Compensating for M8 exposure

[jaapv]

Advertisement (gone after registration)

If it makes your ego feel better to think so, be my guest. Or read the next posting, maybe it will be clearer. If not, e-mail the International Standards Organization and ask them if ISO-whatever# is different for film and digital. Honestly it's the first time I've ever heard it said, but not the first time on an internet forum that some photographer came up with a cockamaymie theory and it ended up becoming a myth.

It only takes 40 ISO to get your knickers in a twist? I personally could not care less... The only thing I know is that I think my photos are correctly exposed - which means to my taste.

[Advertisement]

The Leica Q3 is available for pre-order at B&H Photo Video and Adorama

[mboerma]

The meters in M6, 7 etc use a round spot on the shutter and if you read those manuals Leica's diagram shows a more or less circular pattern. The M8 has a painted shutterblade, and the diagram shows an oblong pattern. If you want to argue, argue with Kaufmann or Stefan Daniel. I don't claim to know better than Leica.

 

It might be my lack of understanding of the English language. But I was pointing at the fact that you were blaming the shutters and the way they were colored for the way the meter is measuring. So I thing we are talking about to different things.

[giordano]

That is not the esposure meter. The way of determining ISO is different on digital than it is on film. ISO 160 in the digital world is equivalent to ISO 200 on film. To compare you must set your exposure meter (and flash control for that matter) to 200 to level the playing field. 360=400, etc...

Sean Reid noted as much in his first M8 review, back in 2006.

 

Hi Jaap,

 

Obviously the way of determining the ISO numbers has to be different on digital than on the various kinds of film, but can you give a citation for "160 in the digital world is equivalent to ISO 200 on film"? I can't find a mention of it in Sean's first M8 piece or in the "Correct exposure and other myths" piece he posted about the same date.

 

The only ISO 12232 text I can find proves that the intention was that ISO numbers should match across film and digital (and it's hard to imagine that anyone would have wanted otherwise):

The methods for assigning an ISO speed rating to electronic cameras should harmonize with current photographic standards and practice. In order to be easily understood by photographers, the ISO speed rating for an electronic camera should directly relate to the ISO speed rating for photographic film cameras. For example, if an electronic camera has an ISO speed rating of ISO 100, then the same exposure time and aperture should be appropriate for ISO 100 rated film / process system. (

Final Draft International Standard ISO 12232, October 1997

)

[peterv]

It only takes 40 ISO to get your knickers in a twist?

 

haha, watch what you say now, the difference between 10 ISO and 50 ISO is 2,3 stops...

[luigi bertolotti]

Hi Jaap,

 

The only ISO 12232 text I can find proves that the intention was that ISO numbers should match across film and digital (and it's hard to imagine that anyone would have wanted otherwise):

 

 

The methods for assigning an ISO speed rating to electronic cameras should harmonize with current photographic standards and practice. In order to be easily understood by photographers, the ISO speed rating for an electronic camera should directly relate to the ISO speed rating for photographic film cameras. For example, if an electronic camera has an ISO speed rating of ISO 100, then the same exposure time and aperture should be appropriate for ISO 100 rated film / process system. (

Final Draft International Standard ISO 12232, October 1997

)

 

If this is the ONLY ISO doc on sensitivity of electronic cameras... well anyone (say, any manufacturer) can decide which "ISO strategy" to implement : it's a "reccomendation" and not a "statement of standard" (years ago I was in charge of an ISO sub-committee, in a completely different field, but have a certain knowledge of their processes). In this sense, Leica can have decided for a certain methodology on assessing their ISO values, and this can surely bring to results that can be interpreted as "160 on M8 is like 200 on film" (this, btw, is my rough feeling, too), but can be different for Nikon and so.

 

With film (though I have never red the ISO docs on the topic) I think a reasonably simple process for sensitivity assessment can be drawn:

- Exposure for a standard time with a standard light (spectrum, power, location) in a standard environment (reflectivity)

- Develop with a standard chemical formula with a standard process (time, temp.)

- Evaluate the graytone to a standard scale.

This is a semplification... but the principle can be so. With a "crude" sensor maybe something standardized can be envisioned (equalized measures of microvoltage variations...?) but the problem is that the "crude" sensor is "dressed" and implemented by the manufacturer to obtain a RAW file with bit strings...with software in the middle... farewell to standard references (and it would be SO easy... bit values...maybe something could be arranged around a standard like DNG... I am thinking I would like to speak with some old friend still into ISO org....)

[jaapv]

You beat me to it, Luigi. This is a weird piece of prose for a body that is supposed to set international standards. It reads more like a wish than a standard sheet and carefully refrains from defining tolerances. (Years spent at Delft Technical University taught me something about things like this ) To me that can only be interpreted in one way: The difference in sensitivity response between film and sensors (and maybe even sensor types) is such that it is impossible to establish a one-on-one relationship and any number must be treated as a rough approximation. In that case a difference of one third of a stop is well within the ball park. After all, correct exposure in a photographic sense is no more than an interpretation, and it is known that even the best shutters can be off by more than 30% at some speeds without detrimental results.

[jaapv]

Advertisement (gone after registration)

Hi Jaap,

 

Obviously the way of determining the ISO numbers has to be different on digital than on the various kinds of film, but can you give a citation for "160 in the digital world is equivalent to ISO 200 on film"? I can't find a mention of it in Sean's first M8 piece or in the "Correct exposure and other myths" piece he posted about the same date.

 

I must confess I quoted from memory. It was in his first M8 (p)review from November 2006, the one with the firemen shot in it. He has subsequently edited considerably and I would have to dig through his whole website to see if it is still there somewhere. I should mail him and ask, I suppose...

[gdi]

Exposure isn't really that complex, but there are things to bear in mind...

• treat the meter as a spot attached to the RF patch. It's job is to turn the reflected light you point it at into 18% (thereabouts) grey. Not black, not white.
  
• If you have trouble imagining what that grey represents, buy a cheap grey card from Kodak (they're under $20).
  
• Learn to meter from significant parts of the scene and exposure lock if you shoot in Av. Me? I shoot in manual so I can place highlights and shadows where I want them (see below).
  
• If you point the RF patch / meter at something white (highlight) that is important to preserve, the Av will turn the white thing grey and will "underexpose" that important white thing, and bring the tonal values down for the rest of the shot. People often have this happen at high ISOs, where they point the RF / meter at the brightest object (which often isn't the subject but a light or something) and the camera consequently underexposes the subject, leading to noise, gnashing of teeth, and complaints about high ISO repsonse.
  
• If you point the meter at something black / dark grey (shadow) that is important to preserve, then the Av will turn the black thing grey and will "overexpose" that important black thing, and bring the tonality up for the rest of the shot. This is often a better technique with digital if the highlights just aren't important, and often they're not. At higher ISOs, shooting RAW, where you can bring down tonal values in post, this often garners better results than "preserving highlights" which often aren't important to the shot anyway. IOW, in a high-contrast situation, you need to make a decision about what's important to the shot and what isn't; if all of it important, you need to modify the light (bring the shadows up with a flash or reflector).

IOW, your meter will turn things "average" grey (which is really quite dark!). That's the way a reflective meter is supposed to work, and the M8's works really well.

 

Now, the hardest thing for me coming from film was to understand high ISOs in digital: you simply don't have the latitude or extra stops of noise-free detail in the shadows you do at lower ISOs with the M8 (or any digicam, actually).

 

At ISO 160-640 the M8 is very forgiving here, due to the fact it captures a lot of data). However, you need to make sure at high ISOs you don't starve the sensor for light. Again, read the above carefully.

 

As an example, if you have a picture of someone blowing out candles, don't meter off the candles! Their face is the important element; let the candle flames blow out and expose for the light surrounding their face (probably darker due to fall off). That way, the whole exposure gets "pushed up" and their face gets a proper exposure; the candles themselves are blown (but who cares?) and you can set a proper blackpoint.

 

But you can see that simply setting -Ev will not work. You need to read how the meter really works.

 

BTW--I recommended this before, and I'll do it again: if this mystifies you buying a cheap incident meter and learning how to use it will improve your exposures till you "get" the built-in meter.

 

 

I thank you for the effort to explain the basics of exposure Jamie, but I am very familiar with exposure, metering and the related concepts such as incident meters. I believe my post was not as clear as it could have been - perhaps implying I was looking for a "formula" was the problem.

 

What I was really getting at is how to best handle the exposure problems present with the M8 in a fairly systematic way. The problems I refer to are - first the tendency of blowing highlights in high contrast lighting situations and then the problem of difficult to manage noise at ISOs 640 and above when shadows need to be brought up in post processing.

 

I have been approaching the first problem the way many other users have been - by using -ev to compensate for the lack of highlight headroom in in-camera metered M8 files. (Of course looking at histograms is fine for after the fact analysis, but it is no substitute for having the camera "dialed in" for such situations when you can't shoot by trial and error.) This works fine and is not usually an issue.

 

But alas, this solution, which works fine at low ISOs because there is the ability to bring up the shadows in post falls flat with higher ISOs. In those situations you have very little leeway with shadows in post due to the high noise levels with the camera. So if you underexpose based on the meter reading to preserve highlights in lowlight situations (yes highlights get blown in lowlight too! ), your shadows can be hard to salvage for color shots.

 

So I think the only solution is to futz with the -ev (or constantly adjust your manual exposures ) - driven by the noise characteristics of the camera based on ISO level. Of course there is usually some compensation that needs to be done with all digital cameras, but the fact that there is such an abrupt tipping point, with respect to noise, with the M8 makes it much more of a problem than I have experienced with other (lower noise) cameras.

 

I know there is no panacea, but I am just trying to gauge how others are dealing with it.

[Chuckcars]

ALL digital camera will blow out the highlights. No exceptions. I set mine to -2/3 for landscapes at iso 160. -1/3 for iso 320.

[gdi]

ALL digital camera will blow out the highlights. No exceptions. I set mine to -2/3 for landscapes at iso 160. -1/3 for iso 320.

 

It is all a matter of degree...

[Jamie Roberts]

{snipped}

So I think the only solution is to futz with the -ev (or constantly adjust your manual exposures ) - driven by the noise characteristics of the camera based on ISO level. Of course there is usually some compensation that needs to be done with all digital cameras, but the fact that there is such an abrupt tipping point, with respect to noise, with the M8 makes it much more of a problem than I have experienced with other (lower noise) cameras.

 

I know there is no panacea, but I am just trying to gauge how others are dealing with it.

 

You're really talking about exposure latitude here, then, and not an "exposure problem." That's what threw me, I guess.

 

In this regard we are actually, I believe, spoiled by the tremendous response of the M8 at low ISOs, where there is no "exposure problem" whatsoever (however, people still overexpose quite easily using Av not understanding how the basics of the M8 meter works).

 

IOW, in a friendly way I completely dispute your assertion there *is* an exposure problem with the M8 There is only an exposure latitude limitation at higher ISOs (and that's to be expected in a 2 year old design now, and the M8 is not much worse than my 5d, in truth).

 

But by dialling in -Ev, at low ISOs, you are simply "sloppily" preserving highlights when metering something that would cause them to blow out in a normal fashion (and when I say normal I mean "as with any other digital camera"). Yes, that's ok at lower ISOs because the M8 saves you from (arguably) bad exposure technique, mainly because of the extra latitude the M8 seems to have in the shadows due to 14bpp AD conversion, compressed or not.

 

So--with the exception noted below--I stand by my assertion that in high contrast light, at higher ISOs, with *any* digicam, you need to:

• choose to preserve the highlight detail and block the shadows / increase noise in shadows OR
  
• choose to preserve the shadow detail and blow the highlights, (even though it's true that at lower ISOs, you can rescue shadow information while preserving speculars).

The only way to have both highlights preserved and relatively noise-free shadow detail at high ISOs with the M8, given it's noise floor and dynamic range, is to preserve the highlights with your exposure and artificially light the shadows.

 

Now, the very newest Canons and Nikons apparently have about two stops exposure latitude on the M8 in terms of maximum shadow response and noise floor (they all blow the highlights equally in my experience, though they meter differently).

 

This does mean that shooting a 1d3 at 3200 is a lot like shooting the M8 at 640... and so is much more forgiving. But not in the highlights

 

Is that what you're talking about? Sorry to be so long-winded. Until there's a sensor or electronics change, I don't see any other options for the M8 (of course, using faster glass makes a difference in getting light to the sensor in low-light situations).

[lars_bergquist]

Here are a few points:

 

• 'ISO speeds' on film cartons are usually no ISO speeds proper. The text normally goes something like 'To be exposed like ...' The reason is that there exists an ISO norm for the determination of film speed, and if the manufacturer said 'this film has the ISO speed of ...' then he would also have to follow that norm. Now, the 'ISO speed' is actually an Exposure Index: 'Expose this film AS IF it were an ISO XXX film.'

 

• Digital sensors, somewhat like slide film, has essentially no exposure tolerance. This is absoluely true on the overexposure side. What you expose for is what you get. A blown-out highlight is blown forever. On the underexposure side, gain can be applied in post-processing, but to the detriment of the tonal scale.

 

• With a reasonable analog camera, you could set the meter to any exposure index you pleased. With a competent camera, bells and whistles were not going on all the time, because the manufacturer knew that pro users would nearly always want to tweak the meter, especially (now listen, Leica!) when using slide film with its narrow dynamic range. In other words, they tweaked not for a momentary 'exposure compensation' but in order to adjust the meter semi-permanently to their needs.

 

I have therefore argued for a way for M8 users too to permanently tweak the meter to their needs, as distinct from an exposure compensation. You say that you can already do that. No, you can't. You can menu in exposure compensation, signalled by a blinking red diode in the finder, so you won't forget to remove it. And that facility should remain, for what it is worth. (I prefer to switch over to manual - my M8 is no less capable on manual than my M6 was). Permanent meter biasing is a different function, and should be signalled in the finder, IF AT ALL, by a separate and NON-BLINKING symbol so as not to confuse the absent-minded and distract the sensitive. Me, I think it would be enough with a heading for meter bias in the main menu.

 

Now listen again, ye cohorts of Solms! This is a needful item for the M8-2 that you will probably present at the photokina this autumn. And of course for the next firmware version.

 

The old man from the Age of DIN Speeds

[giordano]

If this is the ONLY ISO doc on sensitivity of electronic cameras... well anyone (say, any manufacturer) can decide which "ISO strategy" to implement : it's a "reccomendation" and not a "statement of standard" (years ago I was in charge of an ISO sub-committee, in a completely different field, but have a certain knowledge of their processes).

 

It's not the only ISO doc on sensitivity of electronic cameras, it is the only text of ISO 12232 that I could find on the internet without having to pay. The passsage I quoted was from the introduction and expresses the intention, not the methodology. (http://www.opsci.com/AppNotes/ShortOAN-004ISORatingDigitalCameras.pdf)

 

In this sense, Leica can have decided for a certain methodology on assessing their ISO values, and this can surely bring to results that can be interpreted as "160 on M8 is like 200 on film" (this, btw, is my rough feeling, too), but can be different for Nikon and so.

 

Could be. But Jaap seemed to be claiming that digital cameras in general, not just the M8, need a 1/3 stop different ISO setting to produce equivalent photographic results. I found an ISO statement that appears to contradict this and asked for verification.

 

With film (though I have never red the ISO docs on the topic) I think a reasonably simple process for sensitivity assessment can be drawn:

- Exposure for a standard time with a standard light (spectrum, power, location) in a standard environment (reflectivity)

- Develop with a standard chemical formula with a standard process (time, temp.)

- Evaluate the graytone to a standard scale.

This is a semplification... but the principle can be so.

That was tried early on but proved unrealistic. The method that was developed post WW2 and was adopted by ASA, BSI and DIN before being adopted in the ISO standard is rather more complicated. You don't need to read the ISO documents, lots of textbooks explain it.

 

With a "crude" sensor maybe something standardized can be envisioned (equalized measures of microvoltage variations...?) but the problem is that the "crude" sensor is "dressed" and implemented by the manufacturer to obtain a RAW file with bit strings...with software in the middle... farewell to standard references (and it would be SO easy... bit values...maybe something could be arranged around a standard like DNG... I am thinking I would like to speak with some old friend still into ISO org....)

 

The ISO methods for determining film speeds are all based on the needs of practical photography rather than on the underlying physics, and they seem to have taken the same approach in ISO 12232. Maybe this gives more flexibility to digital camera manufacturers than to film manufacturers - or maybe not: as I understand it the "3200" b&w films actually have a speed of about ISO 1000 (i.e. using the ISO method) and successful exposures at EI 3200 depend on push processing.

[luigi bertolotti]

You beat me to it, Luigi. This is a weird piece of prose for a body that is supposed to set international standards. It reads more like a wish than a standard sheet and carefully refrains from defining tolerances. (Years spent at Delft Technical University taught me something about things like this ) To me that can only be interpreted in one way: The difference in sensitivity response between film and sensors (and maybe even sensor types) is such that it is impossible to establish a one-on-one relationship and any number must be treated as a rough approximation. In that case a difference of one third of a stop is well within the ball park. After all, correct exposure in a photographic sense is no more than an interpretation, and it is known that even the best shutters can be off by more than 30% at some speeds without detrimental results.

 

That's my exact feeling, too: all of us know, in practice, what is "sensitivity" in the photo domain, but is a too complex concept to establish a rigorous and unique MEASURE of it, when derived from two completely different physical phenomena (light on a film emulsion, light on an electronic sensor), with completely different ways of "extracting" the basic information (a chemical process, an A/D conversion of voltage values, with some mandatory processing).

And Lars' first observation I think is important in this sense : even for films, manufacturers spec is that "... it can be exposed as a ISOxxx ..." (and probably, years ago, at the times of darkroom passionate, there had been discussions on topics like "Agfapan 100 is really a bit more sensible than an Ilford 125...").

[luigi bertolotti]

....

 

That was tried early on but proved unrealistic. The method that was developed post WW2 and was adopted by ASA, BSI and DIN before being adopted in the ISO standard is rather more complicated. You don't need to read the ISO documents, lots of textbooks explain it.

.......

 

Giordano, can you give me some textbook title/author/editor on this ? This thread has made me curious about the basics of "sensitivity measure" ... would like to read something about...

Thanks Luigi

[giordano]

[quote name=giordano;522993Giordano' date=' can you give me some textbook title/author/editor on this ? This thread has made me curious about the basics of "sensitivity measure" ... would like to read something about...

Thanks Luigi[/quote]

 

Hi Luigi,

 

Two I have are

• Jacobsen (ed.) The Manual of Photography, Focal Press. This is a very detailed discussion of most aspects of photography. Mine is the 7th edition (1978); I think there's been at least one edition since then. The 6th edition (1971) might not cover the unification of ASA, BS and DIN under the ISO standard, but will have the whole theoretical and practical background. Earlier editions (Ilford Manual of Photography, I have the 4th) are interesting but don't cover the later refinements of speed determination.
  
• Dunn & Wakefield, Exposure Manual, Fountain Press. A really good explanation of the issues behind exposure for monochrome neg, colour neg and reversal film - and having read those it's easy to apply the principles to exposure with digital cameras. I have the 4th edition (1981); the 3rd (1974) should be just about as good, but earlier editions will be more of historical interest. I don't know of a more recent edition.

Enjoy!

[jaapv]

Could be. But Jaap seemed to be claiming that digital cameras in general, not just the M8, need a 1/3 stop different ISO setting to produce equivalent photographic results.

 

Well, there have been some posters on this thread with vast Canon experience, and they claimed that Canon tended to underexpose by 1/3rd of a stop at the nominal ISO setting. I went back to my old 10D and 1D shots and I tend to agree, though I confess I did not notice at the time. That suggests Canon assigned a different, more consumer-safe, number to the same sensitivity and that Leica went a more ( as German companies are wont to do) scientific way and assigned what they deemed to be the correct number.

[gdi]

You're really talking about exposure latitude here, then, and not an "exposure problem." That's what threw me, I guess.

 

I guess it would have been more clear had I said latitude rather than "headroom".

 

 

In this regard we are actually, I believe, spoiled by the tremendous response of the M8 at low ISOs, where there is no "exposure problem" whatsoever (however, people still overexpose quite easily using Av not understanding how the basics of the M8 meter works).

 

IOW, in a friendly way I completely dispute your assertion there *is* an exposure problem with the M8 There is only an exposure latitude limitation at higher ISOs (and that's to be expected in a 2 year old design now, and the M8 is not much worse than my 5d, in truth).

Ok I'll stop referring to it as a problem - I agree it is having to fully understand the metering pattern and exposure algorithm chosen by Leica for the M8 with respect to the limited latitude available.

 

 

But by dialling in -Ev, at low ISOs, you are simply "sloppily" preserving highlights when metering something that would cause them to blow out in a normal fashion (and when I say normal I mean "as with any other digital camera"). Yes, that's ok at lower ISOs because the M8 saves you from (arguably) bad exposure technique, mainly because of the extra latitude the M8 seems to have in the shadows due to 14bpp AD conversion, compressed or not.

 

Yes - it is a hamfisted approach to trying to use the M8 in a manner similar to other cameras I use. Other cameras do seem to be designed to call for less exposure for a given scene, thereby reducing the liklihood of blowing highlights. I need to adjust my technique - I use a Mamiya 7, mostly with positive film, so I am used to compensating in a similar way. I think I made the mistake of listening the too many internet "rules" for using the M8!

 

 

So--with the exception noted below--I stand by my assertion that in high contrast light, at higher ISOs, with *any* digicam, you need to:

• choose to preserve the highlight detail and block the shadows / increase noise in shadows OR
  
• choose to preserve the shadow detail and blow the highlights, (even though it's true that at lower ISOs, you can rescue shadow information while preserving speculars).

The only way to have both highlights preserved and relatively noise-free shadow detail at high ISOs with the M8, given it's noise floor and dynamic range, is to preserve the highlights with your exposure and artificially light the shadows.

 

Now, the very newest Canons and Nikons apparently have about two stops exposure latitude on the M8 in terms of maximum shadow response and noise floor (they all blow the highlights equally in my experience, though they meter differently).

 

This does mean that shooting a 1d3 at 3200 is a lot like shooting the M8 at 640... and so is much more forgiving. But not in the highlights

 

Is that what you're talking about? Sorry to be so long-winded. Until there's a sensor or electronics change, I don't see any other options for the M8 (of course, using faster glass makes a difference in getting light to the sensor in low-light situations).

 

There really is no easy answer - beyond making tradeoff decisions you describe. That was the answer to my question, thanks. I have already reset my ev comp!

 

I appreciate your effort - taking the time to describe your approach in the context of the camera and its limitations (and doing so without the defensive posture which is so common on these forums!)

[Advertisement]

Hallo,

Look here: Leica M

[mjh]

It is all a matter of degree...

In this case, it isn’t really. It’s just the way a sensor works, accumulating electrons at a constant rate until the pixel’s capacity for storing electrons is reached and any excess electrons go down the drain (anti-blooming gates).

 

Because of this characteristic of any CCD or CMOS sensor used, one can improve ones chances for an optimal exposure by slightly underexposing; shadows can still be restored later, but blown highlights cannot (unless they only appear to be blown in the JPEG version). But of course, that doesn’t imply that there was a systematic difference between „digital ISO“ and „analog ISO“; it is just that one needs to adapt one’s exposure strategies to another technology with different characteristics.

[gdi]

In this case, it isn’t really. It’s just the way a sensor works, accumulating electrons at a constant rate until the pixel’s capacity for storing electrons is reached and any excess electrons go down the drain (anti-blooming gates).

 

Because of this characteristic of any CCD or CMOS sensor used, one can improve ones chances for an optimal exposure by slightly underexposing; shadows can still be restored later, but blown highlights cannot (unless they only appear to be blown in the JPEG version). But of course, that doesn’t imply that there was a systematic difference between „digital ISO“ and „analog ISO“; it is just that one needs to adapt one’s exposure strategies to another technology with different characteristics.

 

 

Actually, I think you are stating that is can be a matter of degree - based on sensor characteristics!

 

And your comments about the Digital ISO nonsense is misdirected - I'm not involved in that OT debate!

 

Regards...