Sharpness

Pete Perry said:

I'll always use Normal after my tests... I personally didn't like
the noticable loss of detail in landscape shots from using none or
low. As for High, it's alright but the images can tend to look a
bit too sharp on the edges and auto seems to provide excellent
results but also presents an issue with Noise on some of my
pictures.

Click to expand...

I took a bunch of test shots this morning outside and I pretty much concur with you, Pete. However, I couldn't find a subject that liked the High setting.......it's just too much sharpening. The average overall best photos were taken at settings of Normal and Auto. Auto seems to be just as good as Normal for Jpeg's, at least it is for the existing conditions here today. Whatever. As far as I'm concerned, I'll stick to shooting exclusively shooting in RAW format.

Craig

What you've provided are crops from larger JPGs, cut and pasted into a new image, and resaved as another JPG. That's double JPG compression, which creates new problems.

It looks to me that JPG compression (even once, definitely twice) takes a significant toll on the contrast of black against the "railroad crossing sign" yellow of the paper.

My best guess is that people looking at this image zoom in on the "5" in the Permit No., noticing the separation of the stroke in the lower half of the numeral. The reason this is clearly separated in the High case, and blurred in the others, is that High creates a phony white "halo" in the separation, which JPG accurately compresses. Whereas in the none, low and normal sharpening, you don't have the fake halo artifact, so JPG is still asked to compress a contrast between the black and the yellow -- which it can't seem to manage here.

Sorry, but High only looks better in this case because it has a phony white "halo" artifact that has created a "detail" of close-to-white that isn't really there, but happens to be very JPG friendly. I wouldn't draw any conclusions about best JPG settings for landscapes from this trial.

I think you'll be disatisfied with high sharpening images, once you've seen enough of them.

Craig Robinson said:

Jerry Monroe said:

Interesting test. I have one comment and one question.

Although the HIGH image makes the lettering much sharper than the
others, the sharpening appears to have made a mess out of the
orange background.

Click to expand...

The photos are macro-shots of an envelope, so I think on the HIGH
sharpening photo, you are seeing the texture of the paper. I just
did some sharpening of the originals in Capture and the results are
closer together than they were in the first test, but the in-camera
HIGH photo still beats the other three. I'll be interested to see
if NEFrodite can get the LOW setting to beat it without introducing
artifacts.

Craig

Click to expand...

After going outside earlier today and taking some more jpeg test shots, I couldn't agree with you more. It is readily apparent that an in-camera sharpening setting of Normal or even Auto [for the conditions this morning] yielded the highest quality jpeg's(after post-processing). I couldn't seem to scratch out as much detail from the Low setting like I can in RAW, though.

The crops I provided you were saved the first time at the max setting of 12 and the second time at a setting of 10. Since all of them were saved the same way, I assumed(wrongly?) that what was good for one would be good for the rest. I could provide you with the individual crops that were saved just the one time at the max, but there is no need for it. Your assessment of my unscientific test sounds just right to me. All this testing was somewhat illuminating, but since I do all of my shooting in RAW mode, it means little to me. But, then again it's still nice to know. Thanks for your input, for taking the time to review the results and for trying to improve on them.
I'll remove those meaningless images now.

Craig

Why not leave the images where the forum can find them, so that people catching up with the forum for the next couple of days can see what we've been talking about?

I think it's a good example of how High sharpening can be deceptively seductive.

Craig Robinson said:

After going outside earlier today and taking some more jpeg test
shots, I couldn't agree with you more. It is readily apparent that
an in-camera sharpening setting of Normal or even Auto [for the
conditions this morning] yielded the highest quality jpeg's(after
post-processing). I couldn't seem to scratch out as much detail
from the Low setting like I can in RAW, though.

The crops I provided you were saved the first time at the max
setting of 12 and the second time at a setting of 10. Since all of
them were saved the same way, I assumed(wrongly?) that what was
good for one would be good for the rest. I could provide you with
the individual crops that were saved just the one time at the max,
but there is no need for it. Your assessment of my unscientific
test sounds just right to me. All this testing was somewhat
illuminating, but since I do all of my shooting in RAW mode, it
means little to me. But, then again it's still nice to know.
Thanks for your input, for taking the time to review the results
and for trying to improve on them.
I'll remove those meaningless images now.

Craig

Click to expand...

NEFrodite said:

Why not leave the images where the forum can find them, so that
people catching up with the forum for the next couple of days can
see what we've been talking about?

Click to expand...

Too late, I already removed them. But you're right. I'll put them back up.

Craig

NEFrodite said:

Why not leave the images where the forum can find them, so that
people catching up with the forum for the next couple of days can
see what we've been talking about?

Click to expand...

D100 jpeg in-camera sharpening results:
The unprocessed photos http://www.pbase.com/image/8867359
The processed photos http://www.pbase.com/image/8867288
Other test shots http://www.pbase.com/wicopy/test_2

These are all good examples of "What Not To Do".

Craig