult to change this, as shutter speed is based on seconds which is an actual measurement that people can conceptualise, and is useful.

Perhaps cameras should have a 'standard' and 'advanced' modes, standard mode would have auto ISO, etc as suggested by psu, advanced mode would have the ability to turn automatic features on and off, and also full manual control.

I seem to recall back around the early 1970's BMW were racing in a US series where their engines were supposed to be putting out more horsepower than their opponents and they were perennial backmarkers. When a reporter asked one of the German team about this the engineer replied "Deutschland Invented Number". Still makes me laugh.

For the record- you're responsible for 7 of 8 Google "wopplejawed" entries. Just sayin'.

I do sometimes use ISOs below 3200; like on those rare occasions when I'm outside in the sun.

I have to tell BuLiGaS that I have needed ISOs about 12,800. If the otters are in shade AND moving fast as they play, you just gotta do what you gotta do.

Losing the ASA numbers would lose us the sunny-16 rule! Inconceivable!

Probably the important use of the numbers is in comparing cameras. Especially for that, finding an acceptable objective measure of noise is the key. Perhaps image quality overall -- but some people care about color much more and noise much less, say, so one overall measure probably suits only a few people.

Hmmm; if we make "k" = 1066.666 etc., things come out right -- 3k is 3200, 6k is 6400, etc. Then we can have the special pleasure of k=1000, k=1024, and k=1066 depending on the area. That should help avoid confusion!

@psu:

Actually, the ISO numbers for digital cameras are not just a guideline for setting the gain in camera. There's actually an ISO standard for measuring the sensitivity of digital sensors, which aims to give results roughly equivalent to the old film standard. My understanding is that this is the reason the edges of the sensitivity scale are reported as "Low" or "High" values rather than an ISO speed. That's an implicit admission by the manufacturer that those parts of camera's sensitivity range are not well calibrated to the ISO scale and are only approximate.

If anything, it makes more sense to have a calibration standard for digital sensors than for film. Digital sensors have a very linear response, a measured full well capacity, and are developed using a well defined curve. That means it's possible to map a given light value to an output value in the final file much more precisely than it ever was for film, which makes standardizing sensitivity measurements more practical and meaningful.

I think it's just an outbreak of the sensor "megapixel race" again. Different measure, but still the same marketing philosophy behind it.
It's not going to make any difference to film users - they'll continue to do what they've always done and the film manufacturers certainly aren't going to join in the competition.
And as digital cameras (even the 'pro' models) become more and more automated the relevance of manually adjusted sensor sensitivity to light is going to become less relevant to the photographer.
I'd say the best thing is to stick with what we've got, but let the camera manufacturers run riot competing with each other and have fun poking them with sticks every so often. Eventually some form of sanity will prevail but I can't see the august, but bureaucratic ISO organisation taking much interest unless they think that they can make money by imposing a new standard for which everyone has to buy a copy or a user licence from them.

One thing you don't want to do is to get away from a standard, so that every manufacturer uses a different set of numbers and everything turns to chaos. But the problem with creating new standards is that all kinds of international organizations get involved, and there are politics, etc., so changing standards takes forever.

What camera companies could do is take advantage of the fact that their machines are now computers, complete with computer screens. Simply using ISO, they could set up a graphical interface on the back LCD touch screen that would somewhat resemble the current color histogram display.

On three side-by-side scales, the red line would represent shutter speed, the yellow line the aperture, and the blue line the ISO. A fourth green line along the top would represent compensation for under and over exposure. At "normal" exposure, you could push ISO down, and you'd see both S and A (f-stop) fall on the scale. Or you could lock A, push ISO down, and see S drop twice as quickly. Or, you could lock ISO, then push A around, and watch S move around on its own. Or, you could lock one, and let the others move automatically. Once you got used to it, it'd take a couple of seconds to readjust all scales, and to actually see what you had.

Of course, you'd actually have to understand what ISO, aperture and shutter speed mean.

As for making cameras simpler, they couldn't get any simpler, IMHO. All my cameras have an auto setting, which means it's a point-and-shoot. Push the button, and you get something, probably, if you're not shooting in the dark. I think what you mean by simpler cameras is, simpler for advanced photographers.

...psu sounds like a guy who never uses professional series flash, hence shoots until they see something they like on their camera back, which isn't necessarily accurate either, even if you're looking at the tiny chart...making the mistake of thinking that one persons methodology is the same as every one else's...half the time I'm actually doing a hand-held meter reading on a subject, even in continuous light, and getting closer on the first shot than the camera meter...

...for those of us that use professional series flash literally 90% of the time, sometimes even asa 200 is too, too much to try and do exterior balancing, or get the wide open f/stop you want...AND, you want to take an accurate strobe meter reading to get a decent starting point, seems like someone here is making an argument for getting LESS accurate, which is folly...

"Plus one" to every one on here that says it's a gimmick, it might be great for photo-journalists, so it needs to exist, but I can't imagine ever using it...

Changing the understanding or use of ASA may be OK for people that don't want to learn what it means, but are we designing professional cameras for them now? (Hey, you can make 160,000 ASA just read 160K) That's how I'll know that photography as a career is really over...the number one mistake Photoshop made on day one, and the reason you know it wasn't designed for photographers (just pre-press people) is it never used the terminology like 'stops' for density and 'grades' for contrast...let's make another stupid mistake like that, shall we?

Why not replace all ISO, DIN and ASA numbers by the number of photons per square millimeter per second required to achieve saturation?

Well, there's one advantage of the current "ISO" race. Marketeers are now telling the engineers to program absurdly large numbers on the ISO menu. They seem to have stopped telling them to create sensors with absurdly large numbers of too-small photosites. With the former, we can simply ignore the big (noisy) numbers. But with the megapixel race, we were stuck with too many (noisy) pixels.
Letting marketeers obsess over irrelevant numbers seems better than having them muck with the important stuff.

Please, Mike, that's not a good idea. Then I'd have to change my blog name from 'ISO 100' to 'ASA 21'. You see, in my country 'asa' means 'wing'. People will think it's a blog about aeronautics.

I can sort of imagine working with a scale that apparently increments in 1/3 stops, which the DIN scale seems to do, but I'm sure it would slow me down.

However, the cynic in me believes that consumers and marketers are allergic to log scales, and predicts that we'll end up with sensitivity multipliers based on ISO 100 as a standard. ISO 50 would be x0.5, ISO 400 would be x4, and so on.

For a while, some of us old-schoolers would cling to the ISO equivalents in our minds, but the newbies wouldn't care, nor need to. And we'd soon catch on that this is how we always dealt with ISO numbers in the field anyway; and that things like the "Sunny 16" rule and guide numbers not only still work, they're rendered slightly simpler to think about and teach.

Two things that bug me in the digital era

Why can't the raw data from the sensor be stored in floating point, then blown out highlights would go away and all this iso stuff would be
irelavant.

Why can't lenses be speced by their absolute aperture as well as focal length divided by aperture, and resolution in radians as well as line pairs per mm in the image, like telescopes for instance.

At least I can take care of the second problem in my head, even if it makes my head ache.


Why not go back to the APEX scale where ASA100 = 5, ASA200 = 6, etc. The makes the ridiculously high ASAs into numbers like 15, 16, etc.

DIN is like decibels. No one really understands that math.

Who needs ISO, f-stop, shutter speed and color temperature when one can just point and shoot? The only relevant figures are megapixels, frames per second, and oh, price. :)

This all reminds me of a trick question in a photography class asking what the the exposure was in bright daylight for film. rated at a speed of twelve. If the student asked whether it was ASA or DIN , then the student got the question wrong.

I often say ASA and need to correct myself to ISO--showing my age--as the value is expressed on my Sekonic meter (which was pricey on my budget) and several of my cameras with inbuilt meters.

Perhaps I'd have to carry a scrap of paper with ISO/DIN conversions.

For my sloppy technique I see ISO as a rough estimation of image quality when shooting digitally, so any related series of numbers would probably be OK, assuming many books and websites were rewritten.

Still photography, "digital style" looks rather amateurish when compared to cinematography and broadcast television.

A broadcast video camera and high end digital cinematography camera describe the sensitivity as "the aperture required for correct exposure under a certain amount of light (LUX)"

And if there isn't enough light to allow the use of a preferred aperture, you wind up the "gain" which is measured in dB (deciBels) and that's a logarithmic scale. 3db means twice as sensitive. 6db means four times more sensitive. Of course, the signal to noise ratio deteriorates at higher gain settings.

The high end lenses for the cinematographers define the aperture size in "T-stops" which means Transmission stop, in reference to the actual halving or doubling of the light that passes through the lens, taking into account the losses of the lens elements.

So if the inaccuracy and inappropriateness of the digital still photography camera world weary one too much, one can find peace in the high end digital cinematography and broadcast television camera world. That's where photographers get serious and specifications make complete sense. But the entry fee is high.

If we want to continue enjoying today's low entry fee for digital still cameras, we have to put up with the consumer style marketing and product specifications, because it's the sales to non-professional consumers that subsidise our professional pursuits in still photography.

Btw, there is a great survey of the history of film speeds at

http://en.wikipedia.org/wiki/Film_speed

I found it while trying to remember how the General Electric ( Edison patented movie photography after all ) and Weston ( Edward Weston the rival of Edison in electrical enginering , not Edward Weston the photographer ) film speeds compared .

It goes into the train wreck of ISO speeds for digital cameras . 5 different ways to calculate the sensitivity , but only for sRGB jpeg , it changes depending on the lens's vignetting and the metering pattern and there is aparently no standard for b&w .

As a Brit I think we should go back to the old British H. and D. scale. That's rounded off, so a stop faster than 6,400 is 13,000, and a stop faster again is 25,000. Simple.

Of course, it's based on a different system than ASA or ISO. 200 ASA would be 6,400 H & D, and 6,400 ASA would become 200,000. Two and a quarter stops faster would be a nice round million H & D.

It would drive most of us mad, but the advertising execs would love it!

The EV scale (and DIN, which is EV*3) seems to me to be much more sensible, and directly related to the normal camera controls, than the exploding ISO numbers. Zero and negative EV values cover perfectly reasonable but now forgotten sensitivities like ISO 50, 25, and 10 (ah, old Kodachrome). A comment above finds multiplying by two a more "intuitive, natural" way to express a doubling of sensitivity, allowing one stop more exposure, but that probably depends on taste. To me, ISOs in the thousands, millions, billions and trillions are in poor taste, and just an invitation to more innumerate confusion.

Oh no! Not another number to remember!

I'm an available-light photographer who almost never shoots BELOW 1600 ISO, unlike many of the more "serious" photographers who commented initially. I shoot indoor sports, theater and musical performance, and social events. I NEVER use a flash.
The race towards higher ISOs is a welcome development to photography, both professional and popular. Some day we will laugh about the days that we had to have an explosion of light near the camera just to capture a good image.
I agree with the general point that ISO numbers are becoming unwieldy and confusing, although they are no worse that shutter speeds (which get larger with shorter times, unless the "1/" is carefully included) and of course the anachronistic f-stop numbers.
It's time to throw out all three numbering systems out the door and design something that will better express the three parameters of sensitivity, time, and opening.

All I can add is a big resounding YES to higher ISO/ASA's! I constantly find myself shooting in not just low light, but sometimes seemingly almost no light.

Last concert I shot was at ISO 6400, f1.8 and 1/40th. And then I had to bump the exposure compensation by .75 a stop in post to get something usable. Whew - talk about shooting in a black hole! Just yesterday was at a museum and ISO 2000-3200 was the norm for the day, and even then that wasn't enough all the time. My daughters' indoor soccer with the 70-200/f4? You get the idea.

My next body upgrade will be with something with a very nicely usable ISO 12,800 (minimum). As it is, my 50D starts to get unusable (to my eyes anyway) depending on the shot above ISO 2,500. Hopefully I'll be pleasantly surprised with the 7D and/or 60D successors from Canon.

While we're at it, is there any reason not to move away from f-stops? Like the EV proposals, let's just use unit changes to indicate halving/doubling of sensitivity. This would put aperture and sensitivity on the same scale as the proposed DIN/EV suggestions.

I will let my kids worry about this mess.

Mike: Rounding and K-ing the values is a good alternative. Skipping intermediate f-stops is a good alternative too. I used to have my 5D2 programed to skip the 1/3 f-stops since I believe that in digital is not necessary if you shoot raw. But then, I installed Cinestyle and learned that multiples of 160 is strongly recommended. So, I went back to programing my 5D2 for intermediate f-stops. Bummer!

Ditto Scott Kirkpatrick's comment that EVs make much better sense - see the Wikipedia article at

http://en.wikipedia.org/wiki/Exposure_value

The EV100 (EV with a 100 subscript) system shown in Table 1 describes the combinations of shutter speed and aperture that are equivalent exposures for ISO 100. Each EV step is, of course, a one stop change. Any EV100 index number in the table is related to the next higher sensor sensitivity by one step (and one stop in either aperture or shutter speed) in EV. Thus ISO 200 is +1 EV100, ISO 400 is EV +2, etc. This is an easy way, IMHO, to see quickly just how much more sensitivity ISO 51,200 is than ISO 100 (+9 stops).

The *much* more important point, however, is the radical change that such very high sensitivities is making on photography. Suddenly, supplementary lighting of all kinds (shoe-mount flash a la Strobist, studio flash, hot lights, etc.) become primarily artisic tools rather than exposure necessities that need to be very carefully used to achieve artistic goals. The trick now that we have these extended sensitivity sensors with acceptable quality is to choose the settings that achieves the intended artistic goal (e.g., backgound blur, motion blur, depth of focus) by varying the EV (or ISO, if you insist) instead of finding the compromise that provides almost all or most of what you want artistically yet stays within the confines of lower EV (ISO) values that were all we had with acceptable quality until now. What a tremendous change...

As for the UI.... If camera manufacturers really wanted to innovate it would be nice to see an open source consortium started for all camera functionality. It would be amazing to see what would happen if third party developers were allowed to utilize the basic capabilities provided by the camera manufacturers (e.g. ISO, shutter speed, aperture) and extend it in new and unique ways.

Everything above what i used for film i abbreviate, ie 12k, 25k, etc.

But don't forget that there is a good basic reason to continue to supply iso numbers: because some of us still prefer to expose manually. I've used some mighty sophisticated cameras, but i have yet to find one which does a better job guessing exposure than i do.

"You can memorize such numbers easily and learn to translate them into stops easily."

Actually, no, I can't :)

All I've managed to sink into my head is that the number is proportional to how much I'm likely to bump into things because I'm not a cat. Other than that, well...

As for using DIN instead of meaningless ISO - surely you jest? Now that the Great Megapixel War is just about over, all the poor salespeople have left to awe us into buying is high ISO numbers, and you want to take that away for them? O, cruel blogging world.

(Thanks for yet another great year of illuminating - at the very least 25,600 ISO! - resding, Sir.)

Some 18 months ago we proposed at EtL to get rid of the ISO scale and move to a stop-based scale starting at 1, which is equal to ISO 100. ISO 200 would be 2, ISO 400, 3, etc. For more details and rant see here:

http://enticingthelight.com/2010/06/24/is-it-time-to-abandon-the-iso-scale/

Not saying I prefer one over the other but..
If you had the opportunity to change you could change it to...anything.
Start with with a blank piece of paper.....

Jetissoning "needlessly fine distinctions" like 1/256 second makes it harder for some of us. Powers of two are pretty much second nature at this point, after 40 years.

(In fact, the shutter speeds are already messed up. 1, 1/2, 1/4, 1/8, 1/15 (oops), 1/30, 1/60, 1/125 (oops), 1/250, 1/500, 1/1000 is the sequence I'm used to, and it glitches twice.)

While it is interesting to see this escalation in ISO values, as others have pointed out, having a scale that relates well to f stops is handy. Likewise, there has to be a standard for sensor sensitivity for those working with external light meters.

Not every camera feature is relevant to every genre of shooting, but that does not make these features irrelevant overall or silly, they may be essential to someone.

The comments to this entry are closed.