David Jacobowitz: Important article, but that no credit or mention given to Joseph James is sincerely Not Cool.
He was arguing and explaining equivalence on these very forums years ago, and has had a very comprehensive essay on the subject (more so than this piece) for as long.
I find it highly unlikely that Richard Butler did not even look at that piece:
@Horshack - this article was written in response to arguments and misunderstandings I'd seen on this site. I can't know to what extent either of those things stem from Joseph's writing.
audiobomber: Thanks DPR. This article is overdue. Unfortunately some seem to be still clinging to f1.2=1.2=1.2.
To better understand equivalence, ignore ISO. ISO numbers are manipulated by manufacturers and worse, ISO obfuscates the differences in sensor performance. Noise matters, not ISO. ISO is an artificial and flawed construct.
If we were seeing SNR instead of ISO displayed in the LCD panel, there would be no equivalence debate, and cameras could be compared on an equal footing.
We are working on a new noise test that will give a much more comprehensive understanding of noise behaviour. Bear with us.
Michael Piziak: I'm sure this is a good article, but I lost interest around the 5th paragraph.
I did arrange the pages so it's easy to just flick through to the pictures - they make the point more clearly than I could hope to.
DonSantos: So Panasonic is selling their 2.4 equivalent lens for 1599$ :)
I've never quite understood Panasonic's approach to lens pricing.
However, it's worth noting that it does have rather nice bokeh and is the only image stabilized lens here.
David, you may find it unlikely, but this article is not based on Joseph's piece in any way. It was in part inspired by having to moderate the arguments on our forums, but that's not necessarily something I'd give credit for.
@falconeyes - I take your point. I was trying to make sure I explained the idea behind equivalence of total light (which, as you say, is fundamental to the idea), while also having to explain why you don't always see it.
Trying to keep this relevant to a broader audience meant trying to keep it related to real-world results, and trying to relate things to the same silicon design rules risked making it seem even more esoteric, given that people experience many different sensors of different generations.
photoguy622: Let's not forget the flip side, that a smaller sensor has the benefit of having a great area in focus.
mostlyboringphotog - it's a red herring because it creates a circular piece of reasoning.
A large format camera can give the same depth-of-field, diffraction and total light (so what in theory will be a near identical image) to a small sensor camera, so long as the lens on the camera lets you stop it down far enough.
As soon as you demand the same f-number (and there's no good reason to), then the argument becomes circular:
The actual aperture size relates to the focal length which, in turn, relates to the sensor size (if we are to match field of view). So, by insisting on the same f-number, what you're saying is: a smaller sensor gives more depth of field if you use a smaller aperture.
Or, more simply, that a smaller aperture gives you more depth of field. This is clearly the case.
However, matching f-numbers an artificial and arbitrary demand. If the larger sensor system can match the actual aperture size, then it can give the same depth-of-field.
Happy Jones: In the article is shows an APS sensor does not use the full aperture of a FF lens. Since f-stop is focal length divided by lens diameter and has no relationship to sensor size, this is confusing. Couldn't an APS lens be made with a much smaller diameter since much of it is not being used and still have the same light gathering? Are we saying an APS F1.8 lens projects more light on to an APS sensor since f-stop does not change with sensor size? An APS lens will vignette on a FF sensor because it concentrates light more to the center than a FF lens. Say for a 50mm lens couldn't a lens direct all the light of a 28mm opening towards an APS size sensor rather than spreading it out over a FF size sensor? We would see vignetting on the FF sensor with the lens because the light is more focused. To me then this F/1.8 lens would put the same amount of light on the APS sensor (but more concentrated to a smaller area) as a FF lens would on a FF sensor.
"*The article shows an APS sensor does not use the full aperture of a FF lens*" - that isn't what it's showing:
The diagram shows the effect of putting a larger format lens on a smaller format. Both are using the full aperture - the light intensity and f-number remains the same.
However, the diagram shows that the angle of view (equivalent focal length) changes, as does the *total* amount of light being captured.
The lens is still has the same f/ number, so the light intensity remains the same - you are just capturing a smaller area of that imaging circle.
If you made a dedicated APS-C lens, the lens could be smaller and it wouldn't ever project the larger image circle (so couldn't be used on a larger format). However, you wouldn't get any more light - you'd still get a fixed intensity of light over an APS-C area.
quezra: Drama incoming in 3... 2...
Actually, Chippy99, I repeatedly wrote that there is no optimal format. If you want the things full frame gives you, you'd like medium format even more.
Personally, I don't shoot full frame very often but, in terms of light capture, bigger is better. That doesn't mean it always results a better camera system.
Bill T.: Damned calculators! A few full sized raws is worth 1,853.97 charts. And bring back Mr. Robot! Am I too visual for this site?
The full sized Raw files are available at the end of the article.
It's the 'At the same F number' bit that's the red herring, here.
A small sensor (with associated shorter FL, smaller aperture) will have more DOF *at the same F number.*
But, so long as the lens on a larger format can stop down enough, you can match that depth-of-field.
As totemic says, that means increasing the ISO, but you'll still have the same amount of total light so you *should* be able to end up with a similar image. Whether or not this throws away the advantage of the larger system, it shouldn't leave it at a disadvantage.
chbde: Very good article. It is however slightly incorrect on the topic of diffraction:"This is because, like depth-of-field, softening from diffraction depends on the actual size of the aperture, not the F-number."
As "Cambridge in Color" (http://www.cambridgeincolour.com/tutorials/diffraction-photography.htm) explains it very nicely, the size of the diffraction pattern on the sensor depends only on the f-number. This is also the reason why diffraction softening is independent of focal length. But one can see the effect of diffraction if the size of the diffraction patterns exceeds the dimensions of a pixel, so it is dependent on pixel density.
"This means diffraction will have the same impact on two images shot at equivalent apertures"
That sentence remains correct, however, for sensors with equal number of pixels.
As a consequence of this, by the way, on a 16MP 2.3" compact the resolution is already diffraction limited at f2.8!
"Although diffraction depends on F-number, its impact on the image is proportional to sensor size."
(Since an Airy disc of the same size on two different sized sensors will be different proportions of image size)
@mostlyboringphotog - a smaller sensor does not have any advantage in terms of offering more depth of field. While the situation you describe is true, the requirement to shoot at the same F-number is the only thing that makes it so.
Stop down a large-sensor system and, so long as the lens will stop down far enough, you can get the same depth-of-field (and you'll get the same diffraction).
Eric Nepean: Thank you for a good article Richard, On e additional point that may be worth noting is that in the far distance background blur is simply proportional to the size of the entrance pupil diameter of the lens i.e. focal_length/Fstop - for primes, crudely, the diameter of the front lens element. But in the mid distance background, the larger sensors create more blur than than the smaller sensors. This can be seen here (set the far distance slider to 100m or more): http://howmuchblur.com/#compare-1x-100mm-f2.8-and-2x-50mm-f1.43-and-1x-50mm-f1.4-and-2x-25mm-f1.4-on-a-1m-wide-subject
I'm not sure I quite understand why you've selected those numbers in your comparison. But I also think we're looking at different problems (I'm trying to explain what lens/sensor combinations behave the same, not how does the same lens behave on different sensors).
If you look at [these examples](http://howmuchblur.com/#compare-1x-75mm-f3-and-1x-50mm-f2-and-1.5x-50mm-f2-on-a-1m-wide-subject), you can see that the 75mm F3 on full frame gives the same background blur as a 50mm F2 on APS-C, at *all* distances.
Yes, if you mount a 50mm F2 on both cameras and shoot them to match the proportion of the frame your subject takes up, then far distant objects will be similarly blurred, and the large format will have the shallower depth-of-field that we expect for nearer objects.
However, because the two 50mm F2 images aren't shot the same distance from your subject, the perspective effects on your subject will be different and the pictures won't look the same.
Artpt: Thank you DPR for the informative article.
The take home message hear is that maximum depth of field requires the smallest sensor size...clearly cell phones win out here, with their incredibly large depth of field...;)
As Lee Jay points out, angle-of-view is relevant:
Put a lens on a smaller sensor and you'll get a cropped view. You'll then have to step back to get the same image framing. *That's* when your depth-of-field decreases.
50mm f/2 lens on APS-C gives the same framing as a 75mm lens on full frame, so it's reasonable to assume you want the same framing as that 75mm lens.
So you step back and get the framing, depth-of-field and diffraction of a 75mm f/3 lens on full frame.
If you don't include the stepping back part of the process, then you'd get slightly shallower depth-of-field but a rubbish photo because you've chopped the edges off it.
@audiobomber - we're not 'clinging' to anything, we're explaining how this works within the current exposure framework and terminology. Inventing a new way to describe exposure isn't exactly going to simplify it for anyone, or make it seem relevant.
@Just a Photograph - the diagram does indeed show a larger-format lens on a small sensor, to show we're not claiming an f/1.2 stops being an f/1.2 lens when mounted in front of a different sensor.
However, the point is that, while the smaller sensor still receives the *intensity* of light that f/1.2 denotes, it only uses a smaller area to capture it (so less light overall).
It's true that a native lens for a small sensor doesn't project light beyond the image circle of the sensor, but this doesn't mean it's passes more light. With a native f/1.2 lens, you're not 'spoiling' any extra light but you're not capturing it, either. You're still receiving f/1.2 intensity over a smaller area.
How about a high ISO video still comparison? ISO 3200 on the RX100 II vs III ??
Your video still comparison doesn't even list ISO!
The video comparisons are all shot at base ISO, at present.
We will certainly consider a low light test.
chillgreg: Images please DPR tech staff?(Flagged as inappropriate to gain attention)
Sadly it didn't gain our attention at the time.
I've just rediscovered this article, uploaded the images and fixed all the links.
Sorry about the delay.
Just another Canon shooter: You could also try it stopped down a bit.
Slotted down a bit?
ChaosCloud: In the specs comparison table with the RX10, the FZ1000 viewfinder is noted as LCD. I thought it was an OLED, same as in the GH4. Please confirm.
@ChaosCloud - that only detects whether it's a well implemented OLED.
The Samsung NX10 had an OLED rear display but it still gave off light when showing a completely black screen.
Juck: More hipster twaddle. Like fitting wheels to a tomato,,,, time consuming and totally unnecessary.
Thank you, Dr Johnson.
The article acknowledges that some aspects appear to be pure hipster bait, but that as a creative tool it's still interesting.