Ken Norton wrote:
>
> There is nothing magical to "highlight recovery". It is nothing more than
> automated curves adjustment.
Oh dear, I must disagree again. Perhaps the clearest example I know of
is on the bottom of this page.
<http://www.dpreview.com/reviews/canoneos5d/page23.asp>
As you can see, some RAW converters "fill in the colors" using guesswork
from nearby areas where the color channel isn't blown. ACR clearly has
some pretty good algorithms for the purpose.
Dcraw is interesting, as one may vary the amount of attempted recovery,
including levels beyond ACR. Effectiveness varies with subject. A very
high setting will work what appears to be magic on one image and look
terrible on another.
> If you convert to 48-bit format and apply curves to that you can end up with
> the same result--just not as easily.
> Highlight recovery does not actually do anything in the process of converting
> the RAW file itself, but applies the settings to the RAW to internal format
> after the conversion algorithm has been run.
>
> In nearly any converter you can just do your conversion to protect the
> highlights (essentially underexposing the picture) and then apply a single
> upward arching curve adjustment to bring the midtones back to proper
> exposure. This is the poor-man's highlight recovery method.
>
Poor indeed, in many cases. See above.
>> Personally, I have only seen these red and blue etchings when converting to
>> jpeg with ACR in PSE 8.0
>
> This leads me to believe that ACR's demoisiacing algorithm isn't processing
> the RAW file correctly and/or is internally converting to 24-bit mode without
> preserving dynamic range.
>
I don't know what's going on with Candace's particular PSE problem. The
complexities of demosaicing are many and subtle - especially for 4/3
sensors.
Most sensors are in a GRGB array, with identical green sensors. 4/3
sensors are in a RGGB array. That would be no big deal, except that the
filters over the two green sensors are slightly different. So while most
arrays may be processed with both green channels combined in a three
channel algorithm, 4/3 sensors require a four channel, R, G1, G2, B
algorithm,and one which either takes account of the green channel
differences or allows the user to fine tune them. This is probably why
the Oly converters and a couple of high end others do the best color
conversions.
As to the 'etchings', they are not confined to 4/3 RAW files. Here is a
page which shows some similar artifacts, incliding blue-yellow 'etching'
in some NEF files and how a wide range of converters and algorithms
handle the problematic areas. <http://www.rawtherapee.com/RAW_Compare/>
> The more I've been learning about these algorithms and sensor design the more
> interested I am. This is not a static technology, that's for sure. Even
> Olympus in Studio1 brought out the "High-Function" converter which used
> a different algorithm than what was in the cameras themselves and in Master.
> Apparently, Studio2 uses "High Function" as the only algorithm now and the
> improvements over the in-camera converter (for earlier E-bodies) is
> very noticable.
>
> I'm seeing reports from the Lightroom Beta users that the converter has been
> changed and is producing images similar to SilkPix which utilized a totally
> different type of algorithm.
>
The above link shows how different the results of various programs and
algorithms can be. More stuff about the 4/3 sensors here.
<http://www.ojodigital.com/foro/perfectraw-perfectblend/257378-labyrinth-artefacts-green-equilibration.html>
Moose
--
_________________________________________________________________
Options: http://lists.thomasclausen.net/mailman/listinfo/olympus
Archives: http://lists.thomasclausen.net/mailman/private/olympus/
Themed Olympus Photo Exhibition: http://www.tope.nl/
|