Reading mode: Light Dark

New pixel density stat on dpreview

Go to last post
DS_Dave said:
Thanks Gordon!

Although I like the crispier nose of the K20D anyway, but Ill give
the smudging a go later on for interest.
Click to expand...
The "smudging" was just for those that base their decision as to which is better only on the numbers and not on the appearance. I like finer "crisper" noise patterns, too, when the noise becomes visible at all.

Which is the problem with Noise Reduction (NR), as one has the choice of a minimum amount so that the size of the noise variation stays somewhere close to the resolution limits and thus appears fine and crisp, or hammer the noise reduction until you can't see any noise variation in flat textures and live with "cartoon-like" images; medium amounts of NR end up looking "splotchy" and worse than no NR at all (hammered NR could arguably be said to be worse, as well).

Regards, GordonBGood
 
Cameron R Hood said:
The Nikon D700 with a 12 mp full frame sensor scores a 1.4 - the
Pentax K20D with a 14.6 meg sensor at aps size gets a 4.0...

I quote from their introduction: "Pixel Density is a calculation of
the number of pixels on a sensor, divided by the imaging area of that
sensor."

This math just does not work. Compare the Nikon 700 with the 300, the
Pentax K20 and K10, and an Olympus or two in the camera comparometer
and you'll see; if this is truly just a calculation, their math is
just plain wrong. And it seems to be balanced in favour of the Nikons
and Canons (hmm....).

Thanks,
Cameron
Click to expand...
I'm afraid I have to disagree.

The Pentax K20D has an imaging area of 23.4 x 15.6 mm, according to Pentax.
That is 365.04 square mm. (3.65 square cm). And 14.6MP/3.65 is 4.0MP/cm2.

The D700 has an imaging area of 36 x 24mm, which is 864 square mm (8.64 square cm). 12.1MP/8.64cm2 is 1.4MP/cm2.

If you can show me the error in my working, please do.

Also, I'd like to add two further points:

1) It is not "balanced in favour" of anything, it is simply a way of looking at the physical dimensions of the sensor that eliminates the confusion that sensor sizes can introduce. We have no influence on what size or specification of sensor each manufacturer makes, we're simply reporting it.

More importantly, though:

2) This is NOT a measure, score or metric. It should not be used as a measurement of, or proxy for, image quality. It is an easy way of comparing cameras with different (obscure) sensor sizes and number of pixels. That's all.

Richard - DPReview
 
Hi Gareth!
Gareth_R said:
Great signature shot Jens.
Click to expand...
Thanks, not really a sig - would be too big anyway - just to illustrate what we might get if we just go out. Cause that's what I did for that shot, not planned, not dedicated to "have to shoot" something. Just me, my cycle, my camera and some light =) Oh, and some weeds ;)
And here endeth your daily dose of unsolicited philosophy.;)

Anyway. I think a fabulous pano of the Aoraki massive is the better signature shot :)

Cheers
Jens

--

'Well, 'Zooming with your feet' is usually a stupid thing as zoom rings are designed for hands.' (Me, 2006)
My Homepage: http://www.JensRoesner.de
 
DS_Dave said:
I agree completely Lance, so I did ;-)

So I downloaded the images and ran them through PS and Noise Ninja to
measure noise levels, here are the noise index's (lum, chroma) from
the top left corner of the wall:

K20D, native size: 36 (17,19)
K100D upsized to K20D: 18 (7,12)

K20D, downsized to K100D: 32 (14,18)
K100D, native size: 18 (7,11)

I got the same results when comparing my DS and K20D (see my response
to Steve below).

So unless my down sizing/upsizing technique is up the creek (I used
PS and selected what it recommended to use), the K20D has more
quanitifable noise, period. That said in this case the K100D
benefitted from a slightly brighter image to begin with but in my
tests my DS didn't. and it looks like the impact of up/downsampling
from 6-14.5 Mp doesn't have such a large impact on noise as some seem
to think.

HOWEVER (its a big one), there is a seriously large increase in
resolution with the K20D compared to the DS (yes, I use raw)...it is
very apparent. It blew my mind ;-)

ALSO the noise is crisp in the K20D (even after resizing)
compared to the soft "smudgy" noise of the K/D/DS/DL series (well it
used to look good before I got the K20D hehe!) even at native
resolution, let alone upsized.

As a result, the K20D noise is much more acceptable to me - it looks
like film grain more than that of the K/D series. I wont be giving
up my K20D ;-)

My 2c...
Click to expand...
And all perfectly valid, Dave. :-)

I was not necessarily poking holes in Anastigmat's theory, only in his example. He should have posted the photos at the same size which would have made the difference much less obvious.

As you say, the noise of the K20D is more "acceptable" and it looks closer to that of film, especially in the higher ISO's. There is no doubting the resolution advantage either, as you also rightly point out.

Having said that, the 6Mp sensor was a real beauty and I still love the photos I have from the *ist D as I am sure you are happy with the results from the DS as well. :-)
DS_Dave said:
Lance B said:
It might be an idea to compare the same size image with the same
size image, Anastigmat.
--
Lance B
http://www.pbase.com/lance_b

Click to expand...
--
GMT +9:30
http://dave.colourpixels.net/
Click to expand...
--
Lance B
http://www.pbase.com/lance_b

 
Yes the 6Mp sensor was a beauty!

I re-looked at the numbers using Gordons smudge (Gaussian blur) to make the K20D noise smudgy to match the appearance of the DS - it required a 0.3 radius blur (I was being kind to the K100D, a 0.4 was closer really). I confirmed this with my own images mentioned earlier.

Once again, Gordon was right...then the numbers change:

K100D, native size: 18 (7,11)
K20D, downsized to K100D: 32 (14,18)
K20D, downsized to K100D with 0.3 blur: 25 (10,15)
K20D, downsized to K100D with 0.5 blur: 22 (8,14)

Really there is nothing in it!

The REALLY BIG HOWEVER was the difference in resolution and response to sharpening...

With the K20D, even after a 0.5 radius blur (ie. being very mean to the K20D in terms of resolution, while reducing the noise to the K100D level) the larger words on the left bottle and detail in the leaves were still excellent without sharpening and the tones and detail were smmmoootheee. Without the blur the "made in England text was clear).

In contrast, the K100D image (left untouched) suffered from moire, and the text on the bottle "made in england" was completely unreadable (I knew it said that because of the K20D pic LOL) and even the rest of the larger stuff was barely readable. The edge details were jaggy.

Using a conservative smart sharpen made the difference even more obvious: the K20D was very detailed, while it didn't help the K100D that much. Scanning around the image revealed that it wasn't a focus issue either.

To summarise...one can get much more detail resolution with the same noise at 6Mp downsized K20D than the K100D. After a brief play, if you wanted to, I am sure using a 0.6 blur would remove the gap in "noise" but have only a minor impact on detail resolution.

I am very happy with the K20D :-)

Dave
Lance B said:
And all perfectly valid, Dave. :-)

I was not necessarily poking holes in Anastigmat's theory, only in
his example. He should have posted the photos at the same size which
would have made the difference much less obvious.

As you say, the noise of the K20D is more "acceptable" and it looks
closer to that of film, especially in the higher ISO's. There is no
doubting the resolution advantage either, as you also rightly point
out.

Having said that, the 6Mp sensor was a real beauty and I still love
the photos I have from the *ist D as I am sure you are happy with the
results from the DS as well. :-)
Lance B said:
Lance B said:
It might be an idea to compare the same size image with the same
size image, Anastigmat.
--
Lance B
http://www.pbase.com/lance_b

Click to expand...
--
GMT +9:30
http://dave.colourpixels.net/
Click to expand...
--
Lance B
http://www.pbase.com/lance_b

Click to expand...
--
GMT +9:30
http://dave.colourpixels.net/
 
So a higher score is better? The Canon G9 rates at 28...is this better than a 1DS Mk III at 2.4?

I just don't get the usefulness of this stat, no with disrespect intended. Simply trying to understand and discuss this.

Thanks,
Cameron
 
Nice philosophy and the shot is proof. Couldn't agree more though, I sometimes plan shoots and end up being dissappointed, good unsolicited shots are always a nice "bonus".

Was lucky with the Mt Cook pano, the weather was fantastic that day.

--



Annular grooved nails, ribbed for the wood's pleasure.
 
546xatnep said:
my intent was to point out a common over-simplification by prompting
a few questions, not to get answers to my questions ... I actually
knew the answers.
Click to expand...
Strange comment.
546xatnep said:
I tried to be clear that I was referring mostly to image noise, not
sensor noise ...
Click to expand...
Where does image noise come from?
546xatnep said:
they are different and image noise is what people
look at.
Click to expand...
It would be great if sensor noise can be separated from image noise. It cannot.
546xatnep said:
Still, I don't agree that more pixels = more sensor noise
necessarily.
Click to expand...
That is a tricky statement to make. I would not say that more pixels = more noise either. First of all, noise is dependent on a number of factors, including pixel size. More pixels does not mean more sensor noise if the sensor is larger. A 16mm full frame sensor has more pixels than a 14mp APS-C sensor, but the full frame has bigger pixels. So, yes, more pixels can mean lower noise at the same time, if the pixels are bigger too. Further, Canon has its proprietary noise reduction patents that SOny and Nikon do not. So, Canon's sensors can be lower in noise despite having smaller pixels. Besides, noise can also be reduced by using noise reduction software. So, it is indeed possible for a sensor with lots of pixels to have low noise, albeit it may have less resolution than a sensor with fewer pixels due to noise reduction.
546xatnep said:
It seems to me that there are different types of noise
and each sensor design will be affected differently and mitigate
these noise sources differently.

I know that I'm no expert, but ... do we know each other?
Click to expand...
Of course you are no expert. I can tell that.
546xatnep said:
While the K20D may have received a "highly recommended" rating in
DPReview, it has not received such a rating everywhere. A recent
review I read in a magazine rated the Canon 450D higher than the K20D
for completely absurd reasons (in my opinion) .. image noise at high
ISO's was a major reason which I found extremely misleading.
Click to expand...
Personally I prefer a camera that has low noise at high ISO settings. I would give the Canon 450D higher score over the K20D on that count too.
 
Hi Dave!

I don't know how much time you have, but this might be worth a thread of its own. The physical principle is known, but many people look at "100%" crops and then of course, the 6MP APS-C sensor will be much better than the 14MP APS-C sensor.

Well, just printing both at the same size would illustrate the noise-mitigating effect of downsampling.

Cheers
Jens

--

'Well, 'Zooming with your feet' is usually a stupid thing as zoom rings are designed for hands.' (Me, 2006)
My Homepage: http://www.JensRoesner.de
 
It is a bit of a hijack of the OP's thread isnt it?

Did you want this on a new thread...I could possibly post the images and details etc to show the point?

I might try printing a section of each at the same size, say a 2x3" section of an 8x12" sized (not resampled of course, and untouched) full image...might be interesting Jens.

Regards,

Dave
--
GMT +9:30
http://dave.colourpixels.net/
 
Thanks. Of course it is good for me. I like the 6mp APS-C cameras because of their low noise at ISO levels up to 1600, without resorting to destructive levels of noise reduction. The only meaningful upgrade to me is a full frame with similarly sized pixels as these cameras. I have saved a lot of money by not getting on the APS-C upgrade treadmill. A lot of people think they can get better image quality by replacing one APS-C camera with another having more but smaller pixels, but what they are doing is expending a lot of money without getting anything useful in return. They are wasting their time and energy, just like the hamster that is running on a treadmill. LOL.
*isteve said:
Anastigmat said:
I think the K100D image is a lot cleaner. Resolution of the two
images are about the same, but the K100D image is crisper and better
looking. Conclusion: I am keeping my 6mp Pentax bodies. TTL flash
support and clean high ISO images! They just don't make them like
that anymore.
Click to expand...
Good for you.

--
Steve
When I can master technique I'll be a photographer.
When I can realise a vision I'll be an artist.
When I get paid I'll be a professional.
Click to expand...
 
Just reading the poster comments, and not checking them out might not get you the right answer...

Image 1 100% crop
Resized the K20D test shot to K100D size.
Gaussian blur 0.4
Smart sharpen (0.2 rad, 100%)

Image 2 100% crop
Original K100D test shot

I tried some smart sharpen to get the detail back, but it didn't help much, and the noise went higher than the K20D, so I left it alone.

Notes:

The brighter exposure favours the K100D - less noise. Yet the K20D still has more detail and about the same noise.

Sure the leaf in the K20D shot appears to miss some detail, but it doesn't really - what you are seeing is the effect of the lower exposure.

The detail in the resized K20D without the gaussian blur is such that one can read the text plainly. Try for yourself. Selective NR will be great with this...

Dave





--
GMT +9:30
http://dave.colourpixels.net/
 
Hi Dave!
DS_Dave said:
It is a bit of a hijack of the OP's thread isnt it?
Click to expand...
I don't know - the thread is about comparing different pixel densities and that's what you did.
DS_Dave said:
Did you want this on a new thread...I could possibly post the images
and details etc to show the point?
Click to expand...
"Want" sounds so strong :)

I just thought that this might (should?) be interesting to more people than the number who will find it here.
DS_Dave said:
I might try printing a section of each at the same size, say a 2x3"
section of an 8x12" sized (not resampled of course, and untouched)
full image...might be interesting Jens.
Click to expand...
I always wanted to do this myself, I probably would have reverted to the dpreview test shots and have them printed for evaluation. But I'm too lazy ;)
EDIT:

Also interesting: 6MP-camera-shot vs. processed 10MP vs. in-camera 10MP-> 6MP, printed to the same size. And no, I'm not saying you should do this :)

Cheers
Jens

--

'Well, 'Zooming with your feet' is usually a stupid thing as zoom rings are designed for hands.' (Me, 2006)
My Homepage: http://www.JensRoesner.de
 
Hi Jens,

Well, I'll see about a thread. There are a number of things with the K20D that I'd like to bring up for wider input.
JensR said:
Also interesting: 6MP-camera-shot vs. processed 10MP vs. in-camera
10MP-> 6MP, printed to the same size. And no, I'm not saying you
should do this :)
Click to expand...
Sorry Jens, I cant help you there, I only have the 14.6Mp K20D ;-)

Dave

--
GMT +9:30
http://dave.colourpixels.net/
 
... For the very first time, I don' t understand why DPR wrote such a report.

To me, but I may be wrong, DPR should have simply add the pixels density in the respective camera review.

Moreover, their chart has too many other information not useful for a pixels density "analysis". All of these are already published somewhere else in the review and don't add anything related to the original purpose.

And... I don't get why all the major cameras produced are not there? A lighter chart with only pixels density and noise graph about all the camera tested by DPR would be more useful for us, no?

Do we need DPR to divide the pixels quantity by the sensor area? I am saying that because I believe that beginners won't care about pixels density and the others (the DPR aficionados) may be interested in a context of a wider comparison.

My humble opinion.
 
You must log in or register to reply here.

Keyboard shortcuts

f
Forum
m
My threads
Mobile site
About
Editorial content
Cameras & Lenses
Community
All content, design, and layout are Copyright © 1998–2025 Digital Photography Review All Rights Reserved.
Reproduction in whole or part in any form or medium without specific written permission is prohibited.
When you use DPReview links to buy products, the site may earn a commission.
©GPS Media - Guides, Products, Services.
Back
Top