DXO Mark

My summary and a few opinions on these two articles in a series about
data measuring vs subjective camera testing and about the price
difference for relatively little additional performance:
http://www.luminous-landscape.com/es...vs-value.shtml
http://www.luminous-landscape.com/es...-numbers.shtml

Nice concise description of the situation. Of course he throws some
opinion in too.

He says DXO Mark does not include megapixel count... OK that was not
clear to me. Seems easy enough to figure in though. I would probably use
a linear measure rather than area but you can just assess that for your
print size needs if you like.

And he describes how the DXO Mark data is done with raw files on the
same raw converter - and how some cameras like Nikons have pattern noise
reduction processing applied to the raw conversion where others rely on
the jpeg conversion or proprietary raw converters to accomplish that
step. And how the medium format digitals apply antialiasing in post.
Feel free to correct my reading of that or other factors I missed.

Then he says because of these irregularities, the data approach is not
very reliable. The comparison is made to the audio recording industry's
conversion to digital[1] and he asserts that the data model was rejected
by audiophiles in the end, citing at least one example of how
manufacturers cheated to create good numbers at the price of fidelity.
Well, I don't discount that but this is where it gets into religious
territory: saying that it's all just too complex and we can never know
the true nature of the universe. That's of course somewhat true but the
numbers are something we can know and continue to evaluate and learn
more about and it looks to me like they are mostly spot on so far,
especially given the above qualifications.

Lastly, he discusses price for performance ratio. He gives personal
examples breaking the rules in both directions for the advice given in
these discussions so I'll write that off as Rockwellian hyperbola g.
But yes, there is a big jump in price for not much performance
improvement as you go up and everyone will obviously find their own
priorities to solve these issues. It seems about right when he says a
5-point difference is statistically insignificant where today's DSLRs
come in between 50 and 80 on the DXO Mark scale, with P&S at around 35.
I'm guessing a crappy web cam or cell phone might be about a 10 on that
scale.


[1] one might also compare the transition of video from analog to
digital, and to higher resolution, which is really still in it's
infancy, to the extent that file size & compression have more to do with
image quality than anything else as far as I know, again, feel free to
correct and add info here g.

--
Paul Furman
www.edgehill.net
www.baynatives.com

all google groups messages filtered due to spam