I shortend the article a bit in the theoretical background section in order to focus on the main effects and make it easier to read.
Rudy Leos: Hi,
The formula shows that if we use 2x crop sensor, compared to FF sensor, shooting the same Field Of View (same equivalent focal length and same subject framing) and the same aperture, we will eventually lose a total of 4 stops equivalent background blur, which 2 stops come from DOF, plus 2 stops come from Circle of Confusion size (related to magnification or sensor size).
Can we find any sample of real images showing the comparison of background blur of Full Frame vs MicroFourThird, with the same Field of View (same equivalent focal length) and same DOF ? For example sample photos of EM-5 40mm f2.0 vs 6D 80mm f4.0 (same FOV, same subject framing and same DOF).
It will prove whether shooting same subject framing and same subject size, 3 factors which are focal length, aperture and subject size are enough to determine strength of background blur. Or there should be 4 factors which sensor size must be included to determine the strength of background blur.
I guess this is the same issue I mentioned below, i.e. absolute circle of confusion b vs. what you see in the photo which is the relative size of the circle of confusion B = b / w2. The absolute circle of confusion is in fact 4 times smaller for the "same filed of view" scenario. However the sensor is 2 times smaller, so the strength of the background blur is only 4/2 = 2 times smaller in the "same filed of view scenario". Unfortunately I don't have a FF and a MFT to test, but could do so using an APS-C DSLR and a 1/1.7" compact. I know, it sounds counter intuitive, that the sensor size is not included, but indirectly it is by using the real focal length of the lens which is e.g. 2 times smaller for a crop factor of 2. Trust me, it's really only focal length, aperture and subject size ;-)
Rudy Leos: Nice article.
You formula shows Blur = B = f · ms / Nwhile ms (magnification) = sensor width (size) / subject width (size)
But, When you calculate B for the comparison table, why do you omit the magnification (ms)?
Let's say sensor size of Full Frame is "1" and microFourThird is "0.5" (due to 2x crop size, to achieve the same framing). Subject size is 0.6m = 600mm
If on 6D 85mm f1.8 = 85 x (1/600) / 1.8 = 7.87% (agree).Then on OMD EM-5 42.5mm f1.8 = 42.5 x (0.5/600) / 1.8 = 1.97% (not 3.94%).Using Blur Calculator (by Bob Atkins) also shows similar result.
Do I miss something? Please advise.
I have to admit that using small and capital letters b and B may have been a bit confusing. b (= f · ms / N) is the blur disc in absolute terms projected onto the sensor. You need to relate this to the sensor size in order to come to a meaningful metric for the strength of the blur, i.e. relate b to sensor width w2 (please have a look at the first photo above to see what I mean). So B = b / w2 which results in: B = b /w2 = f ·ms / N / w2 = (f ·w2) / (N ·w1) / w2 = f / (N ·w1)You find this towards the end of "Theoretical background" section. This means that the strength of the blur is independent of the magnification, but of course only if you use the real focal length that the lens has and not the equivalent. So for the OMD EM-5 42.5mm f1.8 you really only need to calculate 42.5 / 600 / 1.8 = 3.94%I hope this explains.
Johann3s: Nice article! However, you discard one very important effect, which is the distance between the subject and the background. Your simplified equations are only valid for when this distance is substantially far away.
There is a nice tool to show graphs of the amount of background blur for different sensor sizes and lenses. It can be found at http://howmuchblur.com
You will see that sometimes a certain lens is in the advantage for nearer backgrounds, but loses that advantage for backgrounds which are further away. It is important to also mention this effect.
Indeed you are right. The simplified calculation is only valid, when the background is far away. I personally would consider the estimation to be good enough, once the background distance is let's say 5 times the subject distance (so for a normal portrait maybe larger than 10m), but there is of course no right or wrong. And you also raise a good point, that the behavior of the background blur changes from camera/lens/sensor size combination to combination, as the closer the background gets, the more similar the behavior gets to the DoF definition.Also many thanks for the link. I had not this before. Looks very interesting.
marike6: Let the games of "Name that DOF equivalent" begin.
I personally would not discount the ability of this camera to control DoF (and I actually mean background blur). Besides the fast lens, the zoom reach is also a bit longer than for the rest of the enthusiast compacts. So when you take both into account the XZ-10 is comparable to the XZ-2 which has a shorter maximum zoom of 112mm equiv. Anyway better than Canon S110 or Fujifilm XF1, which have very limited capabilities in terms of DoF control.
Zigadiboom: Great and very relevant review... although it would have been nice to see the Canon G1X included as it is definitely within the same category of camera in terms of pricing, performance, size and target market. The overall assessment of a somewhat unique camera such as the G1X can only really be measured when compared to its peers.
I thought the equivalent aperture graph was a great inclusion and it would have been interesting to see how the Canon G1X with its very large sensor but relatively slowish lens performed within this group in relation to blurring backgrounds at the various focal lengths. When viewed in comparison to something like the Panasonic LX7 it would have also addressed a popular issue concerning small sensor/fast lens vs large sensor/slow lens.
In terms of bakground blur at 90mm equiv. the G1 X and the LX7 are pretty much the same (as you indicated small sensor/fast lens vs large sensor/slow lens) and the same should be true for shorter focal lengths. Though I can't tell from the specs how the exact graph would look like, anyway max tele and wideangle end point create the same background blur. The G1 X has a little extra zoom though.
taotoo: I thought this was a very well written article. Seemingly gone are the days of rating cameras on the numbers that they produce in a laboratory. Great efforts are made to explain that an individual's specific circumstances may make any or all of these cameras a good or poor choice for them. Well done DPReview.
Now what I would like to see is a massive diagram illustrating the depth of field obtainable in real world situations with various classes of camera/lens, from 3x compacts to full frame cameras with fast telephoto lenses. Just like the photos on the first page, but with many more cameras. It can be 5000 pixels high if necessary. We need to see that if your particular requirements are to photograph your kid at 10 metres with a wall 2 metres further afield, e.g. f/2.5 or f/4.0 on a small sensor are much of a muchness. EDIT: and what you really want is a FF camera with a 135mm f/2.0 lens.
If you are interested in how background blur relates to aperture, sensor sized and distance and not just for compact cameras, you might want to take a look at the article I wrote here on dpreview: http://www.dpreview.com/articles/6091822765/background-blur-and-its-relationship-to-sensor-sizeThere are also a couple of sample shots on page 2 ... no portraits, but sample shots of flowers to illustrate how background blur varies across the various settings.
theranman: I think the Sony RX100 is on the right track. Either a slightly bigger aperture or slightly increased zoom would be sweet, even if the camera needs to grow 10%.
I 100% support that!... or if I could even ask for a little more: slightly bigger zoom (120 -140 mm equiv.) AND bigger aperture at maximum zoom (e.g. 2.4)
wetsleet: I'm assuming that to get an equivalent FOV and DOF from a smaller sensor you need a correspondingly shorter focal length (to get the same FOV) and larger aperture (to get the same DOF).
Shorter focal length and smaller image circle mean that smaller sensor sizes yield smaller,lighter lenses, but the requirement for larger apertures works in the opposite direction.
What I would really like to know is this - what is the optimum sensor size to achieve the smallest. lightest lenses - assuming you want to achieve FOV and DOF comparable with 24x36mm sensor area (aka FF).
So when I look at lenses like the Sigma DC 50-150 f2.8 I wonder, if it were instead designed to an aperture to give an equivalent DOF as their DG 70-200 f2.8 would it then be just as big and heavy?
So what is the optimum sensor size to allow the smallest lenses to give FF equivalent DOF and FOV? When I know that, I'll know where to start with my deream camera!
Yes agreed. I am hoping since a couple of years, that someone would take the risk and develop a enthusiast pocketable camera, with the right balance between IQ, sensor size, camera/lens bulk and maintaining the ability to create background blur. Unfortunately this hasn't happened yet. Let's hope one of the camera manufacturers is courageous enough and takes the lead.
I will buy the first camera to get to market with the following specs:- Sensor size of 1” or above- Max aperture <= 2.4 at max zoom- Retracting lens to achieve compactness.- “Thin” body with a maximum of 40-45mm- Zoom Range: Minimum <= 28mm to maximum >= 120mm equivalent, ideally more.
I think this is not an easy question to be answered, but obviously as you state, the smaller the sensor, the smaller the lens. And if you want the have the same background blur (not the same DoF) you are correct, then you will have to make the aperture larger. This would however change the diameter of the lens in the first place, not necessarily so much the length. In addition you need to consider how much the lens can be retracted to make the body compact. So all in all quite complicated.
I personlly would start with a 1" sensor, and an f/2.4 lens at max aperture.
BTW, the relationship between DOF and sensor can be somewhat confusing. If you are intersted have a look at my article at :http://www.dpreview.com/articles/6091822765/background-blur-and-its-relationship-to-sensor-size
Pocketable camera, capable to create some background blur in portrait situations:
- Sensor size of 1” or above (noise performance and background blur capabilities)- f-number <= 2.4 at max zoom (for a 1" sensor calculated using http://www.dpreview.com/articles/6091822765/background-blur-and-its-relationship-to-sensor-size, portrait height = 0.7m, f = 45mm (120mm equivalent), B = 2.25% , i.e. reasonable background blur).- Retracting lens to achieve compactness.- “Thin” body with a maximum of 40-45mm, when the lens is retracted. Ultimately the thickness will determine whether the camera is pocketable or not.- Zoom Range: Minimum <= 28mm to maximum >= 120mm equivalent, ideally more.- As a fast lens will require a bigger lens opening, most likely leading to a compromise with respect to the viewfinder. I could do without it, if it keeps the package small and makes the lens faster.
Thanks to all of you. I am happy to hear, that you think the article is helpful.
Michael_13: Very interesting, thank you, Carsten!
Your results also show an often overlooked disadvantage of FF sensors like in Sony's RX1: Even at 35mm you have to deal with almost 3% blur. This means that indoors, when taking pics of people, you need to use F4.0 or higher and push up ISO. Compacts like XZ-1/2 can use open apertures an still provide enough DOF.
Thanks, Michael!And the additional point you added is indeed a very interesting aspect, I have never thought about before. Thanks.Carsten
By coincidence I just published an article that is related to the section above "Enthusiast compacts: lenses, sensors and background blur". Anybody that is interested in some additional viewpoints (like e.g. which camera is capable to product background blur for a typical portrait situation), please have a look at:http://www.dpreview.com/articles/6091822765/background-blur-and-its-relationship-to-sensor-size