Is this over sharpened?

http://1mpkzq.blu.livefilestore.com/..._8_g/Furry.jpg

PeteD wrote:
http://1mpkzq.blu.livefilestore.com/..._8_g/Furry.jpg

Hmm. For a sensible viewing distance it's not bad. But looking
closely, there are bits (eyes and other high contrast edges) that are
showing haloes, yet other areas apparently in focus look soft and need a
bit more..

My guess is that if you were using PS USM, all the settings were a
little high, especially threshold (maybe 2-3 for a clean image like this).

It also depends a lot on how you got there - you may have lost some
sharpness in downsampling. I presume it is downsized? If so in one
hit? and what algorithm (Bicubic, Lanczos)? Maybe try a few steps, with
a small amount of sharpening each time. If it's a worthy image, I will
often downsize in several steps and experiment with the settings at each
step to get the right final result. Maybe even layer it and get out the
sharpener and eraser to selectively do it..

Mark Thomas wrote:
PeteD wrote:
http://1mpkzq.blu.livefilestore.com/..._8_g/Furry.jpg

Hmm. For a sensible viewing distance it's not bad. But looking
closely, there are bits (eyes and other high contrast edges) that are
showing haloes, yet other areas apparently in focus look soft and need a
bit more..

My guess is that if you were using PS USM, all the settings were a
little high, especially threshold (maybe 2-3 for a clean image like this).

Possibly it's a bit over sharpened with a larger radius like 2 or 3 (is
that what I see on the curly fiber in the lower-right?) but there's
still room to sharpen at a 0.3 pixel radius. I never really messed with
the threshold, does that prevent sharpening tiny little noise grain type
detail? If so maybe use that for the first larger radius round and
mask/erase at the smaller radius to bring out hairs & such.


It also depends a lot on how you got there - you may have lost some
sharpness in downsampling. I presume it is downsized? If so in one
hit? and what algorithm (Bicubic, Lanczos)? Maybe try a few steps, with
a small amount of sharpening each time. If it's a worthy image, I will
often downsize in several steps and experiment with the settings at each
step to get the right final result. Maybe even layer it and get out the
sharpener and eraser to selectively do it..


--
Paul Furman
www.edgehill.net
www.baynatives.com

all google groups messages filtered due to spam

On Sun, 21 Dec 2008 11:10:54 +1100, "PeteD" wrote in
:

http://1mpkzq.blu.livefilestore.com/..._8_g/Furry.jpg

Try Focus Magic on the original.
--
Best regards,
John
[Please Note: Ads belong (only) in rec.photo.marketplace.digital]

Paul Furman wrote:
Mark Thomas wrote:
PeteD wrote:
http://1mpkzq.blu.livefilestore.com/..._8_g/Furry.jpg

Hmm. For a sensible viewing distance it's not bad. But looking
closely, there are bits (eyes and other high contrast edges) that are
showing haloes, yet other areas apparently in focus look soft and need
a bit more..

My guess is that if you were using PS USM, all the settings were a
little high, especially threshold (maybe 2-3 for a clean image like
this).

Possibly it's a bit over sharpened with a larger radius like 2 or 3 (is
that what I see on the curly fiber in the lower-right?)
Yes, I think it is. Also in reverse around specular highlights on dark
areas.

but there's
still room to sharpen at a 0.3 pixel radius. I never really messed with
the threshold, does that prevent sharpening tiny little noise grain type
detail?
It determines how much contrast is required before the sharpening effect
kicks in - the lower the number, the lower the contrast difference has
to be before it decides it is an edge. So it depends on image content
and may need to be cranked up if there is a lot of noise, or for skin
tones where you don't want to exaggerate skin texture. If the image is
clean, and fairly low-contrast edges need sharpening (eg this image)
then I would drop it pretty low (I tend to work around 2-10 for most
images anyway). It's worth playing with, and it's pretty easy to spot
when you have dropped it too low.

If so maybe use that for the first larger radius round and
mask/erase at the smaller radius to bring out hairs & such.
Yep!

PeteD wrote:
http://1mpkzq.blu.livefilestore.com/..._8_g/Furry.jpg

It looks all right to me, depending on what you plan to do with it. It
reminds me of my childhood in West Texas in the 1930s. We would see
actual herds--hundreds or even thousands--crossing the highways; why I
do not know. Everyone tended to lift their feet off the floorboards of
cars when driving through these herds. Ughhhh!
Allen

John Navas wrote:
On Sun, 21 Dec 2008 11:10:54 +1100, "PeteD" wrote in
:

http://1mpkzq.blu.livefilestore.com/..._8_g/Furry.jpg

Try Focus Magic on the original.

I tried FM some time back and was unimpressed. I just visited their
site to see if anything had changed much or improved, and frankly, it
seems to have stood still. And of course it can't go very far because
of the physics involved. For any 'normal' image, with background and
foreground detail, deconvoluting will almost invariably introduce
unpleasant artefacts - you can see them all over the samples at their
own page. And they do not inspire confidence when they skew the
'comparisons' deliberately. On this page:
http://www.focusmagic.com/exampleunsharpmask.htm
...note the bit that says they use Threshold=0 for their USM comparisons?
That's *ridiculous*, especially for the first example (and I would
have quickly erased the the other examples in camera..). On that first
image I get a much better result from a single-pass USM *with more
sensible parameters* - try it yourself at say A=200 R=0.7 T=5. Less
artefacts, hardly any visible haloes/ringing and no increase in noise,
looks just as sharp. Happy to post result if asked.

Deconvolution can be very useful in astronomy, where the scenes are as
simple as possible (mainly point sources, dark background), and of
course they can help do little party tricks like reveal number plate
characters or clock faces when they appear to be lost in a blur. Motion
blur can certainly be ameliorated (my word of the day) reasonably well,
but o-o-f blur is less well handled, and the more complex the image or
o-o-f, the more problematic it becomes and the more artefacts are
introduced.

For 'normal' final sharpening, which usually involves an
already-in-focus image that just needs a little extra 'edge' (due to
bayer, downsampling, etc) I remain unimpressed and wouldn't pay anything
for such a program.

If arguing otherwise (ie in regard to normal sharpening), I'm happy to
see some *real world* FM sharpening examples, but they need to be
compared with the results gained using conventional methods like USM.
Post your results and the original, and let's see who wins...

Mark Thomas wrote:
Paul Furman wrote:
Mark Thomas wrote:
PeteD wrote:
http://1mpkzq.blu.livefilestore.com/..._8_g/Furry.jpg

Hmm. For a sensible viewing distance it's not bad. But looking
closely, there are bits (eyes and other high contrast edges) that are
showing haloes, yet other areas apparently in focus look soft and
need a bit more..

My guess is that if you were using PS USM, all the settings were a
little high, especially threshold (maybe 2-3 for a clean image like
this).

Possibly it's a bit over sharpened with a larger radius like 2 or 3
(is that what I see on the curly fiber in the lower-right?)
Yes, I think it is. Also in reverse around specular highlights on dark
areas.

but there's still room to sharpen at a 0.3 pixel radius. I never
really messed with the threshold, does that prevent sharpening tiny
little noise grain type detail?
It determines how much contrast is required before the sharpening effect
kicks in - the lower the number, the lower the contrast difference has
to be before it decides it is an edge. So it depends on image content
and may need to be cranked up if there is a lot of noise, or for skin
tones where you don't want to exaggerate skin texture. If the image is
clean, and fairly low-contrast edges need sharpening (eg this image)
then I would drop it pretty low (I tend to work around 2-10 for most
images anyway). It's worth playing with, and it's pretty easy to spot
when you have dropped it too low.

If so maybe use that for the first larger radius round and mask/erase
at the smaller radius to bring out hairs & such.
Yep!

The sharpening method you are all discussing is historic. It has little
value in producing well sharpened images today, other than to put image
faults and flaws where there were none before.

There are seriously better methods of sharpening an image than the
rather agricultural 'unsharp mask'.

I can't quite figure out how a portion of that image on the right side
is blurred but on the other side, at about the same distance from the
camera is not. Rita calls it "Tennessee blur". Has this picture been
artificially blurred with Photoshop?

On Sun, 21 Dec 2008 15:59:18 +1000, Jurgen wrote
in :

I can't quite figure out how a portion of that image on the right side
is blurred but on the other side, at about the same distance from the
camera is not. Rita calls it "Tennessee blur". Has this picture been
artificially blurred with Photoshop?

I've seen this happen from lens element decentering.
--
Best regards,
John
[Please Note: Ads belong (only) in rec.photo.marketplace.digital]

PeteD wrote:
http://1mpkzq.blu.livefilestore.com/..._8_g/Furry.jpg

For display on an LCD, possibly slightly.

For display on a CRT, where the high-frequency MTF is slightly poorer,
possibly not.

Excellent image!

Cheers,
David