I didn't pay too much attention to AG's comments when he first posted
them but have been thinking about them some more after viewing the K10D
results (below) side-by-side using the full size images. First, the
difference in sharpness between in-camera JPEG and post processed raw is
pretty dramatic. It almost looks like the difference between an
excellent lens and a just so-so lens. But Pentax certainly knows this.
I'm wondering if the JPEG output is deliberately designed to be
"comfortable" for the average photo but with raw available for the
advanced used. Or, do their software guys just not know how to do a
better demosaicing algorithm.
I'd like to hear from AG about the types of photos he has used where
people seem to prefer the softer version. I'm wondering if subject
matter, size and viewing distance are involved. I can hypothesize that
in some types of photos (portraits in particular) an extremely sharp
photo can force our attention to parts of the photo that really aren't
intended to be examined in detail. In fact, I regularly soften portrait
images except for important facial features such as eyes, eyebrows, lips
and nostrils. For architectural and landscape images, however, I think
you'd want the maximum possible detail to allow for careful inspection
all over. But maybe not.
In any case, it doesn't matter for someone who shoots mostly raw.
Chuck Norcutt
Moose wrote:
> AG Schnozz wrote on 11/22/06:
>
>>I've been doing a little semi-scientific study on ultra-sharp
>>and ultra-detail prints. So far, in my preliminary testing with
>>"joe-consumers" (around 20 so far) is that the slightly less
>>detailed images with slightly softened edges is more
>>"comfortable" to most of the viewers.
>>
>
> http://www.dpreview.com/reviews/pentaxk10d/page16.asp:
> <http://www.dpreview.com/reviews/pentaxk10d/page16.asp>
>
> "We were a little disappointed that the K10D's built-in image processor
> didn't deliver the crisp sharp edges which can be produced by shooting
> RAW and converting using a third party engine (such as Adobe Camera
> RAW). It's speculation but it does appear to me to be a combination of
> the type of demosaicing algorithm used (not isolating edge detail well
> enough) and the type of sharpening algorithm (as discussed earlier this
> appears to work to enhance the 'undershoot', the black side, of an edge)."
>
> and page 25:
>
> "...however the single element of the entire K10D equation which left us
> scratching our heads was just that. Either a poorly implemented
> demosaicing algorithm or a strange choice of sharpening parameters means
> that while the K10D's JPEG images have plenty of 'texture' they can lack
> the edge sharpness we're used to seeing from semi-pro digital SLR's.
> Pentax may well have been aiming for a smooth film-like appearance but I
> at least feel that the inability to tweak this out by increasing
> sharpness is a mistake. That said it's unlikely you'll see this
> difference in any print up to A3 size, it's a 100% view thing so you
> have to decide if that's important to you or not. To get that absolute
> crisp appearance you'll need to shoot RAW..."
>
> Moose
>
> ==============================================
> List usage info: http://www.zuikoholic.com
> List nannies: olympusadmin@xxxxxxxxxx
> ==============================================
>
>
==============================================
List usage info: http://www.zuikoholic.com
List nannies: olympusadmin@xxxxxxxxxx
==============================================
|