EOS and ISO Ponderings

[oak_nj]

For digital SLRs and Canon in particular, is the sensitivity of the

digital sensor (ISO speed) actually a REAL "meaningful" limitation on

the camera and the chip's sensitivity? Or is it built into the

camera's ability to capture an image to replicate the film ISOs in

order to please photographers and make them feel more comfortable and

to maintain the photographer's "laws of physics"? I am wondering if

possibly it is built in just for marketing (such as the Canon spot

meter dilemna)to advertise different ISO scale and to necessitate the

need for consumers to purchase faster lenses for lower light shooting.

Is the 800 ISO in a film camera comparible to the setting in a

digital camera, or could the digital camera capture the same image in

the same light at a different "sensitivity" setting sort of similar in

concept to the ACCEPTIBLE range of depth of field. I am sure that

there is a sensitivity issue in digital cameras, but is it the same

magnitude as in the film? The chemical processes and reaction to

light in the film dictate the ISO speed (I assume), but is that also

true for a digital camera? Why couldn't a digital camera have a

sliding ISO scale that would have the set ISO speeds for

traditionalists, but would also let you to adapt to your light level

needs. Would this function be helpful? It could possibly be set with

an automatic ISO setting similar to auto exposure that could maximize

the shot say on the histogram when in MANUAL mode based on a selected

ISO and shutter speed. Please set me straight... Thanks!

[beauh44]

Hi Joe, Someone else is bound to give you a more technical answer than this, but here goes. "ISO" is a standard - in fact the abbreviation comes from "International Standards Organization". You should find that a digital camera's rated ISO is pretty close to film. If it were not, it could potentially cause a bit of havoc; hand-held light meters would be rendered useless.

 

When you increase the ISO in a digital camera, you will likely increase noise - this is a function of the chip's signal-to-noise ratio. The same thing happens with film too, except it's referred to as "grain". An amplifier kicks up the signal in a digital camera though.

 

You've noticed something that does bug a lot of photographers though: Most digital cameras don't allow the flexibility to *easily* use ISO just as you would shutter speed and aperture. It would be nice to have a digital camera that "promoted" the ISO setting so, for example, you could have "Shutter priority, Aperture priority or ISO priority", in effect. You can of course change the ISO on a digital camera but the setting is usually squirrelled away in a menu and usually not as easy to get-to as changing shutter speed or aperture value. Best wishes . . .

[ben_rubinstein___mancheste]

My 1Ds is spot on according to my meter, the 10D is out by a stop over.

[bobatkins]

Sensors have an inherent sensitivity to light (you can think of it as "x" electrons generated for every "y" incident photons). ISO is determined by the electronic configuration of the sesor including things like the gain of the amplifiers (on or off chip).

 

For it to have any real meaning digital ISO should be the same as film ISO, so you should see indentical exposure readings for a film and digital camera set to the same ISO.

 

Sometimes you don't of course, especially with consumer digicams. I'm sure manufacturers have their reasons to do this. Getting it right isn't difficult!

 

You could have "auto ISO" selection and I'm pretty sure that the DLSR makers are saving that for an "upgrade" at some point. You pick shutter speed and aperture and the camera sets the appropriate ISO setting. It's an obvious mode and something the camera makers must have thought of. However they save these things to give you an incentive to upgrade. They bring out upgrades about every 18 months and they need a string of new features to entice users!

 

Quite a few digital cameras do have "auto ISO" modes, but it's often done in cojunction with program modes and usually limited to 100, 200 or 400.

[mendonphoto]

"Why couldn't a digital camera have a sliding ISO scale that would have the set ISO speeds for traditionalists, but would also let you to adapt to your light level needs."

 

If I understood what you said, this is the essence of your line of questioning, right? That would be nice, but don't count on it. The ISO settings on digital cameras represent the real limitations of the sensors. Otherwise, compact digital camera would feature better quality and faster ISO speeds.

 

I don't think ISO speeds on digital cameras are really limited by the technology. Rather they are limited by the literal number of photons it takes striking a sensor to register a meaningful brightness value.

[damian_tinsley]

I thought Nikon does have an 'ISO priority' mode - set the shutter and aperture and the camera chooses the appropriate sensitivity. Better ask on the other forum - I may be wrong.

[preston_merchant]

The Green modes on the 20D are Auto ISO.

[drfl]

Beau,

 

actually ISO comes from the Greek word "isos" meaning "equal", it does not stand for International Standards Organization. It is pronounced as a word as well, not an acronym.

 

See:

http://www.iso.org/iso/en/aboutiso/introduction/index.html#three

 

As such, the speeds on film should be "equal" to digital by the definition of ISO.

[oak_nj]

In the last line, I mean aperture and shutter speed, but I guess you guys figured that out.

[amir_jahanshahi]

I am a Nikon owner though I like to read the Canon threads as I very much like Canon equipment too. I am just adding this posting for information sake. I own a Nikon D70 and it does have an auto ISO mode that uses the full range of ISO setting available to the camera. You set the Aperture and Shutter speed and the camera adjust the ISO. Though to be honest I never use it as I want full control over my ISO for noise reasons and would generally prefer to adjust one of the other two variables if at all possible to keep a lower ISO. However it does work very well for those that want to use it.

 

It also does have a dedicated button to change ISO. However if I am not mistaken, I think the one canon digital SLR I looked at (the 20D) also had a dedicated ISO button (though I am not certain about this as I don't own one).

 

Hope this helps.

[beauh44]

Thanks Derrald, I stand corrected!

[bobatkins]

Well I guess fully auto ISO will be a Canon "upgrade" since Nikon already have it! Score Nikon 1, Canon 0.

 

I believe the auto ISO on the 10D/20D is limited to 100/200/400 and only available in "program" modes, though since I never venture onto that half of the command dial, I can't say for sure!

[digitmstr]

The ISO settings on my 10D are equivalent to film. I use the same Sekonic meter that I had been using for my film cameras and the results are identical, both with studio strobes and ambient light.

 

At least for my camera I can say there's NO discrepancy in ISO rating between its sensor and film.

[PuppyDigs]

I'd say ISO on my 10D is about 40% less sensitive than film. That is, ISO 100 is behaves like

ISO 60.

If I use the sunny F16 rule on a sunny day, my images are always 1/3-1/2 stop

underexposed, e.g, ISO 100@F16@1/100 sec. They look perfect at ISO 100@F16@1/60

sec. However, the meter almost always gets it right in auto and semiauto modes so I

assume ISO and meter are calibrated together.

[mark u]

One other limitation of the majority of Canon DSLRs is that ISO is only settable in whole stop increments. I think this limitation is built in so that amplification of the sensor data is just a binary shift for higher ISO levels - hence very fast to process compared with multiplication by powers of 1.259921 (1/3 stop). A further reason may be that Canon have tuned the image processing algorithms for each ISO to optimise for noise characteristics. Since a significant portion of these algorithms is hard coded in the DIGIC processor, providing dedicated algorithms at say 1/3 stop intervals would impose significant additional cost and a larger chip.

 

The Nikon solution is certainly handy for those shooting sports in fading or variable light (e.g. part of a pitch in shadow while the rest is in full sun), where the lens may be wide open for preference, and the shutter speed needs to be kept fast enough to freeze the action. I believe the Nikon D70 offers ISO in 1/3 stop increments.

[roger krueger]

<i> One other limitation of the majority of Canon DSLRs is that ISO is only settable in

whole stop increments. I think this limitation is built in so that amplification of the sensor

data is just a binary shift for higher ISO levels

</i><br><br> Camera ISO settings are mostly about analog sensor gain, not post-

processing. There's no inherent advantage to whole-stop ISOs except simplicity.

[jeff medkeff anchorage, a]

<p><i>I'd say ISO on my 10D is about 40% less sensitive than film.</i>

 

<p>40% less than which film? Is this an aesthetic judgment, or a rigorous measurement?

 

<p>I don't think I've ever done anything to calibrate my digital cameras' ISO settings to a fixed standard, but I can easily compare gray levels on digital to density values of film exposed identically, and none of the three of my digital SLRs has nearly as much as 40% variation from Fuji Astia or Kodak Technical Pan. Off the top of my head it is a single-digit figure with a lot of scatter, no strong systematic component. The great unknown here would be development standards for the Astia.

 

<p>I was a little surprised to read this because I would have considered that much delta indicative of something being broken. But I'm not suggesting there cannot be a 40% difference between what exposure is liked with a given film and what exposure is liked with digital. I could see that varying by 40% quite easily, especially when using exposure methods that maximize signal. It is just that these are different matters - density and ADU on one hand, vs. working preferences on the other.

[norm1]

My Digital Rebel 300D goes up to ISO 1600. I would imagine that the 20D would be the same or better.

[amir_jahanshahi]

I believe the 20D is up to 1600 plus a HI mode that is roughly 3200.

[amir_jahanshahi]

As far as I know Roger is correct. The main reason for the whole stops increments only is to be able to move through them quickly at shooting time. I have heard many people complain that the 1/3 stop ISO setting on the D70 take too long to set. I personally like it and have no problem working with the 1/3 stop settings since as I mentioned before it lets me achieve the correct exposure with the absolute minimum increase in ISO and thus noise. To the 20Ds credit, it has lower noise than my D70 in the first place so this offsets somewhat the need for the fine tunable ISO. But there are other advantages with the 1/3 ISO settings too. Ideally it would just be a user defined option like my Exposure compensation is. You should just be able to set a custom function for 1/3 stops, 1/2 stops or full stops. Then everyone is happy. :)

[jeff medkeff anchorage, a]

<p>ISO does change (analog) amplification and so there's no inherent performance benefit to using integer math at this step of the process, but despite any possible technical motives Canon may have, the <i>reasons</i> Canon uses a one-stop ISO increment in the Rebel/XT/10D/20D certainly has something to do with distinguishing these cameras from their "professional" line of 1 series cameras, which support 1/3 stop ISO increments.

[PuppyDigs]

"I'd say ISO on my 10D is about 40% less sensitive than film."

 

40% less than which film? Is this an aesthetic judgment, or a rigorous measurement?

 

Film of same ISO rating of course. For example, Sensia 100 shot at F16@1/100 on a sunny

day is perfectly exposed. ISO 100 on my 10D under the same conditions and setting is

under exposed. However, F16@1/60 renders a perfect sunny exposure (ISO 100 in case

you lost the train of thought).

 

Incidentally, all exposures, within reasonable limits, are subject to aesthetic judgment.

Personally I don't enjoy blocked up shadows and muddy mids.

[jeff medkeff anchorage, a]

<p><i>However, F16@1/60 renders a perfect sunny exposure (ISO 100 in case you lost the train of thought).</i>

 

<p>The question was about how you were determining you had a "perfect" exposure, and what film you were using as your comparison standard (there's a big difference between the compensations people will apply to Velvia and Fujipress, for example).

 

<p><i>Incidentally, all exposures, within reasonable limits, are subject to aesthetic judgment.</i>

 

<p>I personally like what Edward Weston had to say about this:

 

<p><tt>"There is no 'correct exposure' for any subject. Meter merely measures light available. Exposure is correct when it will produce the kind of negative that will enable you to obtain the print you have visualized."¹</tt>

 

<p>But when asking whether digital ISO is meaningfully related to film ISO, which is the basic thrust of the original post, you should compare the density produced on film of ISO rating <i>n</i> with a given exposure, to the level produced by digital at the same ISO rating and exposure (or you can scale the numbers into a common sensitivity standard).

 

<p>A transmissive photometer spot reading of a middle gray on any film that has a linear midtone crossing point should result in a ratio of 0.5 (my apologies for working in units outside of the photographic norm on this one - I'm <i>not</i> referring to the average density values some labs produce), and an sRGB or Adobe RGB rendering of middle gray by a digital SLR should be 128 (in 8 bits). These are just matters of definition, nothing shocking here. What is surprising is that you are suggesting your film of choice pegs 0.5 perfectly, but your 10D renders middle gray at around 107 (again, 8 bits) at the same ISO. Either result by itself would be very surprising, but combined it is baffling. Some of my work requires me to know this relationship and I've never encountered a difference anything like this vast. I'm not saying it is impossible, but it surprises me a lot.

 

<p>Now, film development conditions can strongly bias the actual rating of film. And aesthetic judgments can lead to things like 2/3 or more stop EC very easily. So if you are saying that things just don't look right rating your 10D at its displayed ISO setting, compared to your film results, then I follow you perfectly. But I was curious to know if this was the case, or if the information you supplied was the result of measurement.

 

<p>-

 

<p><small>¹ I've taken that from: Wilson, Charis & Madar, Wendy; <i>Through Another Lens - My Years with Edward Weston</i>, North Point Press 1998. It is a quotation, on p213, of one of Weston's magazine articles.</small>

[jim_larson1]

<i>Since a significant portion of these algorithms is hard coded in the DIGIC processor, providing dedicated algorithms at say 1/3 stop intervals would impose significant additional cost and a larger chip. </i>

<p>

This is the most logical explanation I have ever heard for the hard coded iso setting in Canon dslr's. But I don't believe it.

<p>

In part. . I don't believe it because it does not fit the Occaim's razor test. The simpliest answer here is that Canon is too stuck in standard film thinking. My 10D operates so much like an Elan 7 it is not even funny. So if you are stuck in film thinking. . .how the Canon dSLR's function make PERFECT sense.

<p>

Remember, the 10D/20D/XT/dRebel's are meant to be SIMPLE. The intended use is BASIC modes with all AF points active. The "creative" zone is not meant for every user. Adding an ISO priority mode to a dSLR, (and, more importantly) placing ISO in the viewfinder makes a dSLR nominally MORE COMPLICATED than a film camera. . .thus defeating the "auto" intent of these cameras.

<p>

I was recently on a trip with the other half. . .who was shooting with a film ELAN 7 alongside my 10D. Despite my being there CONSTANTLY, I know the only time the Elan 7 left "P" mode was when I was doing the shooting.

<p>

I submit that "power users" such as ourselves, people who tend to shoot Av mode most of the time, people who swap to "M" when using flash, is a small minority. Our collective voice is just not strong enough (yet) to break the existing film paradigms.

[sjg]

I think it would be very handy to have an ISO 'program mode', where you set an Av and Tv as well as the max ISO. The camera can then allow the ISO to fall lower to adjust exposure. A custom function could let you chose which of Av and Tv take priority when you hit the limit of the ISO range you specified.<p>

For example, I may want f/4 for depth of field, and 1/125s to cope with a bit of subject movement, and I don't mind the noise at up to ISO 400, but that does not mean I want to shoot everything at ISO 400, if the light is varying rapidly I might prefer the ISO auto adjust between 100-400 before hitting my Tv (or Av).