Olympus-OM
[Top] [All Lists]

Re: [OM] ACR in PSE 8.0 was: Who in the world writes this junk? - Raw so

Subject: Re: [OM] ACR in PSE 8.0 was: Who in the world writes this junk? - Raw software rant.
From: Ken Norton <ken@xxxxxxxxxxx>
Date: Tue, 17 Nov 2009 12:05:32 -0600
>
> Well, I know that you have much more knowledge about how these programs
> work than I. The only thing I can say is that there is a noticeable
> difference to my untrained eye as to how these different RAW converters
> handle highlight recovery. I have read many times that it is supposedly
> "impossible" to recover blown highlights from (to use the list
> terminology) these "E-thingies".
>

I didn't mean to imply that the highlight recovery algorithm was ONLY doing
curves adjustment. That would be false. However, there are multiple ways to
slice this loaf of bread.

I have this nasty trait of disassembling things so I can learn how they
tick. I know I'm not alone in this--a coworker who is geekier than me took
apart the electronic ignition module in his wife's brand new BMW to see how
it worked.  He did manage to put it back together just in time to survive.
Oh, having a masters degree in EE, he managed to, uh, "improve" it a little
which resulted in five speeding tickets for his wife. Anyway, this thing
about raw conversion has me in serious deconstruction mode and I'm running
"Mythbusters" like experiments minus the C4.  Although, once in a while I'd
like to blow up my laptop, other than the perverse pleasure of seeing
keycaps flying through the air and imbedding themselves in the ceiling
tiles, that wouldn't be very productive. Entertaining, yes, productive, not
so much.

Anyway, there are many ways to slice and dice the Bayer array. One common
method, early on, was to do a two-pass approach whereby each detector's
value was used for luminance value and on the second pass, color
determination through mixing the triumvirate colors was assigned to a given
pixel. This worked somewhat ok.  Another way of slicing and dicing is to
work in pixel pairs to determine luminance value and three or four adjacent
detectors merged to determine color. Development in converters in ongoing
and is really one of the dark sciences behind digital imaging. The chip
manufacturers and camera manufacturers work hard at perfecting an algorithm
which maximizes the image quality while providing fast performance. This is
why you see differences between identical images from Panasonic and Olympus
E-thingies even though they may use the same exact sensor!  ACR is generic,
and although extremely good, it is not fine tuned to each and every camera
out there.

I say all this because how you demosiac the data is very likely to have a
huge impact in the top and bottom stops of exposure range. Of course,
knowing the bitdepth of the RAW pixel data and which way it is mapped is
also important to know...

Each algorithm has its own advantages.  This begs a question--do we know
what algorithm we are using and why or why not it may or may not be best for
our files?  Is it ACC, VNG, ECW or AHD based?  The exact same file processed
through two different converters will give different tonal qualities,
colors, noise patterns and dynamic range.

A good example of this is how a converter handles high contrast transitions.
When you have greens and reds come together in an image, how does the
converter handle that?  Some converters produce a grey line, others white
and one or two might produce an alien color altogether! How about diagonal
lines?  Some converters will produce stair-step artifacts, while others will
render the line correctly while some will just produce mush.

So, my point in all this is that RAW ain't RAW and for those who poo-poo the
idea that ACR isn't necessarily the end-all-be-all in converters are sadly
mistaken. Maybe it's best for Canon and Nikon files, but those of us
shooting E-thingies must be a little more adaptive and creative.  It's not
that ACR is bad, it's just that there is more to raw conversion than
color-order.

AG
-- 
_________________________________________________________________
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