Too many megapixels? Six is enough for an 8x10.

Started Apr 26, 2009 | Discussions
Ergo607
Ergo607 Senior Member • Posts: 1,354
You're just talking noise...

You're just talking noise in all this. As if that was the most important factor. I couldn't care less... Talk about DR and come back then...

Daniel Browning wrote:

[...continued from part 1]

Unequal sensor sizes.

It's always necessary to consider the impact of sensor size. The most
common form of this mistake goes like this:

1. Digicams have more noise than DSLR.
2. Digicams have smaller pixels than DSLR.
3. Therefore smaller pixels cause more noise.

The logical error is that correlation is not causation. It can be
corrected by substituting "sensor size" for "pixel size". It is not
the small pixels that cause the noise, but small sensors.

A digicam-sized sensor with super-large pixels (0.24 MP) is never
going to be superior to a FF35 sensor with super-tiny pixels (24 MP).

Unequal processing.

The most common mistakes here are to rely on in-camera processing
(JPEG). Another is to trust that any given raw converter will treat
two different cameras the same way, when in fact none of the
commercial ones do. For example, most converters use different
amounts of noise reduction for different cameras, even when noise
reduction is set to "off".

Furthermore, even if a raw converter is used that can be proven to be
totally equal (e.g. dcraw), the method it uses might be better suited
to one type of sensor (e.g. strong OLPF, less aliases) more than
another (e.g. weak OLPF, more aliases).

One way to workaround this type of inequality is to examine and
measure the raw data itself before conversion, such as with IRIS,
Rawnalyze, dcraw, etc.

It's important to be aware of inequalities that stem from processing.

Unequal expectations.

If one expects that a camera that has 50% higher resolution should be
able to print 50% larger without any change in the visibility of
noise, despite the same low light conditions, then that would be
unequal expectations. On the other hand, if one only expects to it be
at least print the same size and the same noise for the same low
light, then that would be equal expectations. Such output size
conditions are arbitrary and in any case does not support the "small
pixels are noisier" position.

Unequal technology.

If you compare a 5-year-old camera to a 1-year-old camera, it will
not be surprising to find the new one is better than the old one. But
in one sense, it will never be possible to compare any two cameras
with completely equal technology, because even unit-to-unit
manufacturing tolerances of the same unit will cause there to be
inequalities. It's common to find one Canon 20D with less noise than
another Canon 20D, even if absolutely everything else is the same.
Units vary.

I don't think that means we should give up on testing altogether,
just that we should be aware of this potential factor.

So that summarizes the reasons why I think the myth has become so
popular. Here is some more information about pixel density:

http://forums.dpreview.com/forums/read.asp?forum=1034&message=31584345
http://forums.dpreview.com/forums/read.asp?forum=1032&message=16107908
http://forums.dpreview.com/forums/read.asp?forum=1029&message=21440105
http://forums.dpreview.com/forums/read.asp?forum=1029&message=23296470
http://forums.dpreview.com/forums/read.asp?forum=1019&message=31512159
http://forums.dpreview.com/forums/read.asp?forum=1018&message=30211624

A paper presented by G. Agranov at 2007 International Image Sensor
Workshop demonstrated that pixels sizes between 5.6 and 1.7 microns
all give the same low light performance.

http://www.imagesensors.org/Past%20Workshops/2007%20Workshop/2007%20Papers/079%20Agranov%20et%20al.pdf

Eric Fossum said that FWC tends to increase with smaller pixels:
"smaller pixels have greater depth (per unit area) and saturate
'later in time'".
( http://forums.dpreview.com/forums/read.asp?forum=1000&message=30017021 )

So the question might arise: what should be considered with regard
to pixel density? There are at least three things to consider:

  • File size and workflow

  • Magnification value

  • Out-of-camera JPEG

File size is an obvious one. Magnification is what causes telephoto
(wildlife, sports, etc.) and macro shooters to often prefer high
pixel density bodies (1.6X) over FF35.

Out-of-camera JPEGs are affected by pixel density because
manufacturers have responded to the throngs of misguided 100% crop
comparisons by adding stronger noise reduction. If JPEG is important
to you and you can't get the parameters to match your needs, then it
becomes an important factor.

Higher pixel densities require bigger files, slower workflow, longer
processing times, higher magnification for telephoto/macro. For me
this is not a factor, but it may be important to some shooters. Lower
pixel densities result in smaller files, faster workflow, and lower
magnification.

I'm sorry this post is so long, I did not have time to make it shorter.

Noise scales with spatial frequency.

 Ergo607's gear list:Ergo607's gear list
Pentax *ist D Pentax K-3 Pentax K-1 II Pentax smc FA 50mm F1.4 Pentax smc DA 10-17mm F3.5-4.5 ED (IF) Fisheye +5 more
misha marinsky3 Senior Member • Posts: 1,330
Re: All true essentially

bobn2 wrote:

Except that the new high MP cameras seem to outperform, in general,
the old lower MP cameras. Annoying when the truth gets in the way of
a good myth, isn't it
--
Bob

The latest Canon G10 is fine at its base ISO, but above 400, yuck.

True of all digicams, due to a small sensor, not small pixels. most
tellingly, the G10 outperforms the previos model at higher iSO's.

The smaller the diode, the higher the amplification.

Complete nonsense. Learn some microelectronics, then look at how the
things actually work. Then you'll know that is rubbish.

DPReview lists pixel density in their database; it is significant.

It's significant in that it illustrates DPR's agenda on this, which
is mainly to never, ever admit that Phil Askey has ever made a
mistake. He has made a big one here because he simply didn't
understand the technology well enough, and now DPR continues to
propagate this myth to defend that position.

Look, I don't want a forum argument. I did some google searching, and found this:
http://theory.uchicago.edu/~ejm/pix/20d/tests/noise/noise-p3.html#pixelsize

"The issue arises because, in general, smaller pixels are noisier...If one compares two cameras having the same sensor size and technologies but different resolutions (pixel densities), the one with the lower resolution (lower pixel density) will have less noise at the pixel level...The bigger pixels of the 1D3 collect more photons than the 40D, and this is reflected in the S/N ratio plots above..."

If this is wrong, provide the link so I can learn. Just don't tell me off - I want to learn. Don't be petulant.

My websites:

http://www.flickr.com/photos/ubereye/

http://ubereye.deviantart.com/

http://newyorkleftist.blogspot.com/

bobn2
bobn2 Forum Pro • Posts: 61,414
Re: Too many megapixels? Six is enough for an 8x10.

Anastigmat wrote:

bobn2 wrote:

So we'll pitch your view about the 'tones' against quantitative
measurements taken by experts and the subjective opinions of may
professional photographers. Right.

Pros are unlikely to use the D40, so they are also unlikely to have
an opinion about it. Besides, many pros are just as ignorant about
photo equipment as you or I.

or him, for that matter. I'd trust properly conducted measurements over some bloke's opinion any day.

I wonder who are the gullible.... Those who keep upgrading to new
cameras with more but smaller pixels thinking that they will get
better image quality because of the increased pixel count. Of
course, there are definitely enough of these consumers to keep the
upgrade market busy and healthy.

Well, they will get better image quality because of the increased pixel count. I would agree, that for many the improvement might not be enough to justify the expenditure, but there are some who alwways have to have the latest and best.

So, you pay oodles of money for a 15-24mp sensor, so you can
downsample it to match the quality of a 6mp sensor?

It doesn't cost any more, and gives better quality when downsampled. True, new models sell at a premium over older ones, but that has little or nothing to do with increased cost of the sensor.

Why not save the
money and use it for something better?

What extra money?

I am not against higher
resolution, but not at the expense of increased noise.

Thats good, because it never seems to give increased noise.

Besides, downsampling is not easy:

Dead easy if you've got a Canon with sRAW. Wasn't it you who was pointing out how much better 10MP Canon sRAW was than 12MP NEF? Or maybe it was Taikonaut - I get confused sometimes.

'Before answering the question in detail of "does it work and how
well" (yes it does, and very well) we need to understand that
"binning" has been done before, but rarely with a colour matrix
sensor and never very well. Monochrome sensors have been binned, but
doing so with a Bayer matrix is a monumentally more difficult task
because each adjacent pixel is of a different colour and has a quite
different spectral response. Also, because binning reduces resolution
by a factor of four, it really doesn't make a lot of practical sense
to attempt unless and until one has a sensor such as the one in the
P65+, with 60 megapixels unbinned. The resulting 15 megapixel binned
size is bigger than many pro DSLRs, such as the Nikon D3 and D700,
and therefore quite usable for many commercial applications.'

'Binning' is cahrge accumulation on the sensor, which I don't believe any camera does (could be wrong, I'd be happy to be informed better). But essentially, you are right. Rather than demosaicing and downsampling, it would be much better to have a downsampling raw converter (which is, I suppose, what sRAW is). I wonder why there isn't one?
(end part 1)
--
Bob

bobn2
bobn2 Forum Pro • Posts: 61,414
Re: Too many megapixels? Six is enough for an 8x10.-part 2

(part 2)

Bigger sensors allow the individual pixels to be big even though
there are many of them. It is the big individual pixels that can
detect a weak signal (while at the same time retain the ability to
record a strong signal without the signal clipping) that allows big
sensors to have good dynamic range. Cram a lot of tiny pixels onto a
medium format sensor, and you would get low dynamic range. Further,
medium format is not dying. Sales has slowed, and will get worse
because of the slow economy, but new products are being introduced.
Leica and Pentax are planning new medium format cameras.

The size of the pixel is irrelevant. People seem to focus on it. Actually a small pixel is more sensitive and can 'detect a weaker signal' in that it converts a smaller number of photons to a given output voltage, but then it needs to be, because it has fewer photons incident. In the end, the sensitivity and size of pixel balances out, and it is only sensor area that matters.

Want to put a bet on the commercial success of the forthcoming Leica and Pentax, if they ever make it at all? The similar Mamiya hasn't fared too well, and there's no great reason why the Leica or Pentax should do a lot better. The Leica probably worse, since it will cost an arm and a leg, and hasn't a legacy lens system to support it. There were also a lot fewer Pentax MF systems around than Mamiyas, so it wouldn't seem to have a much higher chance of survival. The Mamiya couldn't compete against the 1DsIII. I don't expect the newcomers to be able to either, and they've got the D3x, A900 and probably D700x to compete with as well.

Zero logical argument and very poor choice of 'experts'. Why would
you choose the opinion of a web developer over the fellow who
invented the active pixel sensor?

Just because others disagree with you does not mean that their
arguments are not logical.

I wasn't basing my assessment of illogicality on my disagreement with them, just on their lack of logic.

Images are built up by individual pixels. Look at a newspaper image.
The individual pixels are awful in quality. That is why you don't
get a good image from the page of a newspaper. Look at a magazine
page, the images are much better because the individual pixels making
up that image are better than those found on a newspaper.

The 'quality' of the pixels is exactly the same, but they are smaller and there are more of them, that adds up to a better image overall.

You cannot
separate pixel qualiity from image quality. Pixels are the
ingredients of an image. Bad ingredients will not a good image make.

Nice metaphor, but it doesn't work. Small pixels aren't 'bad', they are smaller. They capture an image with more precision, just as you can make a finer casting with a fine sand mould than you can with a course grain one.

ii) Doing theoretical assessments at the pixel level and failing to
analyze the way in which pixels aggregate to form an image.

How do they "aggregate" except by being placed next to each other.
One pixel should not bleed its signal to its adjacent pixel. Pixels
do not mix. A good image depends on each pixel maintaining its
integrity and its goodness. If you start out with inferior
individual pixels, you will end up with an inferior image.

So many flaws in the reasoning here. Firstly the small pixels are not 'inferior'. Secondly they do indeed mix. No-one should view an image so you can see the individual pixels. If you can, you are enlarging it too much. When the pixels are below the limit of individual visibility, their effect does indeed mix. You chose the right example with magazine print. Those are terrible 'pixels' just 1 bit, on or off. But they are small enough to be individually invisible and the effect of their aggregation is anything but on or off. More precisely, what you need to know about aggregation is that signal adds but noise adds in quadrature, so as you aggregate tiny pixels together, the noise averages, naturally and without 'noise reduction'

Small pixels are not necessarily bad. However, if they get too small
so that the signal becomes too weak, then the S/N ratio will become
too low for good dynamic range and good image quality.

What is 'too small' Where is the onset of this effect? How does the signal become 'weak'? What is the effect on SNR? Have you any real examples of this effect?

Noise also
becomes a problem. The D90 relies on strong doses of noise reduction
at ISO 400 and above to keep noise down.

In JPEG, yes. That seems to be Nikon's style (not one I agree with, but leads to good reviews amongst naive 100% pixel peeping reviewers). There's also something funny about it's raw noise curve, but I don't think that's NR, I think it's the unwinding of the trick they use to get best in class noise at low ISO's.

OTOH, the D40's Sony sensor
is well known to be practically noise free at ISO 400. NR is only
needed at ISO 1600 and above in the D40. Alas, to get the D40's
better sensor, you have to put up with an inferior camera body.

The D40's sensor is in no respect better than the D90's. Loo at the DxO ratings, or Clarkvision, or anyone who has actually done some quantitative measurements on them.

To
get the D90's better camera body, you would have to accept a noisier
sensor. That is life. It is full of compromises.

Full of untruths, too.

bobn2
bobn2 Forum Pro • Posts: 61,414
Re: It's all gone quiet over there...

ender21 wrote:

People respond a lot better when you talk across to them and not DOWN
at them. Perhaps the "wrong" ones aren't really the fools. (I know
I know, THEY started it... waaah).

Difficult to talk up when they're down there.

Even people that are right can have zero credibility, and your
proficiency with data (I know this because you're reciting your
resume every other post)

I recited it once to someone who accused me of not knowing what I was talking about. I do, and I demonstrated it. The fact tha he doesn't, and doesn't have a CV to back up his implied claim of expertise is his problem.

is only exceeded by your humiliating
deficiencies in actual human interaction.

Try following the posts if you want an exercise in human interaction. I correct untruths, simply and to the point. I don't talk down or insult. It's the insecurity of those who want to pose as experts and misinform people when they have little realkunderstanding who take correction as an insult, and then they start hurling the insults and corrections, and I can play that game as well as them.

It's made me wish for an
Ignore feature in these forums, despite my agreeance with 90% of your
scientific points.

There is an ignore feature.

Thankfully others have much better delivery than you

Agreed, I am in awe of some of them.

so I can feel free to consciously ignore you anyway.

Your privilege.

Work on that part of your enormous cranium

I've never claimed to have an enormous cranium

and you'll have something new to brag about and
arrogantly proclaim in front of completely anonymous people whom
you'll never meet, yet whose opinions you insecurely address so
routinely. Or as your profile plan states, misinformation in need of
YOUR correction.

So it would be better to let untruths go, to let these people peddle misinformation, for fear of unsettling their insecure little souls? And why don't you apply these principles to the people who first post this nonsense with zero supporting evidence?

Let me let you into a secret, the OP, which I responded to, was a trolling post. He knew full well the controversy of the issue and his post was designed to stir up that controversy. He was saying to all the people who actually know what's what 'you're wrong, because ABC and Gupta say you are'. Well, I've gained a taste for fresh troll, if he steps across a bridge I'm near, he'll get eaten.
Fol-de-rol.

-- hide signature --

Bob

bobn2
bobn2 Forum Pro • Posts: 61,414
that,

misha marinsky3 wrote:

bobn2 wrote:

Except that the new high MP cameras seem to outperform, in general,
the old lower MP cameras. Annoying when the truth gets in the way of
a good myth, isn't it
--
Bob

The latest Canon G10 is fine at its base ISO, but above 400, yuck.

True of all digicams, due to a small sensor, not small pixels. most
tellingly, the G10 outperforms the previos model at higher iSO's.

The smaller the diode, the higher the amplification.

Complete nonsense. Learn some microelectronics, then look at how the
things actually work. Then you'll know that is rubbish.

DPReview lists pixel density in their database; it is significant.

It's significant in that it illustrates DPR's agenda on this, which
is mainly to never, ever admit that Phil Askey has ever made a
mistake. He has made a big one here because he simply didn't
understand the technology well enough, and now DPR continues to
propagate this myth to defend that position.

Look, I don't want a forum argument. I did some google searching, and
found this:
http://theory.uchicago.edu/~ejm/pix/20d/tests/noise/noise-p3.html#pixelsize

"The issue arises because, in general, smaller pixels are
noisier...If one compares two cameras having the same sensor size and
technologies but different resolutions (pixel densities), the one
with the lower resolution (lower pixel density) will have less noise
at the pixel level...The bigger pixels of the 1D3 collect more
photons than the 40D, and this is reflected in the S/N ratio plots
above..."

If this is wrong, provide the link so I can learn. Just don't tell me
off - I want to learn. Don't be petulant.

I'm only petulant when abused, and you haven't done that. On that post, Emil wrote it, I'll let him explain what he meant. I wouldn't have written it like that, because I think it's misleading to imply that the shot noise originates in the pixel. Fair enough, in the way pixel noise is usually analysed, but people do get the wrong end of the stick. That's why I've taken to saying that the shot noise is in the image not in the pixel .

-- hide signature --

Bob

Oly Canikon
Oly Canikon Senior Member • Posts: 1,278
Re: All true essentially

misha marinsky3 wrote:

bobn2 wrote:

Except that the new high MP cameras seem to outperform, in general,
the old lower MP cameras. Annoying when the truth gets in the way of
a good myth, isn't it
--
Bob

The latest Canon G10 is fine at its base ISO, but above 400, yuck.

True of all digicams, due to a small sensor, not small pixels. most
tellingly, the G10 outperforms the previos model at higher iSO's.

The smaller the diode, the higher the amplification.

Complete nonsense. Learn some microelectronics, then look at how the
things actually work. Then you'll know that is rubbish.

DPReview lists pixel density in their database; it is significant.

It's significant in that it illustrates DPR's agenda on this, which
is mainly to never, ever admit that Phil Askey has ever made a
mistake. He has made a big one here because he simply didn't
understand the technology well enough, and now DPR continues to
propagate this myth to defend that position.

Look, I don't want a forum argument. I did some google searching, and
found this:
http://theory.uchicago.edu/~ejm/pix/20d/tests/noise/noise-p3.html#pixelsize

"The issue arises because, in general, smaller pixels are
noisier...If one compares two cameras having the same sensor size and
technologies but different resolutions (pixel densities), the one
with the lower resolution (lower pixel density) will have less noise
at the pixel level...The bigger pixels of the 1D3 collect more
photons than the 40D, and this is reflected in the S/N ratio plots
above..."

It's the picture not the pixel that matters

If this is wrong, provide the link so I can learn. Just don't tell me
off - I want to learn. Don't be petulant.

I agree lets keep it friendly

The conclusion from that article helps with the context:

Bottom line: Among the important measures of image quality are signal-to-noise ratio of the capture process, and resolution. It was shown that for fixed sensor format, the light collection efficiency per unit area is essentially independent of pixel size, over a huge range of pixel sizes from 2 microns to over 8 microns, and is therefore independent of the number of megapixels. Noise performance per unit area was seen to be only weakly dependent on pixel size. The S/N ratio per unit area is much the same over a wide range of pixel sizes. There is an advantage to big pixels in low light (high ISO) applications, where read noise is an important detractor from image quality, and big pixels currently have lower read noise than aggregations of small pixels of equal area. For low ISO applications, the situation is reversed in current implementations -- if anything, smaller pixels perform somewhat better in terms of S/N ratio (while offering more resolution). A further exploration of these issues can be found on the supplemental page. Rather than having strong dependence on the pixel size, the noise performance instead depends quite strongly on sensor size -- bigger sensors yield higher quality images, by capturing more signal (photons).

The other main measure of image quality is the resolution in line pairs/picture height; it is by definition independent of the sensor size, and depends only on the megapixel count. The more megapixels, the more resolution, up to the limits imposed by the system's optics.

Scales USA Veteran Member • Posts: 3,121
Re: Too many megapixels? Six is enough for an 8x10.

I was really amazed by the IQ improvement when I bought my son a 12 mp Canon A-650 IS a year ago. My wife has a 8mp A-710 IS and there is a big difference.

That said, my 3MP Nikon Coolpix 990 takes amazing photos as well, as long as you don't to crop them.

 Scales USA's gear list:Scales USA's gear list
Fujifilm MX-700 Canon G1 X II Canon EOS R
ejmartin Veteran Member • Posts: 6,274
Re: All true essentially

misha marinsky3 wrote:

bobn2 wrote:

Except that the new high MP cameras seem to outperform, in general,
the old lower MP cameras. Annoying when the truth gets in the way of
a good myth, isn't it
--
Bob

The latest Canon G10 is fine at its base ISO, but above 400, yuck.

True of all digicams, due to a small sensor, not small pixels. most
tellingly, the G10 outperforms the previos model at higher iSO's.

The smaller the diode, the higher the amplification.

Complete nonsense. Learn some microelectronics, then look at how the
things actually work. Then you'll know that is rubbish.

DPReview lists pixel density in their database; it is significant.

It's significant in that it illustrates DPR's agenda on this, which
is mainly to never, ever admit that Phil Askey has ever made a
mistake. He has made a big one here because he simply didn't
understand the technology well enough, and now DPR continues to
propagate this myth to defend that position.

Look, I don't want a forum argument. I did some google searching, and
found this:
http://theory.uchicago.edu/~ejm/pix/20d/tests/noise/noise-p3.html#pixelsize

"The issue arises because, in general, smaller pixels are
noisier...If one compares two cameras having the same sensor size and
technologies but different resolutions (pixel densities), the one
with the lower resolution (lower pixel density) will have less noise
at the pixel level...The bigger pixels of the 1D3 collect more
photons than the 40D, and this is reflected in the S/N ratio plots
above..."

If this is wrong, provide the link so I can learn. Just don't tell me
off - I want to learn. Don't be petulant.

Well, I hope it is not wrong, since I wrote that. Indeed the "bottom line" conclusion quoted by Oly states my view of the situation. I have given links to and an explanation of a more sophisticated analysis using noise spectra in my other posts in this thread, based on an analysis I did last Fall when this issue was stirred up by Phil's blog post. I hope to find the time to incorporate these later findings in the noise article at some point.

-- hide signature --
strawman_John Contributing Member • Posts: 967
Bad example

The example linked to is more about ADC system noise than sensor noise. you see there is so little signal what you are seeing is noise in the ADC system and electronics, no surprise a similar result because that test by John takes out the sensor as a significant input.

I can and have done a simple test. Here is examples of why you need no more than 6mp for a 10x8 print. I took the same view with the same lens at the same ISO (ISO 100) on a 300D and a 40D. Printed at 10x8 I could not see any difference in resolvable noise and detail in the print.

In fact do it your self in photoshop if you want. just work out the DPI and get back to us when you discover most inkjets struggle to resolve more than 240 DPI. Even at 300 DPI you are struggling, and remember the lower pixel count image has sharper per pixel results too. Go check out DPREVIEW test results if you want confirmation.

I also have a Canon 570 IS that is of a similar vintage to my 40D. It clearly has lower dynamic range and is noisier, so much so that ISO1600 on the 40D is quieter than the compacts ISO200. But the compact can make good 10x8 prints, in good light.

So to take your arguments apart, I can see the logic of for 10x8 prints about 6mp marks the point of no more benefit for more pixels. In fact as you make the pixels smaller the dynamic range falls off. The noise you can combat, to an extent by applying more filtration, swapping resolution for lower noise.

And if it helps, I do have an electronics degree, a couple of patents and have even designed imaging equipment, so I get the maths behind all this also.

So if you want to convince me that for a given sensor technology step smaller pixels do not mean more noise and lower dynamic range, and that for a given print size more pixels always make for a better picture (once you have factored in lens limitations also) please send me the maths.

strawman_John Contributing Member • Posts: 967
Re: It's all gone quiet over there...

Bob

Reply to my last post, I think it shows you are wrong.

Sorry.

strawman_John Contributing Member • Posts: 967
Evidence.

OK lets take the D700 and D300 same MP about the same level of sensor technology. Now does the D700 produce lower noise at a given ISO, of course it does can I now have my 50$.

Or take the Canon 50D and the 5D MII, the 5D MKII clearly produces lower noise. In fact take the 5D MKII, and even though it suffers from being older technology, it has lower noise than the 50D.

Plenty of evidence that a larger pixel area is better.

You just need to take care that you compare similar levels of silicon development.

strawman_John Contributing Member • Posts: 967
Re: Nikon have some good examples

bobn2 wrote:

The 12MP pixel count for the D3 is not set by the requirement for
high ISO, it is set by the requirement for high speed.

Do you knowthat or are you making an assertion. I think it is for both.

The D3 has been optimised for high ISO in three ways:
i) the sensor is unusually efficient in converting photons to
electrons. This is probably due to effective 100% microlenses.

It might also be due to the fact that with a larger surface area it collects more photo's as well. Is that unusual, no it s a result of the simple fact that bigger pixel area leads to a greater ability to capture photons, its not difficult to grasp unless you have an ilogical ability to not recognise the advantages of greater pixel area.

So, the relative performance has everything to do with the detail of
the design and nothing to do with pixel density.

Can you prove it, I mean prove that last statement. Not just say it, can you show me by maths or actual measurements that the difference in light collection area is not significant.

If not, can they post examples of where (same brand camera / same
chip size) denser pixels reduces noise? Cos I'd like to see that

No-one ever said higher pixel density reduces noise, we said it
didn't increase it.

I think with Sonys A350 and A300 you can see more pixels increasing noise. Read the DPREVIEW results. Then go look at the D50 review again. Now go do the D300 and D700 test and see how the extra pixel area helps.

ejmartin Veteran Member • Posts: 6,274
Re: Evidence.

strawman_John wrote:

OK lets take the D700 and D300 same MP about the same level of sensor
technology. Now does the D700 produce lower noise at a given ISO, of
course it does can I now have my 50$.

Noise as measured how? By pixel std dev of levels in uniform tonality image regions? That is a skewed measure.

Or take the Canon 50D and the 5D MII, the 5D MKII clearly produces
lower noise. In fact take the 5D MKII, and even though it suffers
from being older technology, it has lower noise than the 50D.

Ditto.

Plenty of evidence that a larger pixel area is better.

You just need to take care that you compare similar levels of silicon
development.
--

50D and 5D2 collect about the same amount of photons per unit sensor area. Their photon shot noises will therefore be about the same. If you want we can have a detailed discussion of the correlation or lack thereof of pixel size and read noise, the trend of which is ISO dependent. Did your statements apply to high ISO (say 1600) or low ISO (say 200)?

Note also that your examples are comparisons of crop sensors to FF. I suspect that all the folks who contend that pixel size is not a strong factor in image noise, will readily agree that sensor size is a strong factor in image noise.

-- hide signature --
Daniel Browning Senior Member • Posts: 1,058
Same goes for dynamic range and several other things

Ergo607 wrote:

You're just talking noise in all this. As if that was the most
important factor. I couldn't care less... Talk about DR and come back
then...

Everything said applies to dynamic range as well.

In linear image sensors, dynamic range is simply a function of noise and light. Noise is determined by quantum efficiency (QE, for the photon shot noise) and electronics (for the read noise). Light is determined by the full well capacity and QE.

Full well capacity and QE scales with pixel size, so the only other factor for dynamic range is read noise. Most discussions of dynamic range are with regards to ample-light scenes such as outdoors, studio, etc., so we'll assume ample light and base amplification. We'll use 1:1 S/N as the lower end of the dynamic range.

Compare the 2-micron pixels of the LX3 (10.7 stops DR) with the immensely larger 6.4 micron pixels of the 5D2 (11.1 stops DR). Going by the per-pixel numbers, it seems that the smaller LX3 pixels have less dynamic range. But remember that the LX3-sized pixel samples a much, much higher spatial frequency. [See math below if you want the numbers.]

At the same spatial frequency, the scaled LX3 pixels have 12.3 stops of dynamic range, 1.2 stops greater.
--
Daniel

5D2 maximum signal: 52,300 e-
LX3 maximum signal: 9,000 e-

5D2 read noise at base ISO: 23.5 e-
LX3 read noise at base ISO: 5.6 e-

5D2 per-pixel DR at base ISO: 11.1 stops (log_2(52300/23.5))
LX3 per-pixel DR at base ISO: 10.7 stops (log_2(9000/5.6))

LX3 scaled maximum signal: 92200 (9000 e- * (6.4µm/2.0µm)^2)
LX3 scaled read noise at base ISO: 17.92 (sqrt(5.6 e-^2 * ((6.4µm/2.0µm)^2)))
LX3 scaled DR at base ISO: 12.3 stops (log_2(92200/17.92))

strawman_John Contributing Member • Posts: 967
Re: There is no big manufacturers conspiracy . . .

bobn2 wrote:

Small pixels need no more amplifier gain than large ones.

Do you actually know that or are you saying it. For a given efficiency of micro lens, and the same desired ADC voltage, the larger pixel will capture more photons. Care to share with us how the maths of photons to signal measured by a sensor works.

Please show the evidence to back your assertion.

ejmartin Veteran Member • Posts: 6,274
Re: Nikon have some good examples

strawman_John wrote:

bobn2 wrote:

The D3 has been optimised for high ISO in three ways:
i) the sensor is unusually efficient in converting photons to
electrons. This is probably due to effective 100% microlenses.

It might also be due to the fact that with a larger surface area it
collects more photo's as well. Is that unusual, no it s a result of
the simple fact that bigger pixel area leads to a greater ability to
capture photons, its not difficult to grasp unless you have an
ilogical ability to not recognise the advantages of greater pixel
area.

The Panasonic LX3 records .106 electrons per raw level per square micron (e- µ^2 per raw level) at ISO 400 with 2µ pixels.

The Nikon D3 records .115 e- µ^2 per raw level at ISO 400 with 8.46µ pixels.

When properly looked at as densities of electrons per unit area rather than per pixel, a given patch of sensor of a given size records about the same number of photons (or photo-electrons) independent of the pixel size. That is because the photon sensitivity is a property of the semiconductor materials used, and doesn't care how the photosensitive surface is partitioned into pixel-sized bins. One 8.46µ pixel or 18 2µ pixels, it's the same area and the same number of photons collected, to a very good approximation.

As for well capacity,

The Panasonic LX3 saturation count at base ISO 80 is 2250 e- µ^2
The Nikon D3 saturation count at base ISO 200 is 920 e-
µ^2

Thus the D3 actually saturates at lower illumination than a rather good P&S. I think this supports Bob's contention that the D3 is optimized for high ISO.

(Note: the Nikon data are due to measurements on RAW data taken and analyzed by Peter Facey; the LX3 data is my own analysis of RAW images taken by others according to my specifications)

I'm done replying to your posts in this thread. I think your forum name says it all.

-- hide signature --
strawman_John Contributing Member • Posts: 967
Re: Evidence.

ejmartin wrote:

strawman_John wrote:

OK lets take the D700 and D300 same MP about the same level of sensor
technology. Now does the D700 produce lower noise at a given ISO, of
course it does can I now have my 50$.

Noise as measured how? By pixel std dev of levels in uniform
tonality image regions? That is a skewed measure.

Why? The reason I use it is that it shows that a large surface area gives an advantage when the pixel count is equal. It is something that photographers can observe so is important.

Or take the Canon 50D and the 5D MII, the 5D MKII clearly produces
lower noise. In fact take the 5D MKII, and even though it suffers
from being older technology, it has lower noise than the 50D.

Ditto.

Plenty of evidence that a larger pixel area is better.

You just need to take care that you compare similar levels of silicon
development.
--

50D and 5D2 collect about the same amount of photons per unit sensor
area. Their photon shot noises will therefore be about the same. If
you want we can have a detailed discussion of the correlation or lack
thereof of pixel size and read noise, the trend of which is ISO
dependent. Did your statements apply to high ISO (say 1600) or low
ISO (say 200)?

I did a quick check and a 5D MKII has a pixel density not much different from a 20D/30D. Would you care to show where it is calculated that the 50D and 5D MKII have equal photon collection levels per pixel. I agree the 50D has better micro lenses, perhaps the interesting comparison is the 500D that looks to have a micro lens similar to the other sensors. My statement of course applies to high ISO, where we worry about noise. For a modern SLR, ISO200 noise is not much of an issue.

Note also that your examples are comparisons of crop sensors to FF.
I suspect that all the folks who contend that pixel size is not a
strong factor in image noise, will readily agree that sensor size
is a strong factor in image noise.

I view that size of pixel , efficiency of micro lenses, and electrical performance of ADC etc are important as well as MP count. The best sensors balance all these items out.

For a given print size there comes a point where adding more pixels makes no difference to the print resolution. Achieved somewhere in the 240 to 300 DPI range. And there is a point where making the pixels smaller brings the system to the point for a given light level that signal to noise ratio gets too bad and noise signals dominate. The greater the pixel density the quicker you will hit this.

strawman_John Contributing Member • Posts: 967
Simple question

At low light levels where noise is important. For a given sensor technology, and micro lens technology, can you tell me that having a bigger sensor area does not help gather light.

what you say has elements that are correct, but they do not contradict or negate my point. you are just indicating the many variables. Of course there are sensitivity differences in the various silicon technologies, but does that negate the impact of area?

strawman_John Contributing Member • Posts: 967
Insults never a good thing

It sort of shows that you are not able to continue the debate in a reasoned manner.

Keyboard shortcuts:
FForum MMy threads