Pages: [1] 2 3 4   Go Down

Author Topic: f stop/sensor size realtionship  (Read 16209 times)

Bobtrips

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 679
f stop/sensor size realtionship
« on: April 29, 2005, 01:29:20 pm »

Thanks, both of you....

I don't understand this last bit.

"smaller pixels will produce less dark current noise, some sources of which follow a "per unit area" empirical pattern"

It's actually the last bit of the last bit that I don't understand.  :)

The first part of the 'bit' intrigues me.  If there's less dark current noise doesn't that bode well for small site sensors if they can be read multiple times and blended?  (If photons can be cumulated over frames and divided by a lower concordant noise?)

---

Then.  In the focal length/aperture/sensor size relationship, doesn't the front element size also contribute?  Larger area = more light collection.
Logged

Ray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 10365
f stop/sensor size realtionship
« Reply #1 on: April 30, 2005, 12:29:25 pm »

Quote
a) once controllable noise sources ("total dark noise") are reduced enough and photon shot noise is the limit, there is no advantage to image merging; the S/N ratio at any given pixel is determined simply by the number of photons used to make that pixel, due to that square root law.
BJL,
Perhaps no advantage in image merging to reduce photon noise but wouldn't there be some advantage to improve over all S/N? With such small photosites, most of them would have no signal at all with just a single exposure for the highlights.
Logged

BJL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6600
f stop/sensor size realtionship
« Reply #2 on: May 03, 2005, 03:06:38 pm »

Ray, now I am confused. Though I believe that claims of DOF/speed advantages for either smaller or larger formats are greatly exagerated, I thought that we were agreed that the lower total dark noise levels (in electrons) is, for the time being, a potential S/N advantage to a smaller sensor.

You realy are passionate about denying any possible advantage to smaller formats, aren't you? One seriously wonders why medium format lost so much ground amongst professionals to the smaller 35mm format if the latter had so little in its favor!

The naive phyics of the possible advantage for a smaller format is that with the same aperture diameter and exposure time, a smaller format sensor gathers the same total amount of light from the same subject (same total signal in photons or electrons) but has less total dark noise (less electrons of noise, due to less total bulk of silicon), which points to a potentially higher overall S/N ratio for a smaller sensor.


On the second point, I thought it was decided in a previous discussion that at a given angular FOV, pixel count, and camera weight, the same shutter speed is needed to freeze camera motion adequately. Since FOV counts, not focal length itself, it should be not 1/f, but 1/(35mm equivalent f).

Before you say that smaller, lighter cameras are thus at a disadvantage: nothing stops one from making a camera of any format as heavy as desired, by adding a battery grip or pure "balast" if needed.
Logged

Ray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 10365
f stop/sensor size realtionship
« Reply #3 on: May 04, 2005, 11:10:39 pm »

Quote
I wonder if some people are so out of touch with reality that they still have hopes for a low priced 6MP 35mm format sensor? (Not you Ray; I am sure you have dropped that idea by now!)
Dropped it? I never had it to drop  :D .

I came acroos the LL forum around the time Michael produced his controversial review of the 3mp D30. I recall all the whingeing on the forum then from people who wanted the greater wide-angle coverage that full frame afforded.

The expectation was that Canon's successor to the D30 would be a 6mp full frame. It was almost guaranteed. Michael made several comments that this was very likely.

I wrote quite a few posts in those days extolling the virtues of the smaller format, but like many I didn't really believe it would be possible to maintain the low noise of the D30 with the smaller pixels that would be required to cram double the number into the same space and I thought Canon would probably cave in to the chorus of demand for full frame.

I was surprised but very pleased that the D60 maintained the same format as the D30, so I bought one.

I've since been rather puzzled by frequent statements from 'experts' that pixel size had already reached the resolution limits of 35mm lenses and that increased pixel density will serve no useful purpose. I've engaged in arguments with people who have tried to put forth the view that the D60 pixel size is simply too small and detrimental to image quality compared with the D30 and that the Canon engineers know this and are pulling a fast one.

My role in this current debate about large sensors versus small is more of a 'devil's advocate' role against the ultimate triumph of the small sensor.

I'm really just trying to get a grasp of the fundamental limitations. Can photonic shot noise, the major noise component in small sensors and perhaps the ultimate barrier to further improvement of image quality, be tackled. I don't know. Do you?

Can dark noise, the major noise component in larger sensors and perhaps the ultimate barrier to image quality, be tackled? Well, there's evidence that it can.

Can the 'small well' photosite of the small sensor capture the range of levels of the larger photosite and therefore deliver the same DR whatever the S/N? Apparently not.

Can technology get around this DR limitation of the small photosite by constructing really deep wells and/or merging several shots in-camera? Maybe.

If the larger format sensors employ similar technology to produce extra deep wells and automatic merging of different exposures in-camera, could the smaller format ever compete, eh!?

Perhaps the ultimate triumphal factor will not directly be S/N or DR but P/P (price-performance ratio)  :D .
Logged

Ray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 10365
f stop/sensor size realtionship
« Reply #4 on: May 06, 2005, 01:16:59 am »

Technological development has an effect of 'trickling down'. Of course more people want little things than big things (photo-wise). But humanity as a whole marches on to ever bigger and better.
Logged

Bobtrips

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 679
f stop/sensor size realtionship
« Reply #5 on: May 06, 2005, 11:32:55 am »

Quote
there will never be a good quality, general purpose, interchangable lens camera system in a format smaller than the current smallest DSLR format, 4/3. (Ignoring the debate about whether there is or will ever be such as system in 4/3 format!) The lower size limit set by the "law of diminishing returns" has been reached if not exceeded by current DSLR formats: camera and lens costs and sizes can not be reduced much further, and so on.
You lost me there....

Why won't we see a "FZ20" with a wide angle option?  Why not a very wide ceramic short zoom that can be traded out for the monster zoom?

(Perhaps with Samsung is pushing into 15x territory a second lens won't be necessary to shoot everything.  Down the road a 20x 18 - 360.)

Is there a problem building a very wide lens for a small sensor camera?  Or just a lack of market demand?
Logged

BJL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6600
f stop/sensor size realtionship
« Reply #6 on: May 06, 2005, 03:49:02 pm »

Quote
humanity as a whole marches on to ever bigger and better.
Ray,

   apparently you have not noticed the trends in mobile phones, portable audio equipment, the cameras that most people have been buying over the last century or more, or almost any other hand-held product. My obervation:

"If it carries them, people want bigger [e.g. SUVs],
if they carry it, people want smaller."
Logged

Bobtrips

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 679
f stop/sensor size realtionship
« Reply #7 on: May 07, 2005, 12:12:01 pm »

Quote
The only advantages I can see are at big apertures where there's less DoF.
Design in a 'shallow DOF' mode in a small sensor digital.

Immediately after taking the desired shot move the focus point a fixed percent.  (This could happen very quickly as the camera 'knows where it's going'.)  Take a second shot.

Software (initially in computer, later in camera) could calculate relative distance of objects in the frame.  

Then you could choose your favorite brokeh.  (Probably argues for a post-processing approach.  More flexibility.)

(And someone suggested phase shift, but my knowledge of physics is too limited to deal with this solution.)
Logged

Bobtrips

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 679
f stop/sensor size realtionship
« Reply #8 on: May 08, 2005, 03:29:50 am »

Well Ray, I get the feeling that you are 'young boomer' or less.  Take a look at us old dogs - Didger who's about boomer + 4 and me, boomer + 2.  We're the point men for this huge population bulge that's working its way toward old.  Both of us are looking for lighter cameras for our travels.  

Maybe you young pups are still willing to set off with several pounds of gear, but lots of others aren't.  Neither do we go shooting with car and driver.  

Didger has just spent several thousand dollars to lighten his load a few pounds.  I'm dreaming of more capable one pound or less cameras that will capture what I'm after.  (In the meantime I'm passing on usable high ISOs and shallow DOF.  It's a price that I am willing to pay.)

Because that software hasn't been written yet doesn't mean that it won't.  And because high DR, small sensor cameras aren't on the shelf at this moment doesn't mean that they won't be.

Market pressure has been for smaller, more portable cameras for a long time.  Olympus made its mark by releasing the first compact 35 mm SLRs.  (I dumped my Pentax for an OM1 soon after they appeared.)

35 mm film won out over MF film, not because the film and developing were so much cheaper.  But because the cameras were smaller.

IMO digital cameras will continue to shrink because the market wants small.  

Now I don't think that the shallow DOF solution is 'blending'.  I suspect that it could be done by 'object identification' and determining the relative distance position of those identified objects by how much the edges become sharper or more diffuse when the focus point is slightly moved.

Will that require some 'chewing power' on the part of the CPU?  Quite possibly so.  But Moore's Law seems to be holding for speed.  Stuff is a-changing.
Logged

BJL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6600
f stop/sensor size realtionship
« Reply #9 on: May 10, 2005, 04:40:38 pm »

Reply,

I seriously doubt that either you or I is in a  position to assess the potential of such technologies; if researchers at Standford University say that it has potential, and the technology has already been put into commerical use, I see little reason to worry just because you are cynical!

but I have no idea where you get your 40MHz from. 4x faster than 1/100th of a second is 400Hz, not 40MHz, so you are off by far more than my one zero!.  I repeat that the testing of photosites only needs to be done at a veryeasily attainable frequency of about 10,000 times per second or less.

By the way, the version that I have heard of does a local analog to digital conversion on the chip at each site, applies an adjustment for the different exposure time at different sites to that digital value while still "on site" (probably just a bit shift for each factor of two difference in exposure time), so that the final data read off the sensor after the exposure is finished is already digital, and so immune to any further noise sources.


As to your absurd taunt about 4/3 format, I should not stoop to this nonsensical DPReview forums style rhetoric of "my brand currently sells better than your brand, therefore it and/or its format size are inherently and permanently superior".

Alright, I will stoop a bit, just for fun.

a) The size difference between 4/3 and say Canon's 1.6x is so small that the broad factors I am discussing have so little effect; they are probably drowned by other variations between product lines: it is about 10% to 20% linear, 20% to 40% in area, or less than a half stop.

The jumps in format size that are big enough to have a clear cut effect are ones between levels like
- digicam, up to 2/3"
- the new DSLR formats as a whole, everything from 4/3 to DX
- 35mm
- 645

 A huge array of other factors are in play when comparing Olympus to other DSLR makers, most obviously that back in the 35m film SLR era, Olympus had become a far smaller player than Canon and Nikon, or even Pentax and Minolta, with no established auto-focus lens system and a tiny SLR user base.

c) A far more interesting comparison is to Pentax and Konica-Minolta, who were closer to the size of Olympus in the 35mm film SLR market than were Canon and Nikon, though still substantially bigger. The Olympus E system is reportedly easily outselling both Pentax and Konica-Minolta DSLRs combined, and also outselling both Fuji and Kodak DSLRs, and possibly about matching or surpassing all those brands combined for unit sales volume.

So in fact, the E system has overtaken a number of competitors including several bigger companies who were in the DSLR market earlier.

d) As far as I can tell, Olympus E system models also easily outsell Canon's 35mm format models, in both unit volume and revenues. (Perhaps you will complain that this is due to huge price differences; but once you raise current sales volume as an argument, price is a legitimate factor in comparing the viability and prospects of different formats.)
The figures I have read in official statements from Olympus and Canon are shipments of about 40,000 units per month for the E system in the early months of 2005, versus a reported production level of 2,000 units per month, first for the original 1Ds and then for the 1Ds Mark II.

From other sales figures, 4/3 probably outsells all digital offerings in all formats 35mm and larger.

e) If instead you compare Canon's own sales in different formats, or 35mm DSLR as a whole to smaller format DSLR sales as a whole, then by your flawed reasoning, the glaring conclusion would be that it is 35mm format has totally failed to "overtake anyone" in the DSLR market place, despite the backing of the industry leader Canon. Instead 35mm DSLRs are stuck with a very small and shrinking share of total DSLR sales and revenues. As far as I can tell, 35mm format digital even has a smaller total sales volume than medium format had in the era of film.
Logged

Ray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 10365
f stop/sensor size realtionship
« Reply #10 on: May 12, 2005, 08:53:41 pm »

Quote
  do you really think that I believe that smaller formats are overall superior to larger ones?
I was certainly beginning to get that impression, BJL   , but thanks for clarifying your stance on this.

If the sampling technology you've mentioned really is practicable and would not require unrealistic amounts of computing power and/or significant delays during the exposure whilst photosites discharge and reset; if it really is technically easier and more economical to manufacture a truly diffraction limited f2, small image circle lens, than it is a diffraction limited f8 35mm format lens; then I think there could be justification for claiming the smaller format superior on balance. But not of course in every respect. I don't see how technology could ever hope to 'mimic' the combination of speed and image quality of an 85/1.2 35mm format lens, using a 2/3rds format, unless it becomes possible to make an F0.3 lens at some time in the future. Creating a shallow DoF could be done in software, so the bottom line is really one of speed in favour of the larger format and ultimately resolution if at some point, for example, it becomes possible to make a diffraction limited F4 lens for say, 35mm but impossible to make the equivalent diffraction limited F1 lens for a 2/3rds format.

Actually, I'd be very interested in an affordable 2/3rds format camera with detachable 400mm f2 diffraction limited lens and image stabilisation. Wow! That would be one superior camera.
Logged

Ray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 10365
f stop/sensor size realtionship
« Reply #11 on: May 14, 2005, 11:16:23 am »

This idea of 'the best camera for the job' is interesting. It really highlights what an expensive hobby, or even free-lance profession, photography can be. A camera for each situation. This is getting very specialised and professional.

Those who are well-heeled can of course afford that variety of equipment. I cannot. I'm a generalist rather than a specialist and I want a camera system to be as versatile as possible, which is why I (and clearly many others) have opted for the interchangeable lens system of a medium format camera... er! not medium format as in the old MF 6x7cm etc, but middle of the range from 4mmx5mm to 36x48mm.

We all tend to make compromises between absolute image quality, convenience of operation and portability, depending on our priorities.

I haven't yet got to the stage of preferring a 2/3rds format camera on long walks because it's lighter than my 20D plus a couple of lenses. But I might eventually  :D .
Logged

BJL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6600
f stop/sensor size realtionship
« Reply #12 on: May 17, 2005, 10:14:42 am »

I am amazed to see both of you interested in an idea I have had for a while: a fully professional level Canon DSLR in EF-S format, for situations where "agility" and the longer telephoto reach of any given lens through using smaller pixels is an advantage. The long awaited EOS-3D?

Can anyone else see this being an interesting option for some sports and wildlife, as a complement to the 24x36mm format 1Ds series?


P. S. Ray, it sounds like your MF gear is meeting the same fate as my film scanner!
Logged

Ray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 10365
f stop/sensor size realtionship
« Reply #13 on: May 18, 2005, 10:57:28 am »

Quote
how good can the MTF be at about 50lp/mm and distances of up to 14mm from the center of the frame be for (a) current good Canon 35mm format lenses, and ( future professional level EF-S lenses, if and when Canon makes any such.

My guess: if Canon chooses to do it, they will be able to make lenses that can handle the job.
Well, I've never come across MTF charts at 50 lp/mm, but for what it's worth, the best performance of the 70-200/f4 at 135mm and f8 is approx. 67% (averaging tangential and radial), and the best performance of the 100-400 at 180mm and f8 is 57%. (Photodo results)

Do we drop down about 20% for 50 lp/mm or more?
Logged

Ray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 10365
f stop/sensor size realtionship
« Reply #14 on: May 08, 2005, 10:57:43 am »

Well, Bob, I think you're straying off the point here (or the point that this thread has evolved to). I've already had a discussion with Didger that lighter is better for backpacking purposes, so forgive me if I don't want to get into that argument again  :D .

I merely make the point that during the past 5 or 6 years, despite the apparent take-over, plethora and ubiquity of the small and very small digicam, the rate of increase in absolute image quality of these small devices seems no greater than that of the larger formats. All formats are getting better, year by year.

The star performer so far this year has undoubtedly been the Nikon D2X, but I don't necessarily think the success of this camera will knock FF 35mm out of the market.
Logged

Bobtrips

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 679
f stop/sensor size realtionship
« Reply #15 on: April 28, 2005, 10:35:39 am »

I'd like to ask this question in a calmer, more helpful forum than where the issue originated.  :)

Here's a C&P from a post...
--------
Size of FZ20 sensor = 5.76*4.29 = 24.7104mm2

Size of 350D sensor = 337.5mm2

Ratio: 13.7:1

f-stop multiplier = sqrt(13.7) = 3.70

Hence a lens for the 350D only needs to be 2.8*3.7 = f10.36 to be effectively as bright as the FZ20.

--------

First, true?  (And a link to a simple explanation would be appreciated.)

Second, if true then how does one compare the capability of one type camera to another?
Logged

BJL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6600
f stop/sensor size realtionship
« Reply #16 on: April 29, 2005, 03:25:41 pm »

Quote
I don't understand this last bit.

"smaller pixels will produce less dark current noise, some sources of which follow a "per unit area" empirical pattern"
...

In the focal length/aperture/sensor size relationship, doesn't the front element size also contribute?  Larger area = more light collection.
First a warning, the total "dark noise" in each pixel comes from a mixture of at least three types of dark current in the sensor photosites plus noise picked up later in the process, as the charge is moved out of the photosites, converted from a charge to a voltage, amplified, and fed through an analog to digital converter. There is a detailed discussion of this in the Kodak technical report http://www.kodak.com/global....ces.pdf
The only thing I am fairly sure of is that increasing photosite size will tend to somewhat increase the total dark noise, counted in electrons; how fast it increases is probably slower than proportional to area.

That complicates your interesting question about blending multiple frames from a smaller sensor, which is somewhat related to several technologies currently being used in surveillance video cameras to get great dynamic range out of small photosites. I think you can rest assured that lots of ideas like yours are being worked on; manufacturers love the cost advantages of squeezing ever better performance out of ever smaller sensors. (For example, have you heard about the latest sensor from Sharp, with more pixels than the previous 8MP, 2/3" format champion and in the smaller 1/1.8" format, meaning 2.2 micron pixels? See http://www.dpreview.com/news/0410/04102601sharp8mp.asp)


Suppose a smaller sensor and pixels are 1/4th the size of a larger one. When you push exposures time to as long as they can go within the limits of well capacity signal (electron count), the total signal that the larger sensor can get in a single frame would require blending four frames from the smaller sensor. Due to the "root mean square" way that random noise sources like dark noise combine, this would give twice as much noise as a single frame, not four times as much.  So if all the dark noise varies in proportion to pixel area, it is 1/4 as much in each frame for the smaller sensor and so still only half as much in the blended image.

But as I said above, the overall scaling between photosite size and total dark noise is not clear, so the "winner" here is not clear either.


Front element size is indeed the key factor in light collection rate, and you are roughly keeping it constant when you scale the aperture ratio with the square root of sensor area, since the focal length also scales with the square root of sensor area (linear sensor size) to maintain the same FOV.
Logged

BJL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6600
f stop/sensor size realtionship
« Reply #17 on: May 02, 2005, 05:35:22 pm »

Ray, completely new text, I misunderstood your post! Then again, maybe you misunderstood my point too, or maybe it is a matter of language: what you call improving overall S/N ratio, I think of as improved dynamic range.

Yes, so long as sensor designs limit the maximum exposure (minimum Exposure Index or minimum ISO speed) due to limited highlight headroom, blending can improve shadow noise levels by allowing the use of a longer exposure for the shadows, blended with a shorter one for the highlights. Still true even if photon shot noise were the only noise source.

I start to imagine cameras that automate sequencing of frame for both exposure level blending and the focus distance blending described in another recent thread. For fans of ready to use JPEGs cameras could even do such blending themselves if processing power increases enough. (Outputting in JPEG2000, which seems to allow bit depth greater than 8.)
Logged

Ray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 10365
f stop/sensor size realtionship
« Reply #18 on: May 03, 2005, 12:31:15 pm »

Quote
Quote
I do not see why you keep working so hard to argue that the potential noise advantage of smaller photosites wil become an illusion at some point, because no-one seems to be disagreeing lately.

Is that what I've been doing  :D . I thought I was arguing that the fast lens/greatDoF/fast shutter speed combination of the smaller format is an illusory advantage.

But now that you've mentioned it, there's another factor related to this matter which perhaps you can shed some light on. How far does the 1/FL rule of thumb extend with the shorter lenses?

There were many situations in Italy where photographing art works (in the Vatican Museum for example) was only permitted without flash or tripod. I was constantly faced with the dilemma of using 1/15th at 15mm and ISO 800 or 1/30th at 15mm and ISO 1600. (Not really much of a dilemma. A lot of the time I'd take a shot at each setting.) But I found that I was usually able to get a reasonably sharp image at 1/15th and 15mm, but not at 1/15th and 30mm.

Can we say that the 1/FL rule should still apply to the 4 and 5mm lenses of small format cameras, or does it completely break down due to the difficulty of holding a small, light object steady?
Logged

BJL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6600
f stop/sensor size realtionship
« Reply #19 on: May 04, 2005, 12:23:24 pm »

Ray,

   perhaps I should put it this way. You and many others seem committed to the idea that a smaller format has little or no advantage in image quality in any situation, but perhaps the more relevant question is the other way around:

When does a smaller format has distinct disadvantages in image quality, enough to offset its clear natural advantages? [Added:there certainly are cases where such disadvantages exist and create an advantage for a larger format!]

About these natural advantages, something should be clear, since it has been going on at least since the invention of the enlarger started pushing formats smaller than those needed for contact printing. Higher sensor/film resolution (e.g. smaller photosites), and the smaller formats and shorter focal length lenses allowed by that higher resolution, generally lead to reduced size, weight, and cost. This in turn can increase the chances of being in the right place at the right time with the right equipment for the shot, thus increasing the quality of some images.

A tie on basic image quality is an effective image quality win for a smaller format.

Indeed this is driving a very strong overall trend towards using smaller formats in digital than were used in similar situations with film. (Latest victim; all but one of the Hasselblad 6x6 format cameras have joined most Bronica and Fuji MF cameras in the museum.) While we debate theory, sensor design efforts are overwhelmingly focused on reducing photosite sizeand enablig wider use of smaller formats. I wonder if some people are so out of touch with reality that they still have hopes for a low priced 6MP 35mm format sensor? (Not you Ray; I am sure you have dropped that idea by now!)
Logged
Pages: [1] 2 3 4   Go Up