This article at Luminous Landscape claims that Nikon, Canon, and Sony silently boost ISO when their cameras are used with very fast lenses (f/1.2 and f/1.4 principally), the implications being that (a) you may as well use a slower lens and increase ISO yourself, and (b) this practice is shady.
I'm skeptical, but I had a hard time parsing the article. Are the authors on to something? Is this an unfounded accusation? Or did I misread the article in some other way?
I am also very skeptical about this article. If that was true, then opening the aperture past a certain point should not make any difference in the defocusing ability of the lens.
I tried a small experiment: these are pictures of a couple of street lights close to my home. I set everything to manual and used the exact same settings for all the pictures: same ISO, shutter speed and defocusing. Only the aperture was different from shot to shot.
As you can see, the blur discs increase in size all the way to 1.4. Additionally, the surface brightness is about constant, which would not be the case if the ISO was changing.
Update 1: To address che's point, I tried the same experiment, but this time with the blur circles near the corner of the picture, instead of at the center. This is intended to maximize the light ray's angle of incidence. Here is a composite at f/1.4:
The angle of incidence is maximized in the far corner, because those light rays come from the top-right edge of the aperture and fall on the top-left corner of the sensor.
There seems to be a slightly lower brightness in the corner compared to the center, but it is hard to say whether this comes from the sensor or the lens (or the classical cos^4 illuminance law). Dubovoy's article sounded like the sensor would be completely blind past some angle. I cannot assert from my experiments that there is no angle-dependent sensitivity in the sensor, but if there is, then it is far from being as strong as the article suggests. At least the claim that “the marginal light rays just don’t hit the sensor” seems to be a gross overstatement.
Update 2: I had some correspondence with the author of the article, Mark Dubovoy (not Michael Reichmann, my mistake). After trying to dismiss my evidence with bad arguments (and after my lecturing him on geometrical optics, which got him upset), he now barely acknowledges that “It may very well be that with your camera and with your lens the issue is negligible.” But he still stands by his position, believing this issue may still affect “a significant number of camera/lens combinations.”
For those of you who would like to know whether their camera and lens is among this “significant number”, here is the the way to do a quick test:
If the blur discs increase in size with increasing aperture, then you are fine. You should then notice that the discs have the shape of the aperture (you can count the number of blades). If the size of the blur discs stops increasing past a given aperture, then Mr. Dubovoy is right, at least for your camera and your lens.
There is a well known effect, called vignetting. It depends on lens construction (faster lenses suffer more), and also on how well is the sensor able to capture out-of-axis light rays. You can see the measurements in almost all lens tests, for example EF 24-70 f/2.8 can go as far as 2 EV on full-frame camera.
Recent Canon DSLRs have a function called Peripheral Illumination Correction, which brightens the corners in post-processing. If you want you can interpret it as "silently booting ISO", and if you don't like it you can turn it off in the menu.
First off, I am VERY skeptical of the results provided by DXO-Mark. I have never understood their numbers, and I don't really think their results reflect real-world performance or behavior. They are probably extremely accurate purely scientific results, relative to their own domain, but I don't think that is helpful to normal people doing normal photographic work. My own rather cheap Canon 450D, with its pretty basic, entry-level sensor, was rated as having 10.8 stops wroth of dynamic range, and 21.6 bits of color information. I know that neither of those facets of information are true, as I most certainly do not get 21.6 bits of color information, and I have to work pretty hard to barely get 9 stops of dynamic range...I usually get 7-8 stops at best.
That said, I started getting skeptical with the article when I read the following:
When you look at the structure of CMOS sensors, each pixel as basically a tube with the sensing element at the bottom. If a light ray that is not parallel to the tube hits the photo site, chances are the light ray will not get to the bottom of the tube and will not hit the sensing element. Therefore, the light coming from that light ray will be lost. It appears from this graph that when using large aperture lenses on Canon cameras, there is a substantial amount of light loss at the sensor due to this effect. In other words, the "marginal" light rays coming in at a large angle from near the edges of the large aperture are completely lost.
[Emphasis added]
Outside of considerably older digital cameras, all digital sensors these days use microlenses above their pixels. These microlenses are designed to direct off-axis light down into the pixel well. The "marginal" light rays coming from large angles are not completely lost. Some are reflected, some are captured.
For all of DXO's talk about the accuracy of their tests, and their down-talk of camera manufacturers "cheating", they don't really tell their own customers how their own product really works. How exactly are they measuring this light loss? Is it truly accurate?
In my experience, and admittedly I have only used Canon bodies, so I can't speak for others. If I set my ISO to automatic, I get some oddball ISO values in my pictures based on the EXIF data. ISO 160, 240, 320, 480, etc. If I set my ISO to a specific value, it is always that value in the EXIF data. Granted, it is certainly possible for a camera manufacturer to truly try and cheat, tell you it is using ISO 100 when in actuality it is using ISO 200, but it is a little hard to believe they would actually explicitly change the EXIF data to hide that fact from their customers.
It should also be pointed out that ISO "settings" and actual analog readout levels are never in sync in the first place. On a Canon body, an ISO 100 is close to that, but I've seen various tests that indicate the analog readout is anywhere from 80 to 120 depending on the sensor. There have been similar tests for Nikon sensors as well (which probably apply to all Sony sensors given thats what Nikon currently use.)
I don't think the story is as cut and dry as Camera manufacterers are gaming the system. There are physical difficulties in manufacturing sensors that prevent the analog readout from exactly matching the chosen digital ISO setting, fine microlens structures that mitigate a lot of this supposed light loss at the photosite, and fairly advanced algorithms that, to my knowledge, work to maintain the accuracy of the settings you have chosen, not the other way around.
[NOTE: I would like to provide a more accurate description of what DXO-Mark actually does, however, predictably, their site is not accessible at the moment. I'll have to do some research to see if they do offer any detailed specifications or other information about exactly how their measurements work, to see if DXO-Mark are the ones trying to "game the system" as a marketing ploy.]
If I understand Mr. Dubovoy correctly, he forwards the idea that by increasing the aperture size, the incident angle on the sensor increases (faster lens w/same focal length). With a larger incident angle the sensor detects is less intensity. To suggest that the size of the aperture affects the incident angle at the sensor is technically incorrect - ridiculous. The incident angle at the sensor is determined by the geometric relationship between the focal length and the sensor size. The size of the front aperture has no effect on the incident angle (assuming equivalent focal length and sensor size). If he is suggesting something else, the article is so poorly written that I have no idea what he is trying to say.
Further he goes on to state that the increased angle causes ‘marginal’ rays to be lost off the sensor affecting the depth of field. He states the loss of this information does not produce the desired out of focus blur. Finally he says, considering all of this, one should just save the money and buy smaller lenses.
Boy did I waste big bucks for that big glass. All that increased bokeh that I thought I was seeing is just my failing eyesight. I’ll blame Adobe for that. Too much keyboard time and not enough time out in the uV rays. Them (sp) uV’s scatter at the retina and produce great focus somehow, I’m sure.
If any of this off axis attenuation theory was true, it would be observed in increased vignetting with faster lenses as others have suggested. Them (sp) sinister digital camera companies going around changing ISO without telling us. Sue them for hurting our feelings. Class action that’s the way. Lawyers get big bucks while us minions get $1.50 after filling out a form and using a 44 cent stamp. Oh, I forgot about the equivalent exposure tests I performed on film comparing my big class against old small lenses. The ISO didn’t change with the aperture size – or did it? The film must have molecules in it that determine the aperture size and compensate the ISO. The film companies are involved in the conspiracy too. Get them all - more money for the lawyers.
AxO Labs needs to be careful who they authorize to use their material. I don’t understand their data and what it is supposed to prove. I would think they would fully explain the data on their web site and clarify this article. Until then I consider the third symbol in their name to be a zero. That would make their name A times 0 or in other words, Zero Labs.
there IS some effect there, and it's easy to see it for yourself if you own a fast lens (
put your fast lens on the camera, put the camera on a tripod in a controlled lighting environment. take a picture in manual using the maximum aperture of your lens. now turn the lens in the mount, it doesn't have to be far, just enough to break communication with the camera, and take the exact same picture again.
the second picture will be less bright, because the cam doesn't know you're using a fast lens and hence doesn't apply correction. the difference is easy to see if you expose for some blown highlights - the blown area will be bigger on the brighter of the pictures. the difference will be the bigger, the faster your lens is. a 50mm f/1.8 for instance does show the effect very clearly, but it is no that strong.
I wonder why would camera manufacturers make things that complicated. If you're in Av mode with fixed ISO and fixed aperture, you can simply use shutter speed that would correctly expose the photo (including compensation of lower light transmission). There's no need to secretly raise ISO.
I read that article, and I'm not sure I buy it. DxOMark provides some interesting numbers, but they don't mean a whole lot in the real world, I think, and without a lot more details on their testing process, we're as much taking their word for it. In any case, even if the camera makers are "cheating" a little, I'm not sure I care. ISO in digital is just like a marker on the dial for gain on the sensor and are, in some ways, a holdover that allows us to compare to film equivalencies. It could just as easily be a knob that we turn until we're satisfied with the exposure value. I can see that effect when the camera selects the ISO anyways as I too get some odd values.
I have to wonder had film never existed, and we were just at the dawn of photography with digital being the option, would ISO even exist?
I suspect we have a software developer trying to make some noise to attract attention to their software - which I have found to be less than useful for my professional work.
I suspect that the writer of that article does not take into account the fact that the irradiance on the sensor is realy proportional to 1/(4Fnum^2+1) and not to 1/(4Fnum^2). This difference is negligible for Fnum>=2.8 However, for smaller Fnum one must take it into account.
The ration (4Fnum^2+1)/(4Fnum^2) explains at least some of the difference between that the author expected and what was measured.
Ofer
OK do this simple test. Take a black frame with just the body cap on the camera, with an f/1.4 or faster lens mounted and with a slow f/4 lens mounted. Measure the SNR of the blackframe. You DO NOT get the same result in all the three cases, the first and last test match but the middle test gives a different result and the RAW file comes out different. Thus the manufacturers ARE applying secret boosts to gain for fast glass. The amount applied varies from body to body.
