ISO 100 on M10

[Dirk Mandeville]

Advertisement (gone after registration)

What is more important: higher ISO, higher DR or better picture?

False equivalence. Higher ISO and higher dynamic range don't matter . . . until they do. If the subject needs higher ISO or more dynamic range, the resulting photograph will be better if you have them.

[otto.f]

What is more important: higher ISO, higher DR or better picture?

or being able to use your Noctilux undisturbed by filters

[marblemadness]

This chart shows the dynamics range in ISO 100 is a bit lower than expected from the line over ISO 200.

But still, it has a better dynamic range than 200.

Giving this data is correct, ISO 100 could be pull-processed from an ISO around 150 but not 200.

 

Just my 2 cents.

[adan]

That is our current operating assumption, based on the 3rd(?)-hand statement from Leica of "100-150", previous tests on the M240 sensor, etc. etc.

 

And my own experiments that seem to put DR for 200 and 100 rather close - but at different ends of the exposure range (200 more tolerant of highlights, 100 more tolerant of shadows).

 

And to the questions asked of me on the previous page - I haven't really made a choice yet between 100 and 200 as "normal operating ISO" for myself. I like the clean shadows of 100, but I tend to need a higher ISO for other reasons, like DoF and fast shutter speeds.

 

For the moment I am running a consistent -0.3 ex.comp at any ISO, but that has more to do with Leica's base metering.

[UliWer]

...but I tend to need a higher ISO for other reasons, like DoF and fast shutter speeds.

 

...

 

Perhaps thinking twice of your position that Auto-ISO was something that happens to other people could help.   

[adan]

Not really - I have a nice little knob that I can use to set the ISO manually.

[M11 for me]

Advertisement (gone after registration)

And Saint Sean Reid spoke to us today . . .

 

ISO100 is a Pull.

[UliWer]

Well, not much more than Adan described in this thread: ISO 100 is less tolerant to highlights; one might guess that the base ISO of the M 10  is something between ISO 100 and ISO 200. So if ones aim is to stay away from burned out highlights one should opt for ISO 200, if you look for regaining details from the shadows, ISO 100 might be your choice. 

[Tailwagger]

I'll confess to a certain level of ignorance here, but if this analysis is correct, it strikes me as rather strange that base ISO is therefore essentially unavailable. Given my understanding that ISO multiplication and division is achieved through amplification and attenuation, and that the optimum S/N ratio is achieved at base ISO, can anyone enlighten me as to why Leica might have chosen this route? 

[adan]

These factors, at least, come to mind.

 

1. "ISO" is a loose figure to some extent. Go to DxOMark and check out their measurements of the actual ISO sensitivity of sensors across the industry (Leica, Sony, Canon, Nikon), and there is usually some discrepancy between the labelled ISO(s) of a camera, and the actual ISO DxOmark measurements. Not a huge amount, generally on the order of 1/3-1/2 stop. Some are closer than others - none fall exactly on the "nominal" ISO line.

 

M10 has not been tested yet, so you can pick between the M240 or the SL as an "approximately" equal sensor.

 

2. Go to DxOMark's measurements for DR or S/N ratio, and between the lowest ISO and the next lowest, that ALSO shows a drop of only 1/2-1/3 EV (f/stop) for DR, or 4 db of S/N ratio. So it is not as though the DR or S/N ratio drops through the floor if the "real" ISO is 150 and the selectable ISOs are 100 or 200.

 

3. Let us suppose the real base ISO of the M10 sensor is 135. Leica could, of course, mark their ISO dial or their ISO menu listings as "67.5 (pull) - 135 - 270 - 540 - 1080 - 2160 - 4320"...etc. But I imagine the response of the market would be "Where's my ISO 400??" A few precisionists would be happy, though. I suppose a halfway house would be to list 100-135-200-400-800.... on the dial, and 100-125-135-160-200-250-320-400-500-640... in the menu list. But it messes with the accepted mathematical nature of ISO spacing (1or 3 ISO steps per "doubling" of sensitivity).

 

4. I would imagine that sensors, to some extent, are engineered according to the principle of "cut once, measure twice." That is, you pick a pixel architecture - which is to say, (if you are Leica) something a manufacturer has already designed - and you estimate or prototype and get a basic "monochrome" sensitivity (X many photons deliver X many electrons). So you know you are in the ballpark of, perhaps, ISO 295. Then you spec your color Bayer filter densities, based perhaps as much on purity of color as on hitting an exact target density overall, and prototype a full sensor stack (silicon filters, microlenses, IR filter/coverglass) and measure that - and get 135 (or 137 or 142) as the "real" color ISO.

 

It is extremely unlikely (and expensive) to consider going back to square 1 and re-engineering every production mask and every layer of the silicon architecture (pixel size and performance), and your color filter dye choices (densities and transmission characteristics), just to get your base ISO of 137 to fall exactly on a traditional ISO value (100 or 125 or 160 or 200...) - which are themselves not perfectly linear anyway: ISO 64 x 2 ≠ ISO 125). You take what you get, and adjust the AD converter or other processing to nudge the output to the nearest round ISO number(s).

 

5. Combine 1-4 with the fact that. with silicon, you are working in the realm of quantum effects, and thus what any given pixel, or any given pixel amplifier/transistor (for a CMOS sensor) produces will be within a "cloud" of quantum probabilities, and you realize that worrying about the DR difference between the exact base ISO, and "100" or "200," is going to be lost in the margin of error.

 

Nobody is going to be able to detect the difference. And no other manufacturer does it that way either - for the same reason(s).

[SrMi]

For the interested: http://www.photonstophotos.net/Charts/PDR.htm has M10 dynamic range measurements.

[M11 for me]

Thx. See Post 103 

[sdk]

For what it's worth, Lloyd Chambers compared the performance of Leica M Typ 240 and Leica M10 at various ISOs fro indoor and urban evening scenes. The M10 sensor seems less noisy at all ISOs, and particularly above ISO 800, the M10 is at least one stop better in noise. The M10's ISO 100 seems ever so slightly less noisy than its ISO 200 performance in evening blue sky rendering too. Landscapists may want to use the ISO 100 setting when practical, but avoid blowing out highlights by bracketing exposure.

[Berlinman]

New test at 100 and 200 ISO M10, M240 and SL at Reid Reviews.

 

 

Gesendet von iPhone mit Tapatalk

[jonoslack]

Great article from Sean as usual! It bears out exactly my feelings (use 200 ISO on the M10 if it's very contrasty!)