Olympus-OM
[Top] [All Lists]

Re: [OM] (OM) Resolution question

Subject: Re: [OM] (OM) Resolution question
From: Moose <olymoose@xxxxxxxxx>
Date: Tue, 26 Jul 2011 17:23:48 -0700
On 7/25/2011 2:13 AM, Brian Swale wrote:
> At the bottom of
> http://www.brianswale.com/zuikoholics/July-2011/July-2011.htm
>
> . . .I am not happy with the resolution as delivered by the camera.
> Details on the web-page.
>
> Would you be?

Probably not, but then, that's one reason I don't have an E-thingie.

I generally agree with all the other comments saying that the info provided 
isn't sufficient to render an opinion and 
that sharpening is necessary for ALL digital capture.

Howsomever, you caught my curiosity. Although, the poorest detail in your 
sample is in the area of the finest foliage, 
so it could be subject movement, it is also the area where anti-alias filtering 
could cause that sort of look.

So I took a look at the dpreview test of the E-3. They made RAW image 
comparisons to several comparable cameras, Canon 
40D, Nikon D300, Sony A700 and Panasonic L10. The unsharpened E-3 image was at 
least slightly softer than all four. Most 
telling was the L10, as it uses the a sensor of the same physical and MP size, 
from the same vendor and possibly 
identical to that in the E-3. DPR had this to say:

"This is perhaps the most telling comparison in this entire section. We had 
been led, off the record, to believe that 
the L10 and the E-3 shared a sensor, which should (given the 'level playing 
field' processing used) produce identical 
results in this test (it's the same lens).

Looking at the difference in pixel-level sharpness it's obvious that these 
cameras don't share an identical sensor; even 
if the silicon underneath is the same, the E-3 obviously has a considerably 
stronger anti-alias filter. There are minor 
color differences too, but these could be down to the profile used by ACR, or 
minor white balance variation. The L10 
doesn't really impress with its JPEG output, but the raw files are capable of 
delivering some of the best 'per pixel' 
detail on the market. The E-3's output appears to be far more limited by its 
sensor, and there is little (if any) 
'detail' advantage in shooting raw."

I downloaded samples for the E-3, 40D and L10 and compared them as layers in 
PS. The L10 images are the sharpest, with 
very clean edges, the 40D a little worse and the E-3 distinctly softer.

Applying Focus Magic, the E-3 image can be improved quite a bit. A radius of 
one pulls it up close to the L10 and two 
makes it sharper, but at the cost of subtle edge artifacts. Radius one plus 50% 
of two seems about the best that can be 
done, and a considerable improvement over it as from the camera.

Applying FM of one to the L10 makes it better than the best I can manage with 
the E-3, and simply cleaner.

The samples I used are available here. 
<http://www.dpreview.com/reviews/olympuse3/page32.asp>

In summary, whether your particular sample represents the best the E-3 can do, 
the E-3 is not as sharp at the pixel 
level as other cameras in it's class.

On the other hand, is it really a problem? Need it be one? The size of your 
800x600 sample on my screen is 16.8x22.4 mm, 
almost 7x9 inches. The full frame would then be 30x40 inches, which would force 
me to move back quite a ways to view it 
properly. Display at more common sizes, properly sharpened for that size after 
down sampling, and it should look fine.

Pixel peeping is a useful way to see what your camera, lens etc. are producing, 
but may lead to unnecessary hand 
wringing. I and some others are concerned because we have a tendency to use 
teles and macros, then crop heavily, so it 
does matter. For a landscape shot like your sample, it may not really matter.

Beyond that, there are tools available to greatly mitigate the problem of a 
strong AA filter at the pixel level. Focus 
Magic and other deconvolution tools are well suited to at least partially 
'recreating' the image that AA softened.

Also, you asked "Is this as sharp . . ." on your web page, but "resolution" for 
this post. There is a BIG difference! 
Much more than in analog image tech, digital makes the difference important. 
Resolution is the ability to differentiate 
edges by differences in film density or digital image brightness levels.

Sharpness is a poorly defined quality of apparent detail visibility to the 
human eye. It's some sort of mix of 
resolution, overall contrast and edge or local contrast (analogous to the term 
acutance used for film). With digital, 
it's possible for a lens/sensor combo to resolve extremely fine detail - at 
contrast so low that the image just looks 
like solid gray to the eye. Up the contrast and the detail becomes available.  
This is illustrated in the first four 
samples here, where I lower the contrast so that the detail is almost 
invisible, then recover it. Compare the first with 
the last on the top line. 
<http://www.moosemystic.net/Gallery/tech/Contrast_Resolution/Contrast_Rez.htm>

A lens with excellent contrast, and especially micro contrast, but moderate 
resolution may look very sharp at normal, 
full frame viewing distances, but resolve little detail at the pixel level.

SO . . .

Processing your image can make it show much more detail. In fact, detail that 
was in the data, but not visible as 
presented, may be brought out. 
<http://www.moosemystic.net/Gallery/tech/Olympus_E-3/E-3_cr-resolution-800.htm>

The final step may look overdone, but if you have eyes like mine, or a low 
power reading glass, you should be able to 
see greater fine detail in the featheriest parts. A print at 30x40, would look 
very good. it certainly looks good at 
large web size. 
<http://www.moosemystic.net/Gallery/tech/Olympus_E-3/E-3_cr-resolution-FF_1024.htm>

Someone else would come up with a different combination of processing steps, 
and a different final result. The point 
here is that the end step reveals MUCH more detail than the original image. 
It's in there; you just can't see it without 
help. (you can move your cursor around outside the selection boxes to directly 
compare first and last without going 
through all the steps.)

Aunti Alias 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/

<Prev in Thread] Current Thread [Next in Thread>
Sponsored by Tako
Impressum | Datenschutz