ManuelVilardeMacedo: This camera makes me wonder whether there will be a need for an E-P7.
Just as the E-PL7 matches the E-M10, I think it's likely that the E-P7 will match the features of a future "E-M7". So, maybe next year?
Dimitrios Pikros: Richard is 110% right about the shooting experience. I have to admit that setting up for the first time my e-pl5 for "serious" photography was frustrating (and keep in mind that my previous camera was e-pl2). I think that there is a simple solution to this: two or three different "general modes", for example "beginners", "advanced" and "pro". Each mode could offer different grade of involvement, but obviously the last one should have the Super Control Panel on by default (among other things). It's so simple that it makes you wonder why no one has thought about it...
Maybe Olympus' thinking is that the people liking the SCP are the ones capable of changing to it, but that those who aren't wont to change settings will have a harder time?
In either case the UI isn't very intuitive.
As an example of empirical data for the theory of equivalence, consider this simple graph:
It shows the noise performance of the Nikon D800 and the Olympus E-M1 for equivalent exposures. The values in the graph all come from DxOMark's tests of the cameras. As is clear, for two sensors of roughly the same generation, it all comes down to the total light!
Unfortunately, this seems to have offended some "full frame"-aficionados who think of smaller cameras as mere toys. Especially, since the E-M1 manages a fair bit better dynamic range at the equivalent exposures. I've tried to appease them by pointing out how the E-M1 is simply incapable of matching the high light performance of the D800 (without stacking), or the lack of v.large aperture lenses, but the idea that a "FF" isn't much! better than a "crop" at low exposures is simply too bitter a pill to swallow. :/
But with equivalence you have to take the sweet with the sour.
Paco 316: This is so stupid!
First of all, Micro Four Thirds IS NOT a cropped sensor, BUT a format of its own, read and repeat, a format of its own, just like Medium Format, just like 35mm.
The only cropped sensor is the APS-* type that uses FF lenses on a smaller sensor than that of 35mm.
Micro 4/3 lenses are made for the Micro 4/3 sensor there for NOT making it a cropped format but a FF, yes, read and repeat, FORMAT OF ITS OWN.
And please, F2 is F2 on ANY format, just like going 40mph is going 40mph no matter what you are driving or riding.
STOP misleading and confusing people.
While I agree that four-thirds are full format sensors, it isn't true that we only find "cropped" sensors in the APS dslrs. If you look at medium format photography, they often use cropped sensors, even though the cropped sensors are bigger than "full format 135". But that is language... it's a living thing.
As to your other point, f/2 is indeed f/2, but the point of the article is that this doesn't matter nearly as much as a lot of people think it matters. And for a lot of applications it would be better to compare the actual area of the aperture or something "normalized" like the "equivalent f-ratio".
Easycass: One last go...
Equivalence is a great word, in the cases we wish to use it for here, it essentially means "gives the same result", or at least "very similar". I may not fully understand all this, but surely, whether you have a use for the information or not, we can agree on the following: -
1. Equivalent exposure - Eg Settings of f/2.0 - 1/60s - ISO400 has an 'equivalent exposure' to f/4 - 1/30s - ISO800. Agreed?
2. Equivalent FOV - Eg a 50mm lens on 1.5 crop sensor has an 'equivalent field of view' to 75mm on a full-frame sensor. Agreed?
3. Equivalent DOF/DIF/TL - Eg Settings of f/1.4 on a 1.5 crop sensor has an 'equivalent DOF, diffraction and total-light' to f/2.1 on a full-frame sensor. Agreed?
No one is saying in any of the 'equivalences' above that the 'settings' themselves are the same, only that they produce an equivalent result: whether exposure, FOV or DOF.
Maybe my numbers above are out (late at night), but you likely get my drift.
Let's move on, hey...
I think it's helpful to use "full format" and "crop factor" in some instances. For example, only few (if any?) medium format digital backs are "full format" and it's useful to be able to distinguish between the different sized sensors on the digital backs like that.
I agree that it's basically nonsense to describe new systems designed around a sensor like Fujifilm X, Nikon CX or micro four-thirds as anything but "full-frame".
Equally 135 format cameras with a 135 format sized-sensors are obviously also "full-frame", which can be a helpful distinction since their lenses are also often used on cropped sensors, like the earlier Leica M cameras, and the APS-sized sensors on Nikon and Canon cameras.
If the context is obvious, i.e. if it's obvious if we're talking about full frame four-thirds, full-frame 645, full-frame 135 etc., I don't mind talking about "full frame" without any other qualifier. If not, it's helpful to at least once qualifying with, for example, "full frame 645 format"
dtmateojr: "And this means that, for the same shutter speed, F-number and ISO, the camera with the largest sensor will have more total light to measure. And, unless the large sensor is significantly worse than the smaller one, it will produce a cleaner, less noisy image. It's likely that the large sensor camera will be bigger, heavier and more expensive, but it should provide cleaner images."
That is not right. Noise happens at pixel level not "sensor level". Noise is affected by how much light EACH SENSEL collects and NOT the total light gathered by the sensor.
And because of this, looking at the sensel size as the main variable is a red herring. In the end, what really matters is the total light per image, not the total light per sensel. When making a portrait, how much light was recorded for the bit with the eye? How much for the bit with the nose? The forehead? The lips? The neck? For the landscape picture, how much light was recorded for the bit with the flower? The tree? The lake? The clouds? The individual pixels aren't the important bit. The picture is.
Btw, I'm not a 135 format evangelist at all. I'm very happy with the performance and size of my micro four-thirds system. I think that combination is advantageous to me in most situations. However, in some situations a system with larger aperture lenses would've allowed me to record more light, and in some a larger sensor would've done the same. For some of these situations, I can work around that limitation by doing multiple exposures, for other situations, I'll just live with it.
dtmateojr,I think we agree that sensels have a size that limit the amount of light it gets during an exposure. Where I think you go wrong is think of the data it collects (which becomes a pixel) as having similar dimensions. It doesn't. Perhaps, you're too used to using software that tells you that you're looking at "100%" zoom?
The data from the sensel is just a number with a location. Its size is what we decide it to be, but more importantly is that we don't change any inherent properties of the data point by making the pixel larger or smaller.
Because of this we can have a camera with a sensor array with four times as many sensels, four times as many data points we can transform into pixels in an output image, and it doesn't change the output image size. Whether we do this by higher resolution or downsampling, the end result is the same: For the same section of the image, a similar amount of light has been recorded, regardless of we used one sensel or four to capture it.
SNR isn't just at the individual pixel or grain level. Think of a seemingly random event like a throw of some dice. We may know the expected average of such a throw, but we're unlikely to get close to it with just one throw of the dice. But the more we throw the dice, the closer to the average we get. This isn't entirely unlike light hitting a sensor. Individually the photons behave randomly and noisy as the quantum particles they are, but the more photons that hit the sensor, the more a pattern emerge and the better the SNR becomes.
Now think of two cameras both set up to capture the same image, but one is built around a larger sensor than the other. Though the light per area is the same on both sensors, for every part of the output image, more total light has hit the larger sensor than the small. No matter if you look at the entire image, the top right quarter, or if you divide up the image in 16 million pixels and look at number 43 and therefore the SNR is going to be better.
Dude, of course the noise performance of a film is better in 120 format than 135. Why do you think medium and large format was preferred over small formats like 135?
Regardless, digital photography doesn't use film, and thinking that things work like in film is liable to only confuse matters. A digital sensor is a photon counting machine. The more photons the better the signal to noise ratio is.
No, I think it's fair to say that the article corresponds to the general consensus on this matter. For a given area of an image, the signal-to-noise ratio is dependant on the total amount of light it has received. Regardless if you look at the entire image overall or parts of it, down to the individual sensel/pixel.
Could you explain why you object to this claim?
It's a point of contention, I know, but you should be really careful about how you include the ISO value when you talk about exposure.
At its heart exposure is about the intensity of the light and the time, which roughly translates into aperture and shutter time in photography. (It's worth noting that we can of course also change the intensity of the light by using for example filters, flash guns etc.) ISO is something that comes into play after.
This is an important distinction when we talk about making a larger or smaller exposure. It has to be understood, that we do that by changing the exposure values, aperture and shutter time, not the ISO.
Besides this the fact that f/8 and 1/250 s gives an equivalent exposure to f/11 and 1/125 s already has a "name" in photography. We refer to this as "reciprocity".
cm71td: Why all the passionate opposition to an article that just tells you how to translate between formats? Is it because you don't want to accept that another camera format might be better than yours in certain conditions?
The Gaia space telescope has an image sensor the size of a 37" television. I suppose some will claim that they could have just as easily used the image sensor from an OM-D E-M1.
cm71td: That would be rather impractical since cameras and lenses are almost universally set up to use the f-ratio. While f-ratios aren't useless, I still think it's a relic from film photography, and that the optical characteristics of a lens is much better expressed by its field of view and the area of its aperture. Focal length and f-ratio just make for unnecessary mental arithmetic.
Primarily some people object to the idea that the focal ratio or f-number doesn't tell you anything meaningful about lens. That you should instead be concerned about the actual size and area of the aperture. This goes against long ingrained and practical experience manipulating shutter time, f-numbers and scene luminance to create a certain brightness for a given ISO, and old habit die hard. Especially if people don't really see any advantage in changing their way of thinking.
It also challenges the notion that noise should be considered on pixel-level instead of looking at it on an image-level.
I partly blame the camera manufacturers. They should've come up with a different system than "ISO" for digital cameras instead of trying to mimicking shooting with rolls of film and sending them to the developer.
Klarno: When you have two lenses for different formats that on their respective formats have the same angle of view, and these two lenses have the same ENTRANCE PUPIL (that is: the optical size of the aperture), then the lenses are equivalent. On their respective formats, they will have the same angle of view, same depth of field, and same amount of light falling on the sensor . These images already have for all intents and purposes the same SNR, so they're already equivalent as far as the signal capture is concerned.
What really confuses people about equivalence, particularly in this digital age, are the film-era measurements that are f-number and ISO.To understand equivalence, you need to understand the f-number and ISO concepts and take them for what they are: helpful for the shooting process, but only because that's how the photography industry developed over the past 187 years, and essentially useless for how the photographic signal is actually captured in digital systems.
You're bang on! I have one addendum, though: On most(?) cameras raising the "ISO" a certain amount can bring down read noise and thereby give you a better SNR ("less noise") when doing small exposures.
Erick L: Total light on sensor affecting noise doesn't make sense to me. Don't smaller sensors appear more noisy simply because the image needs to be enlarged more, noise included?
Seems to me that saying a FF sensor has less noise because it gathers more light is akin to saying a telephoto lens "compresses perspective".
Cheng Bao, as you write yourself, the "total light" hitting the sensor is the main variable when looking at the (shot) noise performance. Magnification or enlargement is a red herring.
Output from two sensors exposed to the same total light is going to have similar shot noise characteristics, regardless that one of the images needs to be magnified more than the other.
Aye, I assumed we were okay by using "more noise" as short hand for "worse signal to noise ratio" ;)
Nope, nerd2, you seem to have things mixed up. It's the act of cropping that increases the per-image noise. It doesn't matter if you then take your output and magnify or enlarge it to 8 feet by 12 feet or 8 miles by 12 miles: The magnification doesn't create more noise. That should be obvious, too. :)
GB: The benefit of downsampling (and similarly, visual acuity) on an image in order to improve signal to noise is an interesting phenomenon, but it should be considered separately and not as a result of magnification.
GB: Enlargement magnifies the signal as well as the noise. Is an image noisier on my 27" display than it is on my 13" one?
No, magnification in itself doesn't change how noisy an image is. In the digital world images can appear less "grainy" when you resize them, but that is a different phenomenon that is dependant on pixel count and not sensor size.
The reason why larger sensors often have a noise advantage at the low end of their ISO range is because of total light: For the same exposure, a sensor with a larger area collects more light, and therefore a better signal to noise-ratio.
It's important to note that part of the whole "equivalence" equation is that this signal-to-noise advantage isn't there under every circumstance: Because a FF camera needs to stop down in order to capture the same DoF and diffraction as, for example, a crop sensor this evens out the signal-to-noise advantage, unless it's possible to use a longer shutter time with the FF camera than with the crop sensor.
Jack Fu: Good article. However, the idea about "equivalence of total light" is wrong.
--quote--And this means that, for the same shutter speed, F-number and ISO, the camera with the largest sensor will have more total light to measure. And, unless the large sensor is significantly worse than the smaller one, it will produce a cleaner, less noisy image. It's likely that the large sensor camera will be bigger, heavier and more expensive, but it should provide cleaner images.--end of quote--
Let's say I take my Nikon D600 under low light, with same shutter speed, F-number and ISO, I take a picture at FX mode with 50mm 1.4, then I take another picture at DX mode with 35mm 1.4. Do you think the first one is "cleaner, less noisy"? Or, I take two pictures with 35mm 1.4, one in FX mode and one in DX mode, the one I take in FX mode would be "cleaner, less noisy"?
The key factor here is the density of pixels, and we knew it since long time ago.
It is possible to overstate the importance of shot noise though. It's a contributing factor but not the whole explanation to why when we record light in our cameras to create images, we get less noise when we record more intense light. The fact that more signal "drowns out" the constant "buzz" of electronic read noise is still significant.