Does Sensor Size Matter? The Test Results Are In

Do cameras with larger sensors have a specific look? Last week I posted images taken with four different sensor sizes and let you decide. Spoiler Alert: Nobody could tell the difference. 

The Test

Proponents of full frame and medium format cameras have always said that images taken with a larger sensor look different. Yes, they typically have shallower depth of field, but most photographers say that it goes beyond just that. They say that larger sensors produce "better looking images" but they are usually unable to elaborate. I decided to put this to the test. 

This is not a resolution test, we will do that in another video. This test is strictly to determine if images are inherently different coming from different sized sensors and manufacturers. 

I stood in one spot and shot my buddy Keith Bradshaw with 4 different cameras with 4 different sized sensors. To keep the field of view the same, I used a 50mm "equivalent" lens on each camera. To keep the depth of field the same, I changed the aperture as well. My cameras and settings are below. 

FujiFilm GFX 50R/ 43.mm x 32.9mm sensor/ 64mm lens f/8

Canon 6D/ 35mm ff sensor/ 50mm f5.6

FujiFilm XT-3/ 23.6mm x 15.6mm sensor/ 35mm f4

Panasonic GH5/ Micro 4/3 sensor/ 25mm f2.8

I shot each of the images below in raw, I changed only the white balance, and stacked the images on top of each other. I cropped in on all of the images to hide the 4/3 aspect ratio of the GFX and GH5 and shrunk them to 1920 pixels in width. 

The Results

Image 1: Panasonic GH5

Image 2: FujiFilm XT-3

Image 3: FujiFilm GFX 50S

Image 4: Canon 6D

The results of our polls were all over the map. With each option being chosen right around 25% of the time, it's obvious that the images are almost identical.  Any noticeable differences like color or density could easily be tweaked further to make these four images look even more similar. 

So what does this prove? Larger sensors do not produce different looking images on their own. Some of you knew that, most of you probably didn't. 

But, shooting at the same F-stop will produce a shallower depth of field on a larger sensor. This is because F-stop is a unit of exposure and works on any sized sensor. A medium format camera lens' aperture shooting at F4 will be physically larger in size compared to a smaller lens' designed for a smaller aperture, and this physically larger aperture will produce a shallower depth of field. 

This may lead you to believe that it's easier to shoot shallower images on a medium format system, but that isn't necessarily true. Most lenses that FujiFilm makes for this system are f/4 which will look like f/3.16 on a full frame camera. Of course, most "pro" 35mm lenses are f/2.8 which will produce a shallower depth of field than f/3.16. So, at least at this time, shooting with a shallower depth of field may actually be easier on a 35mm camera. 

This test/post is not meant to discredit larger sensors. There are certainly many perks to large sensor cameras that will produce an overall better image like resolution, color accuracy, bit depth, lens sharpness, ISO performance, and dynamic range. We haven't tested any of that yet... But we will. 

Log in or register to post comments

74 Comments

Previous comments
Adriano Brigante's picture

Basically, this test says: "Look, if we take away all the differences, there's no difference." Well, duh. It's physics, not magic. Nobody said the 'medium format look' is some kind of mystical thing that appears out of nowhere when you use MF. That look comes from all the parameters you cancelled in your test. Simple as that.

Usman Dawood's picture

What are those parameters?

Adriano Brigante's picture

For instance, depth of field and field of view. With MF, you can get a wider field of view with a shallower depth of field. Of course, you can mimic that with the Brenizer method but it's still part of that MF look. This test doesn't allow to show this.
Also, the fact that there's less distortion in MF for a given field of view (especially a wide field of view) because the "absolute" focal length is longer. A 15mm lens on a small sensor will have more distortion than an equivalent 30 or 40mm lens in MF. Again, the test doesn't show that because of the lens choice.
There's also differences in the transitions from in-focus to out-focus parts of the image. This test doesn't show such transitions, so that's another parameter cancelled.
And one more thing for MF cameras with a leaf shutter: the out of focus areas (and the transitions) are rendered slightly differently, because the aperture varies during the exposure, from closed to full aperture and back to closed, instead of a constant aperture.

Usman Dawood's picture

The wider field of view with a shallower depth of field is not true. The reasons for this is because you have much wider aperture lenses available for full frame than you do for medium format. For that reason shooting with an equivalent lens will grant you the same results and you can also open up your lenses even wider than equivalent too.

There is no 50mm f/1.2 equivalent for medium format.

Lens distortion has nothing to do with the sensor size; unless of course, you're talking about perspective distortion. Perspective distortion is the same with any lens on any sensor. It's entirely dependant on your distance to the subject and has nothing to do with the equipment you're shooting with.
Also if you're shooting with an equivalent lens on a smaller sensor then you will be shooting from the same distance to your subject thereby negating any differences.

Leaf shutters are found in smaller sensor cameras too so that's not a factor of the sensor and the GFX does not have any native leaf shutter lenses.

In focus to out of focus is entirely dependant on the lens and optics so again has very little to do with the sensor.

If you're shooting with an equivalent lens then the results you produce will be pretty much identical.

The Brenizer method mimics large format; it's far beyond any digital medium format sensor.

Gordon Cahill's picture

All you did was prove that lowering every shooting parameter below the weakest point of the weakest sensor and reproducing them at 1920 pixels in an 8 bit space on the web nullifies what *all* the sensors in this test are capable of. Fstoopers has really gone to the dogs lately with these weak click bait tests and links to other peoples work dominating what used to be a great resource.

Alex Yakimov's picture

Maybe they did, by the nature of the scientific approach, but nevertheless, I found the content useful and entertaining. It proves that there are no insurmountable differences in certain conditions and it is fun to learn to intuit that differences to have a chance to enhance own craft. Nice work with MF you have. What was your fav image among 4?

Brian Stricker's picture

So you mean it is flawed because they showed images the same way roughly 99% or more of all images are presented now? The "yeah but" comments are the best. BTW, I have read for years, in countless forums and blogs, about the magical quality of "FF" so at any aperture or size it should be noticeable. That was the point of this test.

The "magical" qualities of FF are dynamic range, resolution and shallow dof, all of which they completely eliminated, to the point that they could have smuggled a phone pic in there without anyone noticing. The way the test was done it was just pointless, because there is no significant finding.
Imho they should have kept at least the aperture at f2 on all cameras, or something similar.

Alex Yakimov's picture

Nevertheless, I've been able to pinpoint 6d, MF and MF3. They are not the same.

Gordon Cahill's picture

No Brian. That's not why it's flawed. It's flawed because the test removed all the things the test was meant to test. The differences are in focal length to aperture, resolution, dynamic range, blah blah blah....

What things like this does is prove that often a larger sensor is on no real benefit to a small one. I think tests that show what you have to do until the differences between formats becomes noticeable are useful and educational. This test doesn't try to show that there are no real world differences in out of camera jpegs at low resolution for use on the web.

Instead the conclusion here is stated as "Larger sensors do not produce different looking images on their own." Which is only true under conditions like these but not always when under different conditions. If I were to do some highlight recovery or large prints or heavy manipulation of the raw files the we'd have different results to the same question.

If you do tests you need to answer the question asked. Not a different one.

Gordon

yep.. my Focus goes as fast as LaFerrari if we stay under speed limit ...

Toby Seb's picture

The metadata in your 4 example photos says they all where shot with the canon 6d. Did you upload the wrong photos ?

David Penner's picture

He mentioned he used the 6d as the base. If he opened up the 6d as the "background" layer and stacked the others on top the metadata would show that the image was taken with the 6d.

michaeljin's picture

Can we all shut up and just take some photos?

Kurt Boomer's picture

The one thing that just glares at me is the abysmal highlight details in the background. With the 6d looking like something out of my iphone. The GFX is worse than a Panasonic when looking at the grass and background tree. Im always shocked at the dynamic range claims by manufacturers. It seems they weigh it heavily on the shadow side of the spectrum.

To take advantage of the dynamic range, you still need to expose correctly...

Stuart Carver's picture

The comments on the YouTube video are pathetic, pixel peeping gearheads arguing like their life depends on being right. Luckily it’s a bit more civilised on here.

This may be the dumbest test I've ever seen. You purposely set out to make each image look as close as possible to begin with. Seriously? Why don't you do the opposite and set out to show how each sensor CAN show a difference between photos instead?

Just dumb man, really.

Arthur Morgan's picture

I have a great project for you to do, instead of whining.................

Arthur Morgan's picture

Please give the actual physical size of apertures to go with the article. focal length/size (f stop ) is of interest for exposure.

This was the coolest and best test ever. I am not surprised at the results. Too many people are biased towards their own brand. Tony Northrup did a very nice test and it showed that the famous Canon colour science superiority is non-existent for many users.

Uneternal Van de Dood's picture

AFAIK there is no lens on Fuji or 645 format that produces a shallower DOF than lenses available for full frame cameras - for instance the Mitakon 50mm f/0.95.
The fastest lens I know for Fuji is the Mitakon 65mm f/1.4 which equals ~50mm f/1.1 on full frame.

I think this is more about 'how much difference does it make'? Will the extra finesse of the larger format make a difference in the real world? Sometimes yes, often, no. Reminds me of the automobile performance enthusiasts who worry about a few tenths of a second in 0-60 or in quarter mile. True differences exist... but in a trip to Walmart, or even a cross country road trip, the differences are usually immaterial. Purists will look at pixels in the corners and see real differences, but for the quality of the picture, for whether it sells or wins a prize, that is usually not the important thing.

Photos are mostly about emotional impact. The rest is detail. Pick any iconic photo from history. Does anyone really care about the vignetting or edge to edge resolution? They care about the subject matter, and how it comes through.

Some years ago, professional wine critics were shown to be easily fooled by the label, giving cheap (but fairly good) wines high ratings. Presumably we won't get to that stage.

Too many unknown variables. Does this site compresses images? What would be proper set up to be able to tell a difference, can I see I difference from a phone screen?How about a test that show a more pronounced color and luminance gradation and also would be nice to have the opinion of an expert in the matter, not just someone that takes nice pictures.

Manzur Fahim's picture

I downloaded all four photos, and the exif of all these four photos are showing Canon 6D. Wanna explain that?

A CX / 1 inch sensor with a 12.96mm lens and an aperture of F/2.07 can produce same quality image as a FF camera? Maybe an iPhone 6 with a crop factor of 7.21 can deliver the same image if it has a 6.93mm lens with an aperture of F/0.77?

They already explained it in the original article but you weren't paying attention.
They stacked the images in photoshop and when they saved them they all got the same metadata, so that people wouldn't try to cheat by looking at the metadata.

David Ha's picture

For those MF look believers, if you dont like this test, then you should bring your own test or proofs that MF has its own look in terms of background separation and DOF. Otherwise, it's just a hype.

Wouldnt taking away dynamic range, high iso, and SNR, take away the variables that make bigger sensors better? Is like a swimming competition, but without the water so all have to go walking? Also regarding DOF, is about wider FOV and thinner DOF, not 50mm or 85mm lenses, try getting the dof of a 35mm 1.2 ff for example with a m43 lens, or a 24mm 1.4. The wider and thinner you go, the harder it will be to get the same results with a smaller sensor. Perhaps full frame is the best compromise between bigger and smaller sensors, but it definitively gives you better results than m43 or apsc.

Ahmed Gadou's picture

I guess you are directionaly right, they all look the same in this test and it is kinda impossible to tell which is which. Considering that you mentioned clearly at the beginning of the video what is the idea of the test. I think what is driving the people crazy is that it feels like bringing 4 cars toyota corolla, ford f150, lamborghini and tesla and test if they can all drive at 60 miles per hour and then ask who did it better!?

More comments