I don't reply to private messages.
lumigraphics: Sorry but you guys botched a bunch of this. :sigh:
"Total light" is completely unimportant. Yes, there is less light hitting a smaller sensor, but it doesn't matter because its a smaller total area. Light PER SENSEL is the same.
And lenses don't matter at all. Given a frame-filling grey card (as an example) a 50mm f/4 and 100mm f/4 will give exactly the same 18% grey image on any sensor size. If lenses didn't work that way, you couldn't have external light meters.
Larger digital formats aren't less noisy because they are larger so they collect more light, its because they can have a lower sensel density. In the film days, it was because you didn't have to enlarge so much with a bigger negative.
smdh...and I'm only on page 2.
@GB: "Let's consider a 6 MP sensor and 24 MP sensor where each pixel has the same read noise (say 3 e-)"
But the 'per pixel' read noise is not going to be the same. All things being equal the smaller the pixel the lower the noise power (smaller area of pn-junctions and volume). If you cut that 1 large pixel to the 4 smaller ones, the noise per that area will stay the same (disregarding the defects and losses introduced by such cutting).
@Great Bustard "So, if two systems have the same read noise per pixel, then the system with more pixels will be more noisy if the same total amount of light falls on their respective sensors."
There are many sources that contribute to the read noise (http://www.eng.auburn.edu/~wilambm/pap/2011/K10147_C011.pdf) in semiconductors, but none of them depend on the pixel count (or pixel size). You can broadly divide them into dependent on total area of pn-junctions and total volume. As a first approximation you can consider the read noise as a linear function of the inverse crop factor: O(1/crop).
Here is a hint for you. The sensor doesn't participate in equivalence calculations, other than its area you need to integrate the light flux over. The equivalence will work the same way with any ideal sensor, or any ideal CCD, or any ideal film, or any other ideal recording medium, or even if you remove the sensor altogether and attach a projection optical system that would transmit the image to the viewing screen.
Interesting, every day there are new 'debunkers' here rehashing the same arguments, the difference only in tone, but they all pretty sure they know they are correct. As somebody said, "the trouble ain't what people don't know, it's what they know that ain't so."
There has been more than 1000 comments so far. What can be learned from them?
1) Many people don't read, at least don't engage their minds while reading, but nevertheless they are quick to reply. They don't read the replies to their replies either, or unable to comprehend them, because they keep repeating the same fallacies again and again.2) Many people have difficulty understanding simple laws of physics and elementary school arithmetic. It's really shameful state of affairs. Blame expensive government schools for that.3) the previous 2 problems are exacerbated by a "choice-supportive bias" (http://en.wikipedia.org/wiki/Choice-supportive_bias). Notice that most (if not all) of the objections are coming from the people with small sensors (m4/3, 1").
Now, the really ugly thing is that most of those people are your average voters. I shudder thinking people with such cognitive faculties go to the voting booths.
Babka08: Equivalence of total light is moot because sensor quality, pixel density, etc have much more to do with "clean" images these days. You can certainly generalize that a full-frame camera will have less noise. But my Sony a850 had as much noise at iso1600 as my Sony RX100 (well, not quite, but...).Your explanations of depth of focus are helpful, and the more basic understanding of sensor crop. But other than that, you just confuse and cloud the issues.
Reread the article again. The equivalence doesn't have anything to do with the sensor, other than its area. The equivalence is all about the lens and geometric transformations you need to do to get the same results on different size ideal sensors.
fortwodriver: This is absolute BS... The f-stop is a measure of the ratio of the diaphragm opening to the focal length of the lens. Nothing more, nothing less. If you want to talk about light transmission, please do your research on T-Stops and stop making this stuff up.
You think you revealed something about the article? No, you only revealed something about yourself.
DoctorJerry: Aperture equivalenceI ran a test using my Panasonic LF1 and selected sensor sizes of 12MP, 8MP, 5MP 3MP and 0.3MP. I can select those sensor sizes by using Panasonic’s EZ Zoom which trims off pixels on the perimeter of the sensor to arrive at a smaller sensor. As I read your article, I should have been needing either a higher ISO, slower shutter speed, or faster aperture as I trimmed pixels from the sensor. I found NO difference my test shots, all shot at 1/80sec, F2.0 and ISO 200, they were the same REGARDLESS of the size of the sensor I used.
Where I think you went wrong is in talking about how much less light reaches the sensor as the sensor gets smaller. What you overlooked was that the same quantity of light reached the plane of the sensor but since the sensor was smaller, it captured less light. It did NOT need all the light reaching the plane of the sensor, only enough to cover the sensor itself. According to my test f2.0 is 2.0 regardless of the sensor size.
Why don't you reread the article, because you're clearly demonstrating you didn't understand a thing.
jonas ar: From the Sensitivity equivalence paragraph:" By comparison, in good light, the generally larger pixels of a large sensor will tend to give cleaner images, without any dependence on aperture."
Why was pixel size brought into the discussion?
You are correct, as the first approximation the pixel size doesn't really matter (as long as there is no additional losses on interconnect, crosstalk and other second order effects). If you cut a large pixel to four smaller ones, nothing changes in terms of collected light, noise, or dynamic range.
Polytropia: This is a crock. There is no equivalence in apertures. F/2.8 is always F/2.8 no matter what the sensor size is. It projects the same brightness of light in all cases.
Point is: a speedlight (flash) that has a guide number of 100 feet will illuminate a subject 35.71 feet away at ISO 100 and F/2.8. No matter what your sensor size is, this will not change.
You cannot say that just because the sensor is bigger that changes anything because if you make the lens F/5.6 and do not change the flash guide number or ISO, then your exposure will be off.
Further, the amount of image noise generated is not exactly two stops "better" between, say, Four Thirds and 135-format. Neither is dynamic range. It varies based on the year the camera came out, how many pixels it has, etc.
DOF is also not exactly two stops different because DOF is affected by pixel density as well.
So stop LYING to people, DPReview.
@Polytopia -- it's one thing to not understand the topic, but it's a completely different matter to arrogantly attack those who actually do understand. People are willing to explain to the ignorant, but they have no respect or patience for an arrogant fool.
Here is a suggestion for another article: pixel vignetting. There is also a lot of confusion as to how the effect demonstrates itself on different sized sensors, and whether it effects the total light and DOF in different measure.
quezra: One thing this article doesn't mention is that as sensor tech improves, your lenses are basically getting faster. Today's gutter standard f2.8 lenses can outshine the f/0.95 lenses of yesteryear because you couldn't push film beyond ASA 800 much (aside from being very expensive). When people complain some new 50/1.8 lens isn't "fast" it is actually a whole lot faster than a 50/1.4 lens from just 5-10 years ago because of digital improvements on the sensor. A "slow" FF kit lens (f/3.5-5.6 usually) producing indoor nighttime ambient light shots would simply be unthinkable 5 years ago.
thinkfat is correct, it's a bad idea to mix a sensor performance and a lens performance. Each contributes its own: the lens collects photons, and the sensor converts them to the electrons. The equivalence is all about the geometrical transformation of the lens parameters for the different sized targets (you may consider the sensor to be ideal). What would really be bad if we started talking about the same lens and different generation sensors and recalculating lens performance to some common sensor performance standard. There would be no end to schizophrenic equivalent for DOF but not for light, etc. arguments.
gianstam: About total light.So, I have a question to DPReview?Do the Sony and Nikon FF cameras produce different noise at pixel level when used at different formats for the same sceen?
My point of view: the sensor does not behave as a whole. The electric signal from each photosite is not affected by the the total light. Sensor is not a sun light collector (larger surface, more electric power).Image (and noise and DR) is produced by pixels. Same tecnology and size pixels produce the same results. So the advandage of the larger sensor appears not because of the more light but because of the more (same sized) pixels or because of the larger pixels (for the same Megapixel sensors).
Very informative article by the way.
How can somebody write "Very informative article by the way", and in the same post demonstrate he didn't understand anything?
Fygaren: If only they could make a multi-format equiva-lens...Ok, way past my bedtime!
Well, focal reducers do exactly that (sans imperfections).
Bjorn_L: Interesting write up. Nicely done. This is a topic which gets flogged to death in the forums (typically with more passion than facts); it will be nice to have a well written article to refer to.
I'm afraid it changes nothing. There were nice articles before that, it never stopped flat-earthers from denying the equivalence. If ignorance is a bliss, why would they want to be confused with the facts?
Wilight: I'm really impressed with the patience of dpreview staff to awnswer dumb questions/comments, even after providing an article like this. I can't realize why such a simple thing still produces so much babbling these days. For me, it's like trying to explain why 2+2 equals to 2x2. Are most of dpreview readers kids or what?
No they are not kids, for kids understand these things easily. They are an example of an average voter: stupid and ignorant.
joe6pack: I like to see some mentioning of focal reducers. A focal reducer focuses more light on a smaller sensor. Which means more light and wider FoV. But, AFAIK, the DoF is unchanged.
The DOF and light flux over the sensor surface always behave in tandem. You can't get more of one without proportionally less of the other.
blohum: Great article DPR and it's that last sentence that people always seem to forget: "There's no universally ideal 'sweet-spot' to that trade-off: but understanding equivalence can help you work out which balance is best for you."
I chose m43 knowing full-well what the limitations are. Will FF give better IQ... absolutely, will FF give thinner DOF, of course... do I care, not a bit... m43 is MY 'sweet-spot'.
There isn't really a well defined sweet-spot, not in an absolute sense anyway. As Sony keeps demonstrating it, the camera size/weight isn't really a function of the sensor size, it's ergonomics is mostly determined by an average palm size. And as the physical principles explained in the article demonstrate that most equivalent lenses are also going to be the same in size/weight no matter what the sensor size is. When Sony finally releases FF camera in a NEX body with a tiny 24-70/5.6 lens (equivalent to 12-35/2.8 m43 lens) where that sweet-spot is going to be?
lacikuss: Excellent article...
So why then a Sony RX-100 III is priced at $850? It should cost $450 according to this article...
"There is no "depth of field advantage", as it states in the article." -- lightpainterx is a writer, not a reader :)
Because it's been repeated 24-70mm f/1.8-2.8 so many times, an average clueless consumer thinks they are getting a better deal than Canon 5D3 with 24-70/2.8.The blatant manufacturers lies (i.e. advertisement) starts backfiring. They were happy advertising cameras with crop sensors until Sony started releasing FF mirrorless. Sony needs to get across the truth to the consumers now or they won't understand why they should pay more and get bigger equipment.