Pages: [1] 2   Go Down

Author Topic: Megapixel Count, Megapixel Density x Camera Size  (Read 12302 times)

Peter_DL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 544
Megapixel Count, Megapixel Density x Camera Size
« on: January 27, 2011, 05:04:26 pm »


Just a thought,
don’t take it too serious:

The Megapixel count was a strong sales proposition for many years:  the more, the better. Then, we realized the downsides: smaller pixel-wells typically mean more noise, less dynamic range and an inferior high ISO performance (without intending to ignore technological progress in this regard). Actually some camera product lines already made a step 'backwards' to less Megapixel and a lower Megapixel density.  So we started to watch the MP/cm2 index e.g. at dpreview (at least speaking for myself).

Another valid question is about the size of a camera - maybe not decisive for everyone, but important for some of us.  Hence, the logic suggests to multiply both numbers which are essentially "the lower the better":

Megapixel Density   x  Camera Dimensions

Some examples are given below.

Even when ignoring the size and weight of DSLR lenses, and without intending to advocate for the S95, its Sensor Size seems to be more in line with the Camera Size compared to the other cameras. Maybe nothing new, however, somehow it makes a nice unit: Megapixel-centimeter.

Commets ?

Peter

__

Canon D60:
5.4 MP/cm2  x (145 x 106 x 79 mm) =
5.4 MP/cm2  x (14.5 x 10.6 x 7.9 cm) =  6557 MP cm

Canon G12:
23 MP/cm2  x (112 x 76 x 48 mm) =
23 MP/cm2  x (11.2 x 7.6 x 4.8 cm) =  9397 MP cm

Canon 1D Mark IV:
3.1 MP/cm2  x (156 x 156.6 x 80 mm) =
3.1 MP/cm2  x (15.6 x 15.66 x 8.0 cm) =  6059 MP cm

Canon S95:
23 MP/cm2  x (100 x 58 x 30 mm) =
23 MP/cm2  x (10.0 x 5.8 x 3.0 cm) =  4002 MP cm

Logged

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto
Re: Megapixel Count, Megapixel Density x Camera Size
« Reply #1 on: January 28, 2011, 02:30:56 am »

Hi!

Contrary to common belief, more uis actually better. Reason is that photography is dominated by shot noise and shot noise cares much about sensor size but little about if the pixels are small or large within reasonable limits. What is reduced with small pixels is dynamic range which is affected by readout noise, but even that is less than expected.

With smaller pixels there is less need of AA-filtering which reduces aliasing artifacts.

Check this excellent article: http://www.dxomark.com/index.php/en/Our-publications/DxOMark-Insights/More-pixels-offset-noise!

Lloyd Chambers made a comparison between Nikon D3 and D3X and arrived at a similar conclusion. That said, Nikon D3X has much better DR than D3 due to sensor implementation and D3 has much better high ISO performance that D3X also due to sensor implementation. So, Nikon D3 is better in the dark and Nikon D3X is better for anything else.

Best regards
Erik


Just a thought,
don’t take it too serious:

The Megapixel count was a strong sales proposition for many years:  the more, the better. Then, we realized the downsides: smaller pixel-wells typically mean more noise, less dynamic range and an inferior high ISO performance (without intending to ignore technological progress in this regard). Actually some camera product lines already made a step 'backwards' to less Megapixel and a lower Megapixel density.  So we started to watch the MP/cm2 index e.g. at dpreview (at least speaking for myself).

Another valid question is about the size of a camera - maybe not decisive for everyone, but important for some of us.  Hence, the logic suggests to multiply both numbers which are essentially "the lower the better":

Megapixel Density   x  Camera Dimensions

Some examples are given below.

Even when ignoring the size and weight of DSLR lenses, and without intending to advocate for the S95, its Sensor Size seems to be more in line with the Camera Size compared to the other cameras. Maybe nothing new, however, somehow it makes a nice unit: Megapixel-centimeter.

Commets ?

Peter

__

Canon D60:
5.4 MP/cm2  x (145 x 106 x 79 mm) =
5.4 MP/cm2  x (14.5 x 10.6 x 7.9 cm) =  6557 MP cm

Canon G12:
23 MP/cm2  x (112 x 76 x 48 mm) =
23 MP/cm2  x (11.2 x 7.6 x 4.8 cm) =  9397 MP cm

Canon 1D Mark IV:
3.1 MP/cm2  x (156 x 156.6 x 80 mm) =
3.1 MP/cm2  x (15.6 x 15.66 x 8.0 cm) =  6059 MP cm

Canon S95:
23 MP/cm2  x (100 x 58 x 30 mm) =
23 MP/cm2  x (10.0 x 5.8 x 3.0 cm) =  4002 MP cm


Logged
Erik Kaffehr
 

Ray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 10365
Re: Megapixel Count, Megapixel Density x Camera Size
« Reply #2 on: January 28, 2011, 05:08:16 am »

I'm not sure it's particularly useful to equate pixel size with camera size. Also, because of evolving technology, there doesn't seem to be a strong or consistent relationship between pixel size and pixel quality, with respect to SNR and dynamic range.

The D7000 has a pixel size of less than 5 microns yet, as regards dynamic range, together with the Pentax K5, it seems to have the highest DR of any camera on the market, including all DBs (although it has to be said that the true DR of many DBs remains a trade secret).

If Nikon were to produce a full frame 35mm sensor consisting of D7000 pixels, it would be 40mp with extraordinary performance.

It's true, according to DXOMark, that the larger pixel of the Nikon D3 (8.4 microns as opposed to 4.78 microns for the D7000) seems to have marginally better tonality and color sensitivity than the D7000. However, when the larger file size of the D7000 is downsized to the D3 file size, those marginally better specs regarding SNR at 18% grey, tonal range and color sensitivity, disappear.

But the DR of the D7000 is significantly better than that of the D3 at both the pixel level and the downsized, normalised print size.

These facts demonstrate that evolving technology is a more significant factor for pixel performance than pixel size; within reason of course. No P&S pixel comes even close to the pixel quality of the best DSLRs or DBs.

Logged

Peter_DL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 544
Re: Megapixel Count, Megapixel Density x Camera Size
« Reply #3 on: January 28, 2011, 03:39:26 pm »

Contrary to common belief, more uis actually better. Reason is that photography is dominated by shot noise and shot noise cares much about sensor size but little about if the pixels are small or large within reasonable limits. What is reduced with small pixels is dynamic range which is affected by readout noise, but even that is less than expected.
With smaller pixels there is less need of AA-filtering which reduces aliasing artifacts.

Hi,

Not sure if I get the point.

More Megapixel are certainly good, but typically not when it is done on a fix area and same sensor size. It just reduces the pixel size and increases pixel density. The assumption was that a low Megapixel Density as reported by DPR is basically something good, ranging from 50 MP/cm2 with some P&S down to 2.x MP/cm2 in the direction of MF cameras. Wrong ?

However, thanks for the DxO link.



These facts demonstrate that evolving technology is a more significant factor for pixel performance than pixel size; within reason of course. No P&S pixel comes even close to the pixel quality of the best DSLRs or DBs.

Valid concern.

The basic idea and question probably was, if it wouldn’t be possible to combine a Camera’s ground image quality (IQ) with its size to one single Rating Parameter.

One option could be to include a division by a Camera Sensor-evolution Index, maybe corresponding to the DxO Sensor Overall Score:

(Megapixel Density x Camera Dimensions) / DxO Sensor Overall Score


The lower the final number, the better.

Makes any sense ?

Peter

--
Logged

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto
Re: Megapixel Count, Megapixel Density x Camera Size
« Reply #4 on: January 28, 2011, 05:34:55 pm »

Hi,

There are essentially two factors, noise and DR. Noise consists of several parts:

- Shot noise, random variation of photons, this is a property of light and statistics
- Thermal noise (not very important)
- Readout noise (significant in the deepest shadows)
- Pixel nonuniformity (which I have not looked into)

Shot noise is essentially proportional to the square root of the number of captured photons. Let's assume that a small area on the print is eight covered by either:

- a large pixel detecting 32000 photons
- four small pixels containing 8000 photons each

Either way the area on the print would be described by 32000 photons, so shoot noise would be the same. SNR (Signal Noise Ratio for shot noise would be 179). So shot noise would be the same on both.

If photon count is low, readout noise would play a significant role. DR (dynamic range) is a signal processing term that is defined as maximum signal, divided by readout noise. If we assume a readout noise of three electrons and one electron per photon the dynamic range of the aforementioned pixels would be:

- 13.3 steps for the large sensor
- 11.4 steps for the small sensor

This is per pixel. The small pixel sensor would have four times the pixels, however.

DxOmark has two different values. One is actual pixels and the other is "print". In print mode the values are scaled for an 8x10 print corresponding to 8 MPixels. The DxO mark is based on the "print" values so it is normalized to 8 MPixel. Simply enough, DxO is essentially image quality. The DxO-mark figure is based on three scenarios.

- Landscape which is essentially DR
- Portrait which is color sensibility
- Low light which is based on usable ISO

The enclosed figures compare SNR och DR in "screen" and "print" mode between Nikon D3X (24.5 MP full frame) and Nikon D3s (12 MP full frame).

The reason that the Nikon D3X has better DR than the D3s at low ISO is less readout noise. The D3s curves are typical for a system with relatively high readout noise.

Best regards
Erik



Hi,

Not sure if I get the point.

More Megapixel are certainly good, but typically not when it is done on a fix area and same sensor size. It just reduces the pixel size and increases pixel density. The assumption was that a low Megapixel Density as reported by DPR is basically something good, ranging from 50 MP/cm2 with some P&S down to 2.x MP/cm2 in the direction of MF cameras. Wrong ?

However, thanks for the DxO link.



Valid concern.

The basic idea and question probably was, if it wouldn’t be possible to combine a Camera’s ground image quality (IQ) with its size to one single Rating Parameter.

One option could be to include a division by a Camera Sensor-evolution Index, maybe corresponding to the DxO Sensor Overall Score:

(Megapixel Density x Camera Dimensions) / DxO Sensor Overall Score


The lower the final number, the better.

Makes any sense ?

Peter

--

« Last Edit: January 28, 2011, 05:36:35 pm by ErikKaffehr »
Logged
Erik Kaffehr
 

Ray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 10365
Re: Megapixel Count, Megapixel Density x Camera Size
« Reply #5 on: January 28, 2011, 06:59:00 pm »


The basic idea and question probably was, if it wouldn’t be possible to combine a Camera’s ground image quality (IQ) with its size to one single Rating Parameter.

One option could be to include a division by a Camera Sensor-evolution Index, maybe corresponding to the DxO Sensor Overall Score:

(Megapixel Density x Camera Dimensions) / DxO Sensor Overall Score


The lower the final number, the better.

Makes any sense ?

Peter

--



Not much. If it is meaningful to relate basic image quality to camera size, then one should also include camera weight which, for some of us who go trekking, is more important than camera volume.

However, the choice of camera is likely to be dominated by the range, quality and cost of available lenses, as well as the weight of such lenses.

Weight and size are the two qualities about which there can be little dispute. I've never seen a thread on this site arguing about the claimed weight and size of any piece of photographic equipment. Weight and size are the easiest specifications to appreciate and understand.

I don't think it is useful to conflate such clear and precise information as weight and size, with the much more nebulous and endlessly disputed qualities of sensor and lens performance.

Logged

Peter_DL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 544
Re: Megapixel Count, Megapixel Density x Camera Size
« Reply #6 on: January 29, 2011, 07:07:43 am »


...

DxOmark has two different values. One is actual pixels and the other is "print". In print mode the values are scaled for an 8x10 print corresponding to 8 MPixels. The DxO mark is based on the "print" values so it is normalized to 8 MPixel. Simply enough, DxO is essentially image quality...

Hi,

Great explanations. Many thanks.
Wouldn't it also mean to accept that for example the sensor / "print" quality of a Sony 580 is the same as with a Canon 1Ds Mk III. Both cameras show a DxO Overall Score of 80. Hmm...


Logged

Peter_DL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 544
Re: Megapixel Count, Megapixel Density x Camera Size
« Reply #7 on: January 29, 2011, 07:16:23 am »

Not much. If it is meaningful to relate basic image quality to camera size, then one should also include camera weight which, for some of us who go trekking, is more important than camera volume.

However, the choice of camera is likely to be dominated by the range, quality and cost of available lenses, as well as the weight of such lenses.

Weight and size are the two qualities about which there can be little dispute. I've never seen a thread on this site arguing about the claimed weight and size of any piece of photographic equipment. Weight and size are the easiest specifications to appreciate and understand.

In a first order the Camera Dimensions may correspond to the Weight. But then, at same Camera Dimensions a higher Weight can speak for a more solidly built body. So there's a possible disconnect regarding the "the lower the better". At the end of the day, the Camera Dimensions seemed more important for us, because it finally determines pocket or bag in which the camera is carried.

In all the discussions about a best possible trekking camera, for example when a GH2 + lenses is suggested, some people will say that they prefer a 5D Mk II with a small prime lens. But then the GH2 could be used with a pancake lens as well. So we left out the lenses here.


I don't think it is useful to conflate such clear and precise information as weight and size, with the much more nebulous and endlessly disputed qualities of sensor and lens performance.

Aren't we doing this all the time when reading through any camera specs.
More or less intuitively we place bookmarks, at least in a notional sense, then trying to interpret and to interconnect this information:

a.) many Megapixel are basically good
b.) large Sensor Size is basically good
c.) large Pixel Size is basically good
d.) but then high Megapixel Density is typically bad
e.) high DxO Overall Score is considered to be good
f.) large Camera Dimensions are bad (for me)

So the logic suggests that:
high (a x b x c x e / f) is good
Megapixel Density d = Megapixel count a / Sensor Size b,  makes b = a / d
Megapixel Density typically corresponds to the inverse Pixel Size, so the Pixel Size c is redundant.

= high [a x (a / d) x e / f] is good
= high [(a^2 x e) / (d x f)] is good
This is the inverse of the equation in my previous post which was set up to "low is good",
except that the absolute Megapixel count was missing.

Considering that many photo magazines are rating cameras by one single score number which mixes all kinds of attributes,
above approach is probably more at the conservative side.

Regards, Peter

--
Logged

Ray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 10365
Re: Megapixel Count, Megapixel Density x Camera Size
« Reply #8 on: January 29, 2011, 11:13:51 am »


Aren't we doing this all the time when reading through any camera specs.
More or less intuitively we place bookmarks, at least in a notional sense, then trying to interpret and to interconnect this information..............

Regards, Peter



Yes, but in accordance with our own biases. The trouble with single scores derived from a combination of a number of different factors, is that there's always an unavoidable and inherent bias according to the weighting or significance given to each individual factor by the reviewer.

Such single score ratings might be useful for complete beginners, and might also be useful for more experienced photographers if the biases of a particular reviewer are clearly understood.

Speaking for myself, the main factors that influence my choice of equipment are:
(1) Cost, (2) sensor image quality (3) range of lenses and lens quality, (4) general flexibility regarding usefulness of features, (5) weight and size.

The significance that I attach to each of those factors will likely be different to yours or any other photographer's. For example, the cost factor alone would be sufficient to deter me from buying an MFDB system, not to mention its weight and bulk, but obviously this factor does not deter others whose circumstances are different to mine.

What I require in order to make a purchasing decision, whether a new camera body or a new lens, are reliable test results from the application of a consistent methodology which allows meaningful comparisons to be made at the performance level rather than using a weighted score.

DXOMark provide much of this information and are a valuable resource for me, but I'm more interested in the actual test results than over-all scores.


Logged

Peter_DL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 544
Re: Megapixel Count, Megapixel Density x Camera Size
« Reply #9 on: January 29, 2011, 04:32:18 pm »

Yes, but in accordance with our own biases. The trouble with single scores derived from a combination of a number of different factors, is that there's always an unavoidable and inherent bias according to the weighting or significance given to each individual factor by the reviewer.


True.
It can however be intersting to go through such simplifying rating,
to agree or disagree, depending on further insights or individual preferences.

Best regards, Peter

--
« Last Edit: January 31, 2011, 02:07:59 pm by Peter_DL »
Logged

douglasf13

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 547
Re: Megapixel Count, Megapixel Density x Camera Size
« Reply #10 on: January 29, 2011, 05:20:27 pm »

Hi,

Great explanations. Many thanks.
Wouldn't it also mean to accept that for example the sensor / "print" quality of a Sony 580 is the same as with a Canon 1Ds Mk III. Both cameras show a DxO Overall Score of 80. Hmm...


The new Sony 16mp sensor (A580, K5, D7000) is considerably newer than the 1DsIII, and so it's a battle between size and newer tech.  It doesn't surprise me that they test similarly.
Logged

Ray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 10365
Re: Megapixel Count, Megapixel Density x Camera Size
« Reply #11 on: January 30, 2011, 04:53:13 am »

Contrary to common belief, more uis actually better. Reason is that photography is dominated by shot noise and shot noise cares much about sensor size but little about if the pixels are small or large within reasonable limits. What is reduced with small pixels is dynamic range which is affected by readout noise, but even that is less than expected.

Erik,
This is a good point which some photographers seem confused about. Shot noise has no direct relationship with pixel size, only sensor size.

The bigger the sensor the lower the shot noise at any given ISO. However, it needs to be stressed that shot noise comparisons between different size sensors are only meaningful when equal exposures (in terms of ETTR or some other precise standard) are used at equal and real ISO values.

This is another reason why it can be so deceitful (or ignorant) when photographers attempting to demonstrate the superior noise performance of MFDB systems compared with DSLRs, use different real ISOs and different shutter speeds at the same f stop, resulting in different DoFs for each camera, and the creation of a bias in favour of the MFDB.

What is also interesting is that there can be a penalty for the lower shot noise of the larger sensor, and that penalty is either a slower shutter speed or a shallower DoF at equal shutter speeds.

In other words, a FF 35mm sensor will record 1.4x more shot noise than a sensor which is double the size (in terms of area), provided the real exposures used are the same for both sensors.

However, when equalising shutter speed and DoF, the FF DSLR will have a one stop advantage over the sensor which is double the size. This means that the MFDB must be used at one stop smaller aperture and one stop higher ISO to achieve equal shutter speed.

These conditions completely negate and cancel the lower shot noise advantage of the larger sensor.
Logged

Erick Boileau

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 251
    • http://
Re: Megapixel Count, Megapixel Density x Camera Size
« Reply #12 on: January 30, 2011, 07:17:43 am »

and what about f/22 with MF
or f/16 with  a 35mm ?

big or small pixels ?
Logged

Peter_DL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 544
Re: Megapixel Count, Megapixel Density x Camera Size
« Reply #13 on: January 30, 2011, 07:27:26 am »

DxOmark has two different values. One is actual pixels and the other is "print". In print mode the values are scaled for an 8x10 print corresponding to 8 MPixels. The DxO mark is based on the "print" values so it is normalized to 8 MPixel. Simply enough, DxO is essentially image quality. The DxO-mark figure is based on three scenarios.

- Landscape which is essentially DR
- Portrait which is color sensibility
- Low light which is based on usable ISO


The most pronounced differences between all the cameras seem to be in the Low light / High ISO chapter.
P&S cameras fail by nature. Whereas the numerical differences in Landscape and Portrait appear less significant.
Hmm.

Best regards, Peter

--
Logged

Peter_DL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 544
Re: Megapixel Count, Megapixel Density x Camera Size
« Reply #14 on: January 30, 2011, 07:31:55 am »

The new Sony 16mp sensor (A580, K5, D7000) is considerably newer than the 1DsIII, and so it's a battle between size and newer tech.  It doesn't surprise me that they test similarly.

Ok, I'm confused.

Do we have to combine the DxO Mark with the Sensor Size or with the Megapixel count,
in order to derive an absolute score for sensor performance / image quality ?

Peter

--
« Last Edit: January 30, 2011, 07:41:58 am by Peter_DL »
Logged

Peter_DL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 544
Re: Megapixel Count, Megapixel Density x Camera Size
« Reply #15 on: January 30, 2011, 09:23:38 am »

Anyway, here's my preliminary conclusion (see attachment).
Your mileage may vary.

Thanks.
& Cheers! Peter

--
« Last Edit: January 31, 2011, 03:53:22 pm by Peter_DL »
Logged

Ray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 10365
Re: Megapixel Count, Megapixel Density x Camera Size
« Reply #16 on: January 30, 2011, 05:54:42 pm »

and what about f/22 with MF
or f/16 with  a 35mm ?

big or small pixels ?

When diffraction kicks in, it makes little difference if the pixels are big or small as regards resolution, but it can make a difference with regard to DoF. In other words, the sensor with the higher pixel density might be able to deliver the same resolution at the plane of focus at say F22, as another camera with larger pixels, and equal size sensor, used at F16.

For example, I imagine the new 80mp Phase DBs will show very little resolution improvement at F22, compared with the P65+ or even P45+ at F22, but they might be able to deliver the same resolution at F22 as the P45 at F16 (at the plane of focus), in which case there is at least some DoF advantage for the sensor with the higher pixel density.

I doubt that we'll be seeing any P180 comparisons with the 1Ds3 which attempt to show off the superior resolution of the P180 when used at F22, compared with the 1Ds3 at F14, which would be the correct F/stop for equal DoF.

Nevertheless, I would expect the P180 to produce noticeably better resolution at F22 than the 1Ds3 at F14, just not spectacularly better.  ;)

Logged

degrub

  • Guest
Re: Megapixel Count, Megapixel Density x Camera Size
« Reply #17 on: January 30, 2011, 08:21:40 pm »

i thought the D7000 was a ~16Mpixel sensor ?
Logged

Ray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 10365
Re: Megapixel Count, Megapixel Density x Camera Size
« Reply #18 on: January 30, 2011, 10:37:20 pm »

i thought the D7000 was a ~16Mpixel sensor ?

I think Peter has got the megapixels of the D7000 and D300s mixed up.
Logged

Peter_DL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 544
Re: Megapixel Count, Megapixel Density x Camera Size
« Reply #19 on: January 31, 2011, 03:55:36 pm »

I think Peter has got the megapixels of the D7000 and D300s mixed up.

Right. Thanks.
Did some further amendments in the above table
(not sure if it finally makes it better).

Thanks again for discussion.
& Best regards, Peter

--
Logged
Pages: [1] 2   Go Up