Olympus-OM
[Top] [All Lists]

Re: [OM] Proper Exposure [was Best ISO for landscapes with E-M1 Mk11?]

Subject: Re: [OM] Proper Exposure [was Best ISO for landscapes with E-M1 Mk11?]
From: Ken Norton <ken@xxxxxxxxxxx>
Date: Sat, 28 Dec 2019 15:56:06 -0900
I revisited some E-1 pictures from 2005 to see what may have changed
in Adobe's latest-greatest converter engine. I do this periodically as
there are enough changes along the way that it may be worthwhile to
either change methods or tools. I also do it as a comparative of
sensor-tech to better understand the changes in modern sensors.

Modern understanding of camera exposure is that there is no such thing
as ISO and exposure is a relative concept. In the 2019 digital
photography world, this seems to be the new truism. This mythology was
partially perpetuated by DXO and their bogus redefinition of ISO. The
result is the belief is that there are no longer any absolutes. That
may be fine if you are a "Universalist", but doesn't always apply
here.

Remember the arguments over "Is RAW really RAW?"

Well, guess what, we're right back there again.

The fact is, there are SOME sensors that have technology on the chip
that adjusts the effective sensitivity of the sensor itself. It may
also include adding a voltage bias to the read result in order to keep
the rest of the system's SNR under control. But the actual sensor
itself may include the ability to effectively capture more or fewer
photons during the exposure. The voltage bias adjustment takes
whatever the read value is, adds to it and then passes this higher
number on to the D/A converter. An actual sensor adjustment bias is
different as it simulates additional photons reaching the sensor and
boosts the sensitivity of the sensor above a threshold where the
photons would normally fall below a threshold. This is the digital
photography equivalent of pre-sensitizing film. Standard voltage bias
offset is more akin to using non-diluted developer. The QE (quantum
efficiency) of the sensor is effectively variable when pre-sensitizing
techniques are used.

Modern converter engines, such as Adobe ACR, are unlike other
converters in that the basic S-Curves Adjustment is more like a guide
to a result rather than a baked in equation. Using fuzzy logic, the
converter profiles are using known camera file characteristics for a
normally exposed image and looking at the camera selected ISO to apply
an offset. Most converters use fixed profiles to achieve the result,
but Adobe's latest/greatest is throwing some additional thinking
behind the process. The result is that there is a pretty decent
"normalizing" of the images to then manipulate.

The geniuses at DPR have essentially assume that today's cameras are
all mono-ISO and linear. While that initially appears to be somewhat
true for some cameras, it's really not true. While sensor-tech is
getting MUCH better, these advances may make you think that special
in-camera sauce isn't being applied when it really is. Sensors are
still using dithering noise, dark-noise subtraction, and exposure bias
(as well as a myriad of other really fancy tricks, like pre-exposure
to kick the electron count to a threshold).

The primary change that DPR is advocating is that the four stages of
exposure (brightness, time, sensitivity, amplification), which used to
all be done in-camera, are now shifted to a new split structure. In
camera: brightness, time, fixed-sensitivity. In computer:
amplification. By moving amplification over to the computer and out of
the camera, DPR is claiming essentially that you can shoot at pretty
much any exposure and adjust in post to get the same result with the
only damage being at the ends of the exposure scale.

Let that sink in for a minute. Is it true?

A modern day Ansel Adams would say "Zone System" and he/she/they would be right.

A very simplified interpretation would be to "expose for the shadows,
process for the highlights". I think this applies to this modern
thinking. Expose the image in-camera in such a way that the shadows
are captured as close to final need (no lifting), and lift the
highlights to achieve desired final results. This is the OPPOSITE of
shooting Canons where ETTR is so incredibly critical. "Expose for the
highlights, process for the shadows". A Canon file must NEVER be
lifted whenever possible as the shadows go very nasty. Always pull
shadows down, never up with Canons. The old E-1 actually behaves very
much like these new cameras, in that you expose for the shadows,
process for the highlights. Try to not pull highlights back in an E-1
file, pull shadows up all day long, but don't pull highlights down.

Understanding the specifications of your camera really is the
important thing. How many stops of DR at what "ISO". That's really
where the rubber meets the road. While DPR has oversimplified and
stated that all other ISOs are derived from the base ISO and just
trims off DR, that isn't always true and the curves reflect that. Why
the curves are not linear is that special sauce the manufacturers put
into the cameras that mean that RAW ain't RAW.

Probably the best advice I can give is to pretend that the modern day
(non-Canon) camera behaves more like the E-1, just with more forgiving
highlights. Expose the image as close to final result as you can get
without clipping the ends of the histogram. In doing so, the
converter's fuzzy logic doesn't have to overcompensate and move the
curves around quite as much. Remember that moving curves around
effectively eliminates much of the bit-depth we strived to get in the
first place. Expose the image so those mid-tones are as close to
mid-tone as you can get. This is ABSOLUTELY true when it comes to
portraiture!

AG Schnozz
-- 
_________________________________________________________________
Options: http://lists.thomasclausen.net/mailman/listinfo/olympus
Archives: http://lists.thomasclausen.net/mailman/private/olympus/
Themed Olympus Photo Exhibition: http://www.tope.nl/

<Prev in Thread] Current Thread [Next in Thread>
Sponsored by Tako
Impressum | Datenschutz