On 11/7/2017 4:23 PM, Mike Gordon via olympus wrote:
Kudos to Maestro Moose for such an elegant analysis and the lengths he went to
perform such an experiment. Can't say I fully understand the results, but
probably his method is the best way to test.
Working on a summary and another example.
Such "older" sensors such as on the E-1 were noted for color accuracy and some
other pleasant rendering features but it is more difficult to quantify objectively than
one would think.
The thing is, testers would most likely use the larger aRGB color space, and find decent color accuracy. At the same
time, some users can like the inaccurate reds of Oly's version of sRGB.
The sensitivity metamerism index (SMI) is defined in the ISO standard 17321 and
describes the ability of a camera to reproduce accurate colors.
The SMI varies depending on the illuminant
Sure, green light and all bets are off.
and is independent of the converter.
I don't understand. Esp. with Bayer array, there are no colors, just three incomplete monochrome matrices, until after
demosaicing. The converter creates the RGB colors. It could be argued that a Foveon or an HD Oly image has colors, but . . .
Sonnie tends to do well here with a denser CFA
Is that how they get away with financial shenanigans, a denser Certified Financial Analyst? Seriously, what's CFA in
this context?
but truth be told the measure is noisy above 80 and not very good at
discriminating the sensors.
If it doesn't work, what's the point? I can see gainfully employed scientists playing with it. I can't see why anyone
else would pay much attention.
Purples and true violets are notoriously difficult.
Part of the reason I chose the example example test subject I showed, purples and violets. As Ken has pointed out
before, and as you know about flutterbys, there are colors that don't exist in printed form. How does one create a
standardized test target for those?
Oddly with OM5DII files I found taking Marnie with true violet in the image
worked much better with DXO conversion then ACR in being able to portray the
true violet with accurate skin tones.
Again, how can the ISO standard test be independent of converter??
Portra 160 had no trouble with a good scan. Admittedly I may not be practiced
enough with the HSL features, but I could not get it right in PS if I stood on
my head with the ACR conversion.
I don't understand the colorspace finding as usually found going from raw/tiff
16 bit to 8 bit jpeg in srgb an issue.
I was careful. I chose a subject and exposure that left the top of the histogram v. sparsely populated. The aRGB to sRGB
conversions didn't blow out any highlights. That's also why they are rather dark/low key. I didn't want to apply any
post conversion editing that might muddy color comparison waters. That subject can be made to sing and dance, but that's
not the point here.
As long as none of the larger gamut colors crash into the top, things are good. I don't know how deeply you have looked
at the PS conversion tool. There is a choice for "Dither", which will use adjacent colors to simulate the colors that
don't directly convert, or some such thing.
I tried another subject, bright red/orange pelargonium in direct sun. That was going to be impossible unless I reshot at
much lower exposures*, converted from Raw with lots of negative EV, or worked to aRGB JPEGs. As CH showed, there are
folks out there using non-color profile aware browsers. Interesting images, as differences are extreme, but won't see
the light of day.
I had an action from an FM'er to prevent clipping but the aRGB processed
always looked a bit washed out after conversion.
There are tricks in PS . . ., but some colors/subjects are really difficult.
like with the E-M5II in the Moose example. No clue why that doesn't occur with
the e-1.
It does, unless one is careful in subject and exposure choices.
Smugmug does not support tiffs or aRGB so I I abandoned trying to establish an
aRGB gamut workflow though I would prefer to keep the files 16 bit with a
wider gamut.
I process to ProPhoto in my regular work flow, and store the processed file as 16 bit PSD. Then downsize and convert
color space for the web (all in saved Actions).
I don't member precisely what Zone-10 supports now but I believe supports at
least the wider gamut. As long as the image is going to be crammed into 256
flavors of each color it just seemed easier to keep things in sRGB like Chuck
recommended years ago not long after I joined the list. I hope to be able to
change that workflow some day.
That's why I save full gamut work.
Full Gamut Moose
* Which would mean additional processing, and defeat the whole "pure out of the
camera" thing.
--
What if the Hokey Pokey *IS* what it's all about?
--
_________________________________________________________________
Options: http://lists.thomasclausen.net/mailman/listinfo/olympus
Archives: http://lists.thomasclausen.net/mailman/private/olympus/
Themed Olympus Photo Exhibition: http://www.tope.nl/
|