Far behind on the list after another weekend busy with wonderful things. I did
have a break Sunday afternoon, but took a
nap to re-energize myself for the evening, rather than play here. :-)
I've replace the letters in the samples with titles.
On 12/18/2010 12:14 PM, Ken Norton wrote:
> Moose, I had my preferences as to which one I like out of the three sets. But
> the problem is, that the processing I liked for THIS shot wouldn't fly for
> other types of pictures.
You are absolutely right! I'm not at all sure any of the samples are what I
would do for a display image. Certainly the
most sharpened 100% samples may be overdone. However, my point there was simply
to show how much is possible, not how
much may be desirable.
My primary purpose was to evaluate the DxO deconvolution sharpening. Mike has
more than once needled me about DXO's use
of camera/lens specific measurements of images at all focal lengths and
apertures, in order to allow them to make
superior lens aberration corrections. His comments about peanut butter smears
finally spurred me into action.
I've always stayed away from DxO for a couple of reasons. First, it seems
overpriced to me for what I imagined the
benefits to be. Second, until recently, it didn't cover any camera lens combo
of use to me. After Mike's post, I looked
at the site - hey, on sale of $99; maybe worth checking out. But . . . the 5D
is a more 'pro' camera, requiring the
"Elite" edition, for $200.
Nevertheless, I decided to check it out - if only to get Mike off my back. ;-)
So, I chose an image with lots of subjects at different focal distances and
some DOF unsharpness. AS long as I was at
it, subtle coloring and wide DR with bright highlights seemed like a useful
idea.
I started processing it in DxO using default RAW conversion settings and image
processing settings. It was immediately
apparent from the histogram that I was going to get red channel highlight
clipping. So the processing settings were:
Exposure compensation = Auto - Highlight Priority - Strong, rather than the
default Medium
DxO Lighting = Auto (Much like Fill Lighting in ACR, I think.)
Vignetting = Auto
Distortion = Auto
DxO lens softness = Global, -0.5, Details, 50 (Apparent defaults,a s I set
nothing.)
Unsharp Mask = off
Noise = Auto
CA = Auto
The result was 'G'. Pleasing colors, although possibly over saturated relative
to the subject. Some over compressed
highlights, to my eye.
I then processed the image using ACR. -1/3 EV, Recovery and Fill Light avoided
highlight clipping while leaving the
highlights uncompressed. the result, 'A', is a little flat, which is how I
usually like my RAW conversions, leaving
other adjustment to PS.
'B' is after applying PTLens. For this subject, I probably wouldn't have
bothered, but DxO applied distortion
correction, and I wanted comparable comparison images. The corrections are
slightly different. It's not too obvious on
the full image, but there is considerable difference away from the center at
100%. It's interesting to me that DxO
enlarges the image to cut out the black edges from pincushion correction. This
seems wrong to me. The correction itself
is already introducing subtle changes by moving pixels around. Why add another
process of slight upsampling? I
ordinarily crop without expanding. Most images are going to be cropped and/or
down sampled later anyway. For the
comparison, I used the same option in PTLens.
'C' adds modest LCE and Curves.
'D' applies NeatImage. There's enough noise that later sharpening is going to
accentuate it at 100%. Customs settings,
some resharpening applied to bring up more than the original sharpness.
'E' is Focus Magic, set at 4 pixels, one less that the FM recommendation.
'F' has post down sampling USM applied. I left it a bit too crunchy on purpose.
'G' It's hard to make visual comparisons of various other factors in images
with markedly different color and
saturation. This is the DxO version with color matched more closely to the
ACR/PS versions.
On the 100% samples the choices are:
A - ACR & PTLens
B - LCE & Curves
C - NeatImage
D - DxO, color adjusted
E - Focus Magic
It's my opinion that DxO, at its default settings, while an improvement, is
slightly less effective at making details
visible and so increasing apparent focal sharpness as the combination of LCE,
NI. Add Focus Magic and DxO looks poor by
comparison.
Can the DxO results be improved through learning how to use its settings? I
would think so, but am not sure I'll go any
further with testing it. I just doubt that it can end up improving enough to
best what I already have and know how to
use. For an on sale price of $200, I can't see it making sense.
To be fair, as Mike has pointed out, the 5D doesn't record focal distance in
EXIF, so DxO's potential may well be
greater with other cameras. For me, there are thousands of 5D RAW images and
more to come . . .
Sharper 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/
|