Reading mode: Light Dark

Pixel density: when Moore is less - discuss!

Go to last post
J

jrdu

Forum Pro
Messages
21,371
Reaction score
135
Location
Huizhou, CN
Sometimes I feel I must be crazy, because if something doesn't make sense to me, I can't accept it, no matter if all the experts in the world disagree with me. I've always had this annoying characteristic even when I was a child. When I used to argue with the people on TV my mom used to worry about me, not just because I was talking to a box, but because TV was considered magical, almost God-like in its early years.

I can't see any problem with putting more and more megapixels in each sensor. More megapixels mean more details. It's all good.

However, all the expert reviewers on this site, continually bemoan the megapixel wars, and blame Sony most of all. They have taken to giving Sony cameras lower ratings primarily, I think, because Sony keeps packing in more megapixels.

Try this mental experiment: (I borrowed this idea from a thread I will share with you later). Imagine a one pixel camera. Low noise in that pixel, right? Not very much detail though. It can only show me one colour. Looking out my window, that would be grey, (sky).

Now imagine a four pixel camera with the same sensor size. There may be more noise in each of the pixels, but the overall amount of noise will be the same, because noise depends on the size of the sensor! However, the four pixel camera would be able to show that the sky is grey here, and the trees below are green. So more pixels give the same amount of noise but more detail in the image.
More pixels is a good thing.

Architecture is different from engineering, but engineering strives continually to approach the ideal of 100% implementation of architectural advantage. An 100 megapixel camera is architecturally more desirable than a 10 megapixel one, in the same way that a 10 pixel camera is better than a 1 pixel camera. Sony continues to lead the engineering challenge to meet that architectural ideal.

So, the new spec, pixel density, that Dpreview has introduced to their reviews does the opposite of what they intend it to do. It demonstrates their mistrust of new technology and propounds the foolish idea that smaller pixels create more noise in the image.

I was surprised to find that other some people have the same idea. For example read through this thread for a while: http://forums.dpreview.com/forums/read.asp?forum=1018&message=28315327
Maybe I'm not completely crazy...

ps - a more useful spec in the reviews would be a graphic showing the size of the sensor in relation to the sensors of other cameras. It's sensor size, not pixel size that's important.
--
John Dunn

Portraits: http://www.fototime.com/ftweb/bin/ft.dll/pictures?userid= {8B9B811D-AD1C-4A7D-923E-A4D0930BB5EE}
 
5 mega pixels is plenty for me. I get crisp 8 x 10 prints. That's all I can ask for. I own 2 cameras with higher mp ratings, but I keep going back to my Sony 717. I'm not sure just what you're doing with your photos, but 5 mp is enough for me.
--

 
with all respect its known that more pixels will give more noise its not brand Specific...just compare the output from a 5MP against a 8/9/10MP camera.

i think there is something else that need to be thrown into the discussion ....how does a brand processes or try's to eliminate the noise ...there it's going wrong.

some brands do a clearly better job on that point (less NR or different appraoch )

what kind of output u like depends mostly on personal taste.

for me the 5/6 MP camera's deliver the cleanest output in the P&S section....u know my shots ............i'll never apply any noise reduction with Neat image or programs like that.

today i will go out with my trusty H1 ( 5MP) and my Nikon P80 (10.1MP )

i like both camera's for different reasons.

same counts for my DSLR's.....

cheers Marti

--
  • living in harmony with nature and other beings...will create an better world for all * marti58 -2006
http://www.flickr.com/photos/marti58/
http://www.flickr.com/groups/worldwidefriendship/
 
jrdu said:
Sometimes I feel I must be crazy, because if something doesn't make
sense to me, I can't accept it, no matter if all the experts in the
world disagree with me. I've always had this annoying characteristic
even when I was a child. When I used to argue with the people on TV
my mom used to worry about me, not just because I was talking to a
box, but because TV was considered magical, almost God-like in its
early years.
I can't see any problem with putting more and more megapixels in each
sensor. More megapixels mean more details. It's all good.
However, all the expert reviewers on this site, continually bemoan
the megapixel wars, and blame Sony most of all. They have taken to
giving Sony cameras lower ratings primarily, I think, because Sony
keeps packing in more megapixels.
Try this mental experiment: (I borrowed this idea from a thread I
will share with you later). Imagine a one pixel camera. Low noise
in that pixel, right? Not very much detail though. It can only show
me one colour. Looking out my window, that would be grey, (sky).
Now imagine a four pixel camera with the same sensor size. There may
be more noise in each of the pixels, but the overall amount of noise
will be the same, because noise depends on the size of the sensor!
However, the four pixel camera would be able to show that the sky is
grey here, and the trees below are green. So more pixels give the
same amount of noise but more detail in the image.
Click to expand...
Sorry, you are totally wrong. I think you don't know the origins and sources of noise. Please try to read this:

http://theory.uchicago.edu/~ejm/pix/20d/tests/noise/index.html

Olaf
 
I also have difficulties to accept things if I don't understand the logic behind them. Also I am not too good in scientific theories and I try to simplify thins as much as I can otherwise I'm getting lost.

The way I see it is very simple:

More megapixels can give you more details

The more pixels you stuff into a certain sensor surface, the more noise you might get.

Therefore, the final amount of noise is not only related to the number of pixels but also to the sensor size.

I see it like trying to stuff 100 people into a small room, they won't be able to move. In order to improve the situation, you have either reduce the number of people or to enlarge the room.

Maybe it is too simplified but I think that is the way it works with pixels too.

In film camera days, If you wanted to produce more details, there were two ways to achieve it. Either use more fine grain films - which is equivalent to more pixels, or use larger format cameras - which is equivalent to larger sensor size. I believe that t works exactly the same way with digital photography.

So basically, you can already now have a 100 or 500 MP cameras, only, based on actual technologies, the sensors and the cameras would be very big and very expensive. I guess that future technologies will allow more pixels in slammer sensors with less noise but we are not there yet.

But again it is exactly like in film days - a 4x5 camera is much bigger than a 35mm one and it will give you much more details and a much better IQ.

All this lead me to believe that deep down there, not much has changed in photography. Of course, we use a different "film" and different terms and more bells and whistles and we don't have to wait anymore two weeks in order to see the results but basically, a lens is still a lens, an F stop is an F stop and photography is still photography...

just my 2c.

Cheers
Moti

--
ARPS - ABPPA
 
Marti58 said:
with all respect its known that more pixels will give more noise its
not brand Specific...just compare the output from a 5MP against a
8/9/10MP camera.
Click to expand...
With all due respect to you too Marti, the point that I'm making is that what "is known" is, in fact, wrong. The noise is related to the size of the sensor, not to the size of the pixel. When I compare the output from the H series cameras it keeps getting better as the megapixels go up. High ISOs get better and better. (Can you shoot in ISO 400 with your H1?). There's more detail and the noise gets more fine grained and less ugly as the megapixels rise.
John
 
Nobody needs to read the complete article linked to in my last posting. If you try to understand the paragraph concerning photon shot noise you will know, what happens on small sensors with too much pixels.

But I will try to give you another example, that's not the same situation, but contains a lot of what you have to understand:

Take a radio with one big antenna and replace it by four radios with small antennas. If the signal of the radio transmitter is strong possibly there's not much difference in the audio reception. But if the radio transmitter is weak you will get the better reception with one radio and a big antenna.

The antennas are the sensor pixels and the radios are the camera electronic needed to amplify the sensor signal and form an image.

Olaf
 
olafk said:
Sorry, you are totally wrong. I think you don't know the origins and
sources of noise. Please try to read this:

http://theory.uchicago.edu/~ejm/pix/20d/tests/noise/index.html
Click to expand...
Thanks, Olaf. I tried to read it. Here is an excerpt from that article:

An amusing extrapolation of the analysis of the effect of pixel size results from consideration of digicam raw data. The Panasonic FZ50 is a 10MP superzoom digicam that shoots raw files. An analysis by John Sheehy, as well as the author's rough measurements on this FZ50 raw file using the noise vs. exposure graphical method outlined on page 2, yields a gain g of roughly .29 photons/12-bit ADU at ISO 400. Dividing by the square of the 2 micron pixel size yields .072 photons per ADU per square micron, comfortably in the middle of the efficiency table above. However, just as the Canon ISO calibration was off for its earlier models, digicam ISO calibrations differ from those of DSLR's. Typical DSLR's leave about 3.5 stops between metered middle grey and raw saturation; digicams put middle gray about 0.5-1.0 stop closer to raw saturation, due to their lower dynamic range. This means that their ISO calibration is about 0.5-1.0 stops understated in relation to DSLR ISO calibration, and so the FZ50 efficiency figure is actually higher than .072 by a factor 1.4-2.0, making the FZ50 sensor about the most efficient per unit area in capturing photons of any digital camera sensor currently available! Of course the sensor is only about 5.5mm x 7.3mm in size, so the photon noise referred to the frame size is rather poor, as it is for any digicam; but the photon noise at fixed spatial scale rivals or betters the 1D3, 1Ds3 and D3. If the FZ50 sensor could be scaled up to the size of full frame, it would indeed rival these cameras for photon shot noise performance; and the resolution -- the 2 micron pixels translate into a 216MP (!) full frame camera (of course, for many applications the actual resolution will be limited by diffraction and lens aberrations). It is currently unclear whether this performance could be maintained as the sensor is scaled up by a factor of nearly five in linear dimension -- practicalities of supporting electronics for the sensels, speed in readout and processing, etc, may make it difficult to maintain the FZ50's performance in a scaled-up version.

The part of the excerpt separated by stars seems to show exactly what I am saying: That noise is related to sensor size, not pixel size. Thanks, Olaf, for helping me to prove my point.

--
John Dunn

Portraits: http://www.fototime.com/ftweb/bin/ft.dll/pictures?userid= {8B9B811D-AD1C-4A7D-923E-A4D0930BB5EE}
 
working with large format or medium format cameras? They are a bit cheaper to buy now than they used to be. Karsh, the famous Canadian portrait photographer always used a large format camera. The output is bound to be better than my pocket camera.
Do you think I should consider a Hassy?
--
John Dunn

Portraits: http://www.fototime.com/ftweb/bin/ft.dll/pictures?userid= {8B9B811D-AD1C-4A7D-923E-A4D0930BB5EE}
 
jrdu said:
olafk said:
Sorry, you are totally wrong. I think you don't know the origins and
sources of noise. Please try to read this:

http://theory.uchicago.edu/~ejm/pix/20d/tests/noise/index.html
Click to expand...
Thanks, Olaf. I tried to read it. Here is an excerpt from that
article:

An amusing extrapolation of the analysis of the effect of pixel size
results from consideration of digicam raw data. The Panasonic FZ50 is
a 10MP superzoom digicam that shoots raw files. An analysis by John
Sheehy, as well as the author's rough measurements on this FZ50 raw
file using the noise vs. exposure graphical method outlined on page
2, yields a gain g of roughly .29 photons/12-bit ADU at ISO 400.
Dividing by the square of the 2 micron pixel size yields .072 photons
per ADU per square micron, comfortably in the middle of the
efficiency table above. However, just as the Canon ISO calibration
was off for its earlier models, digicam ISO calibrations differ from
those of DSLR's. Typical DSLR's leave about 3.5 stops between metered
middle grey and raw saturation; digicams put middle gray about
0.5-1.0 stop closer to raw saturation, due to their lower dynamic
range. This means that their ISO calibration is about 0.5-1.0 stops
understated in relation to DSLR ISO calibration, and so the FZ50
efficiency figure is actually higher than .072 by a factor 1.4-2.0,
making the FZ50 sensor about the most efficient per unit area in
capturing photons of any digital camera sensor currently available!
Of course the sensor is only about 5.5mm x 7.3mm in size, so the
photon noise referred to the frame size is rather poor, as it is for
any digicam; but the photon noise at fixed spatial scale rivals or
betters the 1D3, 1Ds3 and D3. If the FZ50 sensor could be scaled up
to the size of full frame, it would indeed rival these cameras for
photon shot noise performance; and the resolution -- the 2
micron pixels translate into a 216MP (!) full frame camera (of
course, for many applications the actual resolution will be limited
by diffraction and lens aberrations). It is currently unclear whether
this performance could be maintained as the sensor is scaled up by a
factor of nearly five in linear dimension -- practicalities of
supporting electronics for the sensels, speed in readout and
processing, etc, may make it difficult to maintain the FZ50's
performance in a scaled-up version.

The part of the excerpt separated by stars seems to show exactly
what I am saying: That noise is related to sensor size, not pixel
size. Thanks, Olaf, for helping me to prove my point.
Click to expand...
I read this paragraph with much interest because I'm owning a FZ50. You are right: the most important point is the sensor size. But the problem today is that the typical 1/2.5" or 1/2.3" sensors are at their limit concerning pixel count.

Olaf
 
jrdu said:
The most efficient radio telescopes we have on earth are not one
large structure but they are arrays of many smaller ones.
Click to expand...
You are right, but the only reason to do that is the technical limitation to build one radio telescope with the same area. And the size of each of the smaller ones is not at the limit.

The problem with sensors is: you have to collect a minimal number of photons to get an information and if the pixels are too small the information is too imprecise.

Another point is the dynamic range. The smaller the pixels the smaller the dynamic range.

By the way - the pixel density is a useful information to compare cameras. It's not the only one but more useful than the megapixel count.

Thanks for opening this interesting discussion.

Olaf
 
broadly speaking, (over various types of sensors and software) once pixel size drops below a certain dimension, quality goes straight out the door.

The changes in image quality of the "H" series as the models progress, is as good as an example as you can find and well documented here.

Basically, density, sensor size and pixel count are just a way of establishing pixel size.

Would have been easier for DPR to show that size, or, a ratio of it VS the optimum size as a "quality" guide.
regards,
--
Ron

 
..but there are 2 factors that seem to compromise the theory in reality IMHO.

One of them is NR, as mentioned by Marti. Many people pixel peep (and/or crop) regardless of pixel density and pixel amount. Manufacturers know this and came up with NR. The downside of NR is unevenly spread detail (low contrast smearing) and artifacts. The larger the pixel density, the more NR required to make things viewable at 100%. Whether viewing at 100% is a good thing or not, is another discussion, but fact is, we have NR and the only way to turn it off is using RAW (if provided).

In short, reality is, that we crossed a point where NR is starting to limit (low contrast smearing and artifacts) output (detail), which in many cases, makes older cameras with less pixels the better choice for blowing up your pictures or cropping (this mainly goes for smaller sensor cameras, since they are equipped with more agressive NR than for example DSLR's). Ofcourse there is some tech advance that somewhat polishes away the effects mentioned above when comparing newer to older cameras, but the main advance is in NR algorithms at higher ISO's IMHO.

The second factor that compromises the theory you mentioned, is the fact that especially smaller cameras seem to have hit a certain wall. It reminds me of water. Water is good for your health. Drink a lot an your kidneys will be grateful. But up to a certain point. The endless NR battle most small sensor manufacturers are fighting is a testimony to that. Too much NR results in a smear fest, too little and people start complaining about grainy blue skies even on bright days. Sony isn't the only one fighting that beast for many years now, while being battered by many. I'm not sure what the technical consequences of hitting this "wall" are and whether there are any. But what I do see is a clear change in noise since we passed the 6MP border. In the 6MP cameras I mainly see chroma noise in blue bright skies. Easy to clean without affecting detail. Yet, those cameras didn't have much NR to begin with. Todays small sensor cameras either have more or less NR, but fact is they usually do show a little less detail at comparable sizes, but in most cases show a lot more luminance noise too. Even the most sophisticated external noise removal programs have a hard time cleaning that up without compromising low contrast detail beyond their 6MP (comparison after resampling) brothers and sister. That still puzzles me (I have both a 6MP and a 8MP superzoom) and doesn't fit the theory you mentioned.

Regards,

Jort
 
... Don't ask me - I don't know either.

What I do know is I can take my 717 out in very, very dull light and get an excellent image. In the exact same circumstances, I can't do that with the 828 or the H1.

Just pointing out, there are no easy answers and 'measuring' something is not always going to give the answer.

Pixel density can be measured, but how good are those pixels ? How well are they processed ? Etc., etc.

Not even sure this would be included as being important. I think the end result will be that folk end up confused. There will probably be whole thread 'wars' over it.

--
Rgds, Dave.
Have fun - take lotsa pix.
http://www.redbubble.com/people/pixplanet

S100fs Examples - http://www.pixplanet.biz/Posting-stuff_5.htm
 
yes, of course one tiny pixel standing in the middle of the Gobi desert, one tiny grain of sand, can't reflect or absorb much light by itself, but it doesn't have to. The dynamic range and the signal to noise ratio of entire sensor is what you have to look at and that is architecturally the same in a one pixel sensor an a hundred megapixel sensor.
--
John Dunn

Portraits: http://www.fototime.com/ftweb/bin/ft.dll/pictures?userid= {8B9B811D-AD1C-4A7D-923E-A4D0930BB5EE}
 
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