JanKritzinger
Forum Enthusiast
Absolute, unadulterated BS.
Noise increases and DR decreases as pixel pitch is reduced.
-
Welcome to the new forums! Please read this first. For known issues we are working to resolve, click here.
Small sensors gather less light? Wrong!
- Thread starter Chippy99
- Start date
Noise increases and DR decreases as pixel pitch is reduced.
zemes gaoson
Well-known member
You're right as far as the optics is concerned, but you made a mistake on a critical matter which is very commonly misunderstood. This relates the dead zones between the photo cells. Due to manufacturing reasons, the percentage area occupied by the dead zones is not a constant. It increases when the individual pixel size decreases. To understand this, just consider an extreme scenario where the sensor class just one cell, in which case, there would be no dead zone. When the sensor has multiple cells, there will be areas between the cells which are not active for receiving photons. The current technology of course always tries to minimize the dead zones, but there's a limit. For this reason, for a given number of pixels, the total absolute size of the dead zones is roughly about the same for a given manufacturing technology, but the relative sizes is not the same, and as a result, the percentage area occupied by the dead zones relative to the total sensor size is greater for a smaller sensor.
Second, although the point you made is theoretically correct, in practice however smaller sensors are used in systems that have smaller lenses, and therefore receive less amount of light. That's the reality. In the case of 35mm full frame and APS-C, although the physical size of lenses used are the same (except for the lenses that are designed for APS only), APS sensor does not use all the light that comes through the lens, so essentially that corresponds to a smaller lens as well.
Second, although the point you made is theoretically correct, in practice however smaller sensors are used in systems that have smaller lenses, and therefore receive less amount of light. That's the reality. In the case of 35mm full frame and APS-C, although the physical size of lenses used are the same (except for the lenses that are designed for APS only), APS sensor does not use all the light that comes through the lens, so essentially that corresponds to a smaller lens as well.
Filip Wiltgren
Member
There would still be a drawback to the small sensor - you need to assume that lens quality for the lens optimized for the small sensor is sufficiently larger than for the one optimized for the large sensor that you'll be able to utilize the information.
In a "smaller sensor" lens the errors in the lens glass would become magnified and the drop off in light intensity from those errors would become larger. While this isn't as crucial as the crop factor it would play a part.
Also, those same errors would cause a degradation in image quality (you'd get a grainier low light image) due to the randomness of the light passing through the lens being "concentrated".
JanKritzinger
Forum Enthusiast
No, when you talk about pixel pitch, and especially with regards to smaller sensors, you compare per pixel. The very reason you use finer pixel pitch is because you want to get away with using a smaller area. Why would you compare per area if the very heart of the argument is that sensors with less area are worse?
"ceteris paribus" dictates same number of pixels, and not same area. The whole discussion revolves around the diffence in area. A small sensor is a sensor with less area. That is why its pixel pitch is decreased. That is why it yields more noise and less Dynamic range.
TrojMacReady
Veteran Member
Thanks, was wondering when the old myth would return and my suspicion was right. We have a believer.
TrojMacReady
Veteran Member
Like I said, ceteris paribus smaller pixel pitch isn't an issue. It's unrelated to the heart of the argument because per area noise levels aren't higher (often rather the opposite).
A larger sensor with the same number of pixels will not automatically perform better as proven over and over by measurement. Area is the heart of the discussion. Pixel pitch isn't a problem.
No, both are mainly related to area, not pixel pitch.
JanKritzinger
Forum Enthusiast
If it's a myth I cordially invite you to debunk it.
Chippy99
Senior Member
Understand all that.
This thread wasn't about the merits of large vs small sensors and all that goes with it.
It was merely a reflection (pardon the pun) that a small sensor can in theory capture exactly the same number of photons as a larger one, given the same lighting conditions. And therefore the large sensor = lower noise view that is almost universally accepted as being a fact is not necessarily always correct.
That is all.
TrojMacReady
Veteran Member
Let's stay close at home. The Sony A550 and A500 both have a similarly sized sensor, the main difference is the fact that the A550 has more pixels. Yet the A550 does not show worse noise levels or less dynamic range. Area is the keyword, not pixels.
JanKritzinger
Forum Enthusiast
ceteris paribus, a larger sensor with the same number of pixels will perform better as proven over and over by measurement.
JanKritzinger
Forum Enthusiast
You're trying to use a difference of 14% in pixel pitch to prove your point?
Are you being serious?
Are you being serious?
Chippy99
Senior Member
All sounds very reasonable.
Apart from the fact that your argument overlooks the inclusion of micolenses over the photocells, whose purpose is to ensure that all the light hitting the sensor falls into a photocell and not onto inter-cell wiring. I am not sure how efficient this is, but I would imagine if you look at the sensor from the front, you see nealy all cells and hardly any wiring!
Are you able to state with confidence that all p&s cameras with different sizes of sensor - 1/1.7" and 1/2.3" have the same lens size? If not, then the small = bad, large = good argument is meaningless.
True enough.
A lens with an aperture of f:2.8 or whatever will "deliver" the same amount of light across its image circle (disregarding vignetting). If we consider the 24x36 mm area as base area, then the APS-C format receives less than half the total amount of light, and the 4/3 format receives one quarter the amount of light. If all three sensor formats have identical resolution, then the individual pixels receive the same fractions of light.
If you use an "inverted teleconverter"/"wideangle converter"/tele compressor, then you "squeeze" the total amount of light to fit in a smaller image circle, which will increase the amount of light per area unit (like a magnifying glass used to start a fire), but that's hardly an option with lenses made for small sensors, because they're already designed with the smaller image circle, and nobody makes them for full frame lenses, because Kodak hold the necessary patents.
TrojMacReady
Veteran Member
Yes. Measurements should show the differences that follow your theory. They don't. The A550 sensor actually performs slightly better at high ISO and shows better DR figures. The A300 and A350 perform similarly aswell in both regards.
If you want a more mathematical or scientific approach I suggest reading posts like those by John Sheehy, Ejmartin, Eric Fossum (inventor of CMOS) in these forums.
Example of a comparison:
http://forums.dpreview.com/forums/read.asp?forum=1018&message=28607494&q=sheehy+fz50+canon&qf=m
And here's a thread all 3 posters mentioned above make their point:
http://forums.dpreview.com/forums/read.asp?forum=1000&message=30002827
DXO has a good explanation on their website (click the different tabs at the top for the rest of the article):
http://www.dxomark.com/index.php/eng/Insights/More-pixels-offsets-noise !
(add the exclamation mark manually to make the above link work)
TrojMacReady
Veteran Member
I would love to see that happen. There is none, as pointed out here (including sources):
http://forums.dpreview.com/forums/read.asp?forum=1037&message=35530803
Zamac
Senior Member
I think there is a certain amount of talk at cross-purposes going on.
The amount of light falling on a sensor depends on the effective diameter of the lens (not aperture number), the loss when passing through the glass elements and filters and the proportion of the light cone occupied by the sensor.
The efficiency of photon capture by the sensor is another issue. I think the OP was more concerned about a geometric missconception when talking about the size of sensors.
The amount of light falling on a sensor depends on the effective diameter of the lens (not aperture number), the loss when passing through the glass elements and filters and the proportion of the light cone occupied by the sensor.
The efficiency of photon capture by the sensor is another issue. I think the OP was more concerned about a geometric missconception when talking about the size of sensors.
JanKritzinger
Forum Enthusiast
Firstly, 14% difference in pixel pitch is practically negligible.
Secondly, you're comparing sensors of different resolution.
Thirdly, you're comparing sensors of the same size.
We are discussing smaller sensors with finer pixel pitch.
So if you want to convince me, compare a 12 Mpix sensor with 300% (which is still moderate) greater pixel pitch than another 12Mpix sensor.
Secondly, you're comparing sensors of different resolution.
Thirdly, you're comparing sensors of the same size.
We are discussing smaller sensors with finer pixel pitch.
So if you want to convince me, compare a 12 Mpix sensor with 300% (which is still moderate) greater pixel pitch than another 12Mpix sensor.
Chippy99 wrote:
What flaw ?
But you can plainly see the max f-stop ranges on any digicams, and the best of them (a couple of $400-500 models) are in the f/2 range. That means they project the same intensity of light onto a given square mm of the sensor as an f/2 lens on a large sensor camera. Or less total light.
1/1.7" enjoys a one stop advantage over 1/2.3". If the 1/2.3" camera has a 1-stop faster lens, then they're on an even footing. It seems to be the case, though, that the cameras with the larger sensors offer faster lenses.
Gallery at http://kingofthebeasts.smugmug.com
What flaw ?
Sure. And if someone ever invented a small sensor camera with a lens that did that, then it might be great for low light.
But you can plainly see the max f-stop ranges on any digicams, and the best of them (a couple of $400-500 models) are in the f/2 range. That means they project the same intensity of light onto a given square mm of the sensor as an f/2 lens on a large sensor camera. Or less total light.
If that were true, you'd see f/0.5 lenses on them.
The so-called "myth" is factual and measurable - you can calculate it from sensor size and aperture and see it in the high ISO results. Or, given that you seem to acknowledge that you'd need huge lenses, the myth persists because it applies in practice even if could be violated in theory.
This is one of the big arguments for expressing apertures in "equivalents". If you use the crop factor to come up with an equivalent f-stop, you see a relative measure of how much light reaches the sensor.
1/1.7" enjoys a one stop advantage over 1/2.3". If the 1/2.3" camera has a 1-stop faster lens, then they're on an even footing. It seems to be the case, though, that the cameras with the larger sensors offer faster lenses.
Not bizarre at all. In fact, it can be interesting to compare something like a Panasonic LX3 to a micro 4/3 camera with kit lens, as the m43 camera may still enjoy some advantage, but not as much as the high ISO tests suggest. And if you're shooting anything handheld in low light where you want DOF, it's pretty much a wash because you have to stop the bigger sensor camera down as many stops to match the small sensors DOF as the number of stops in light gathering ability you gain.
- Dennis
Gallery at http://kingofthebeasts.smugmug.com
JanKritzinger
Forum Enthusiast
If what you're saying is true, a 12mpixel compact would have the same sensor performance as a 12Mpixel Full frame camera, which is simply absurd.
Comparing two aps-c cameras is nothing short of silly.
Comparing two aps-c cameras is nothing short of silly.
JanKritzinger
Forum Enthusiast
PS: we are not talking about the (number of pixels: pixel pitch) ratio for a given sensor size.
we are talking about different sized sensors at roughly the same resolutions.
we are talking about different sized sensors at roughly the same resolutions.
Keyboard shortcuts
- f
- Forum