k20D "fine sharpness" SUCKS! Too much noise!

rburgoss · 06-21-2008, 12:22 PM

Still playing with my new K20D and trying to discover what REAL changes every single mode or adjustable setting will actually do.

Well, to my surprise, the FINE SHARPNESS feature, explained in page 155 (item 5) of the manual, when activated, indeed produces a bit sharper images, but with unacceptable noise levels.

See the results:

Attachment 14296

No doubt the "chain" looks sharper on the left image, but the noise level at the smooth wall paint is a lot higher. Also note the "edge" (left in diagonal). On the left shows like a highlight that "pretends" to render a sharper image, but in reality, such highlighted edge does not exist (see right image)

Anyone has noticed such?

RB

PentaxPoke · 06-21-2008, 01:00 PM

You may want to be a little more guarded in making such a sensational claim in your thread title before you understand the process you are criticizing. You should not have expected any different with the example you posted. You enlarged to 100% crop, a dark shadow region in the photo. That is always going to be the noisiest part of a digital image, even at 100 iso, and you were at 400.

Sharpening can be thought of as a "derivative process" In other words, it uses variations in light levels to highlight. If you sharpen something that has noise, then by default, the result will be noisier. Smoothing is exactly the opposite. It is an "integration process" where light variations are "averaged" in some respect to reduce noise.

The sharpening did exactly what it was supposed to do: it increased the local contrast of your image based on the variations in local light levels. You can't expect your camera to sharpen and smooth at the same time. If you want to be able to sharpen more while minimizing noise, you need more signal to noise ratio. In other words, more exposure (light). The problem is that if you upped the exposure of this photo, you would probably have blown out the clouds. Such are the compromises we have to make.

This particular shot is an example of where fill flash would have been helpful, but that is for another thread...

*Wheatfield* · 06-21-2008, 01:02 PM

With sharpening comes noise. With more sharpening comes more noise. Your original looks a bit noisy to begin with, I'm not surprised by your result.

**Peter Zack** · 06-21-2008, 01:31 PM

I completely agree with PentaxPoke & Wheatfield. It is excatly as it should be given the image, exposure and ISO settings. I'll assume that this was also shot as a Jpeg in camera as well. Since Jpeg is not a lossless format, you will always see artifacts in a processed image whether done in camera or on a computer.

cputeq · 06-21-2008, 03:41 PM

For these types of operations, doing it on PC with quality software is going to yeild infinitely better results than in-camera, regardless of brand.

That being said - Sharpness always yeilds noise. Always*.

*-- Unless you're masking and sharpening in Photoshop or something instead of full-image sharpening.

ckanthon · 06-21-2008, 05:18 PM

Originally Posted by rburgoss

Still playing with my new K20D and trying to discover what REAL changes every single mode or adjustable setting will actually do.

Well, to my surprise, the FINE SHARPNESS feature, explained in page 155 (item 5) of the manual, when activated, indeed produces a bit sharper images, but with unacceptable noise levels.

See the results:..

No doubt the "chain" looks sharper on the left image, but the noise level at the smooth wall paint is a lot higher. Also note the "edge" (left in diagonal). On the left shows like a highlight that "pretends" to render a sharper image, but in reality, such highlighted edge does not exist (see right image)
Anyone has noticed such?
RB

Hi,
Your information given were too limited for people to study. Please let us know more information such as the focal length, settting of the lens, the Aperature (F/?) and the exposure time etc.

First feeling is that when you apply the "sharpeness on", then you have got JPEG artifact. (you can get rid of this by taking RAW)

I suspect what K20d did for you were:

Sharpeness ON: The amount degree of noise reduction was reduced (to retain more details) + edge sharpening process was applied on that picture .

Sharpeness OFF: Noise reduction process took more action + in fact your picture was a little bit out of focus or somehow a very little bit "blur" (at least, see the chain).

If it were not slighty out of focus, your 14M pixels camera could get a photo even more sharper (even without the shapeness process turned ON). I saw such effect on my K200d.
(attached two pictures with 100% crop)

RiceHigh · 06-21-2008, 09:00 PM

No Text! (Why I have to enter something here?)

ricardobeat · 06-21-2008, 09:48 PM

Originally Posted by rburgoss

Still playing with my new K20D and trying to discover what REAL changes every single mode or adjustable setting will actually do.

Well, to my surprise, the FINE SHARPNESS feature, explained in page 155 (item 5) of the manual, when activated, indeed produces a bit sharper images, but with unacceptable noise levels.

See the results:

Attachment 14296

No doubt the "chain" looks sharper on the left image, but the noise level at the smooth wall paint is a lot higher. Also note the "edge" (left in diagonal). On the left shows like a highlight that "pretends" to render a sharper image, but in reality, such highlighted edge does not exist (see right image)

Anyone has noticed such?

RB

Not the camera's fault. Do the same on any image editor and you'll get similar results. Sharpening always affects noise, and too much of it creates halos/highlights. Learn how and when to use it.

rburgoss · 06-22-2008, 06:59 AM

Of course, every response is right. I have no doubts that any sharpenning (via software, in camera or PP) will increase noise levels, especially in shadow areas. My point is (I should have stated it in my original post) is that my istD is not as noisy using the same settings, except of course, the final resolution.

Yes, indeed I used JPEG capture mode (finest ****) at 14 megapix. Exposure was with FA 16-45 at f/16 and 1/320 sec. ( I used close aperture because my original testing was gonna be about sharpness). Focal length was at full wide (16 mm).

BTW, I also agree that shooting RAW will give the best possible results, but to be honest, I'm a bit lazy for doing PP.

Thanks to everyone for your responses.

RB