On 4/17/2014 5:15 PM, DZDub wrote:
> Thanks Moose and Chris. I worked quite hard on them.
And it's a really first rate gallery!
> Are you certain that they should not look as I have rendered them?
"Should" is a tricky word. If you think they look the way they should, you are
right.
I do believe that a rendition more true to the original subject is possible.
It's entirely possible that you did a perfect job - right up until downsampling
and/or conversion from aRGB to sRGB,
either by you or the gallery software. I have certainly had many cases where
downsampling has caused clipping of what
was a perfect histogram at full size. I've also had occasional trouble, with a
'hot' red channel, in particular, with
conversion for the web. Some traffic cones drove me wacko a few years ago.
I don't consider what I did to be a complete solution. That's not possible with
these JPEGs. It's intended only as an
illustration, to show in which directions highlight retention will take images.
> If so, what convinces you of it?
It's obvious to my eye. I probably wouldn't have said anything, but for Chris
apparently seeing the same thing and
commenting on it as something possibly unavoidable. I see blown red channels on
red-orange-yellow flowers on the web
(and in print) all the time. If I mentioned every instance I see posted here,
I'd be much more annoying than I already
am. :-) As I said, it's a great looking gallery, with a few, relatively minor,
flaws in red channels.
It's also obvious, and easier to explain, in the histograms, so let me start
there.
<http://www.moosemystic.net/Gallery/Others/Wilcox/Orchid_Red_Channels.htm>
I hope it's easy to see the clipped red channels. The big bump of red against
the right that turns into a tall red line
at the very end. There's only so much I can do with a clipped JPEG, so I left a
little clipping on the red one, to avoid
making the image look funny. I'm really only stretching out compressed
highlights. The ones that are well and truly
clipped to 255 can't be fixed.
As to viewing:
On the red one, notice the slight overall change of color. It appears to my eye
that it is a more natural shade of
flower red. In any case, the histo confirms at least that the color is wrong.
It's always wrong if a channel is clipped.
Assume an area of red is r245, b200, g180, that proportion gives a certain
color. If you lower them all proportionately,
the shade remains the same, but darker. If you raise them all proportionately,
and the red clips, the mix of color
channels changes, and so does the color.
You can see where the recovery is incomplete where the color changes less, as
just to the sides of the central veins of
the petals
A second effect, both of clipping and of JPEG engines/conversions that compress
the top of the histogram, is a loss of
visual detail. The variations in the red channel that help define details of
the subject all go to the same or only very
slightly different colors, and the detail disappears.
You can see the effect pretty clearly in the detail of the vein structure. Also
in the rough, grainy structure of the
underside of the orange ones. In both cases, there are still areas where tonal
detail was not recoverable, and thus they
remain without detail.
Does that answer your question?
Shaggy Moose
--
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/
|