[Drake Avenue]

I picked up my entire 6x7 medium format kit for under $700

[Tony Belding]

The price is the concern there. For around $700 more than the K3 you can get a FF camera. Entry-level medium format is about $7000 more. The lenses are also extremely expensive... and not especially fast, if you'e looking for a low-light advantage.

And thus why some people are still shooting medium format -- and large format -- film.

This, to me, is one of the great unsolved technological problems in photography. Somebody will figure it out and make a fortune.

[Dartmoor Dave]

I've been casually following this thread as it has worked through the usual jsherman999 and Class A vs normhead and clackers routine (for the record, I'm with normhead and clackers). I'm left wondering if there has ever been a properly conducted double-blind test to establish whether or not people can consistently see a difference in quality between FF and APS-C.

A Google search just throws up endless discussions on various forums in which somebody posts a couple of random shots in the two formats, and then everyone argues about the methodology. The same thing has often been done here on PF, but it's not what I'm looking for. The gold standard is a well-conducted double-blind test under as close to lab conditions as possible, so that we would finally have some meaningful data about whether or not the human visual system can consistently perceive the tiny variations in technical quality that everyone is arguing about here.

I'm reminded about the endless debates in audiophile forums about whether cables and amplifiers and power supplies can really sound different. And of course there have now been enough double-blind tests to prove that they don't.

(Of course someone will counter with the argument that there are no measurable differences between audio cables, but there are measurable differences between FF and APS-C. But my point is that we need double-blind tests to establish whether or not human beings can reliably SEE those differences. If nothing else, it'd be fun to discuss how such a double-blind test would have to be conducted.)

[max_pyne]

double-blind

What would this require?
* two cameras APS-C and FF with the same amount of Pixels?
* the same lens on both cameras?
* two different Lenses 35 / 2.8 vs 50 / 2.8 try to create the same image under Lab conditions?
* post the images without exif?

just a start..:-)

[rawr]

I'm left wondering if there has ever been a properly conducted double-blind test to establish whether or not people can consistently see a difference in quality between FF and APS-C.

There have been lots of formal and informal tests that show that 'normal' people (and even many a photographer ) will struggle to tell the difference between 3MP or 12MP poster sized prints. So I doubt whether most people would be able to discern any difference between FF vs APS-C when the scene is shot the same way, and without any attempt to play to the specific strengths or weaknesses of either format.

[Rondec]

I've been casually following this thread as it has worked through the usual jsherman999 and Class A vs normhead and clackers routine (for the record, I'm with normhead and clackers). I'm left wondering if there has ever been a properly conducted double-blind test to establish whether or not people can consistently see a difference in quality between FF and APS-C.

A Google search just throws up endless discussions on various forums in which somebody posts a couple of random shots in the two formats, and then everyone argues about the methodology. The same thing has often been done here on PF, but it's not what I'm looking for. The gold standard is a well-conducted double-blind test under as close to lab conditions as possible, so that we would finally have some meaningful data about whether or not the human visual system can consistently perceive the tiny variations in technical quality that everyone is arguing about here.

I'm reminded about the endless debates in audiophile forums about whether cables and amplifiers and power supplies can really sound different. And of course there have now been enough double-blind tests to prove that they don't.

(Of course someone will counter with the argument that there are no measurable differences between audio cables, but there are measurable differences between FF and APS-C. But my point is that we need double-blind tests to establish whether or not human beings can reliably SEE those differences. If nothing else, it'd be fun to discuss how such a double-blind test would have to be conducted.)

The easiest shots to tell the difference are in wide-ish angle shots with relatively narrow depth of field. In landscape shots, or shots with similar depth of field or, shots taken with longer lenses, it is really tough to tell the difference. Pixel peeping you probably can (assuming similar number of pixels between the two), but I don't know what that really means.

Jay has posted several shots taken with roughly equivalent lenses between the two formats and the full frame version clearly has less depth of field. But I don't know how easy it would be to really sort that out, if you didn't know what lens took a particular photo and you just had to guess format.

















I think it is probably pretty easy to tell full frame shots based on the differences between film and digital, but I doubt if they were shot on fairly similar cameras, say a D600 and K3, that you could easily sort them out.

[Dartmoor Dave]

Thank you to those who have responded so far for taking the double-blind suggestion in the spirit it was intended.

My own suggestion would be to use one full-frame camera and lens for the test -- let's say a 24-70mm zoom. Lock the camera securely on a tripod and take a photograph in full frame mode with the zoom set to 51mm. Now switch the camera to APS-C mode if possible. If not, we can just crop to APS-C size later. Take another photograph with the zoom set to 34mm, using exactly the same aperture, shutter speed, and ISO. You've now got two shots with the same camera and lens, with identical fields of view, with one full-frame and one APS-C.

Advocates of equivalence will argue that the APS-C shot should be made one stop wider to make the depth of field match, but that would mean changing either the shutter or ISO, which might bias the results. My suggestion is to limit the test shots to subjects in which everything in the frame would be within the depth of field at both 34mm and 51mm, so eliminating depth of field as a variable.

Take a number of shots using this method, then show them to as wide a selection of viewers as possible, with the person who is showing the photographs to the viewers unaware which ones are FF and which are APS-C. Ask the viewers to express an A-B preference between the pairs of photographs, and make sure that sometimes they are shown the FF or APS-C image twice.

If you could do this with at least few dozen viewers, you would start to get some meaningful data to establish whether or not human beings can consistently see a difference between full-frame and APS-C.

[clackers]

My own suggestion would be to use one full-frame camera and lens for the test -- let's say a 24-70mm zoom. Lock the camera securely on a tripod and take a photograph in full frame mode with the zoom set to 51mm. Now switch the camera to APS-C mode if possible.

Personally, Dave, I think a great test of Total Light is to leave it at that.

To remain scientifically valid, only one variable changes - format size/field of view. Crop, if you like.

Changing two variables at the same time is the way of the charlatan, so distance, aperture and focal length must be kept constant.

[Clavius]

In my honest opinion, it would be a bad thing if we could actually see the difference between a picture taken with APS-C and one taken with FF. And being able to see the difference shouldn't be one of the requirements to need an FF camera. Both because an FF camera will not improve the image you take at all. It will give extra headroom whilst taking the photo in various areas. It makes accomplishing certain things a bit easier. A little bit of extra light sensitivity, a little bit better noise performance, better (U)WA performance, etc... All those little bits combined makes it easier for the photographer to take the photo he is aiming for. That's the second reason why I don't like the equivalence discussions. (The first reason being that it confuses me.)

[rawr]

I don't like the equivalence discussions.

I don't either. Equivalence is a useful general yardstick, but different cameras and formats have different properties and capabilities. That's why we choose one over the other sometimes.

It's not a competition to see which is best, which is how the dialogue often seems to drift.

[Class A]

Advocates of equivalence will argue that the APS-C shot should be made one stop wider to make the depth of field match, but that would mean changing either the shutter or ISO, which might bias the results.

Not using the same aperture size for both shots would also bias the result, even if you choose a scene in which everything is "in focus".

Lenses have different performance characteristics at different f-stops and if the shooting is done anywhere near wide open then the APS-C lens that has not been opened up properly gains an advantage. If the shooting is done anywhere near where diffraction starts then the FF lens that has not been stopped down sufficiently gains an advantage because it does not create the amount of blur due to diffraction that it should in a proper comparison.

Keeping two f-ratios the same is meaningless, when you switch formats. You need to keep the aperture sizes the same and that is only possible by using different f-ratios (as the focal lengths are different).

The shutter speeds need to be the same, so that leaves only the ISO setting to compensate for the lower exposure the FF-camera receives with equivalent settings. Set the ISO value of the FF image crop-factor^2 times the ISO value of the APS-C image and then you have a fair comparison.

Unless you use different f-ratios, your experiment would not be convincing to anyone who believes that apples should be compared to apples. You can of course still run the experiment as you envisioned it, but I wonder how compelling any results would be.

My suggestion is to limit the test shots to subjects in which everything in the frame would be within the depth of field at both 34mm and 51mm, so eliminating depth of field as a variable.

Whether something is "within the DOF" is not that straightforward. Only a single plane is in focus, everything else is out of focus to various degrees. Different manufacturers have a different ideas of which degree of blurriness can still be deemed to be "within the DOF". A suitable DOF definition also depends on the viewing distance, for instance. DOF calculators make certain standard assumptions, but if these do not hold then the calculators will not correctly predict what is "in focus".

To cut a long story short, when you use different aperture sizes (by using the same value for the f-ratios), you will create different degrees of blurriness away from the plane of focus. If the subject(s), viewing conditions, etc. are such that no difference can be seen nevertheless then the test isn't very practical, is it? For instance, the normal advantage of a larger format regarding AF accuracy, becomes meaningless when the DOF becomes so deep that even AF outliers won't matter. The extreme example of an indiscriminate scene would be a closed room with no light source. All formats would take the same black picture then.

BTW, the idea of equivalence quite clearly states that there are many scenes for which both formats can produce an identical picture. This does not need to be demonstrated. As Clavius indicated, a larger format just provides more headroom (e.g., regarding DR, shallow DOF, AF accuracy, lens abberrations, enlargement potential, low-light (with the respective lenses)) that can be very useful to have, but does not influence each and every single image.

[str8talk83]

I shot APS-C Canon (7D and 40D) for 5 years before switching to FF Sony (a7). Here are my thoughts.

One thing that I haven't seen mentioned is the ability to retain decent depth of field control with "slow" lenses or when stopped down. I use an older FD 50mm f/1.4 and stop down to f/2. This provides me with sharp images and better depth of field control than if I was shooting on APS-C wide open. I have actually found that inferior lenses actually are better on FF due to better per pixel detail and (generally) less pixel density.

I also feel that for prime shooters you have a wider array of focal lengths to choose from. There are a lot of primes in focal lengths like 24mm and 28mm that can be used as wide angles with minimal distortion and small sizes. While it is true that you can buy "equivalent" focal lengths to cover your intended field of view, this can sometimes limit your options. This is more true for wide angle, but can be true for other purposes as well. I also love shooting my Nikon 105mm f/2.5 lens for portraits and FF. It gives me great depth of field control and IQ for around $100. Granted it's manual focusing, but I don't mind since my camera has focus peaking and zooming due to EVF. YMMV. I also prefer the larger viewfinder although I think this is sometimes overstated.

I think that there are limitations with FF and the grass isn't always greener. I have found that I have to be much more mindful of depth of field. When I shot APS-C I used to shoot my 17-55mm f/2.8 lens wide open almost all the time and rarely had to worry about too little depth of field. For landscapes I always use hyper focal distance now to avoid missing the shot. Before I was much lazier and would stop down to f/11, focus a third of the way into the image and be done with it. On portraits you have to be careful to avoid the "one eye in focus" situation.

The bottom line is that FF can be nice depending on your needs. I don't think that any of us really "need" a FF camera. It's just another option with Pro's and con's like anything else.

[Dartmoor Dave]

Whether something is "within the DOF" is not that straightforward.

For the purposes of the double-blind test that I've got in mind, I'd rather apply an updated version of the classic depth of field standard. The classic definition was that, if a point of light in the real world is reproduced as a point of light in an 8"x10" print, then it's within depth of field.

I'd envision using something like a 28 inch, 3840x2160 monitor for the testing. So, if a point of light in the real world appears as a point of light at that monitor resolution, it can be said to be within depth of field for the purpose of the double-blind test.

[Parallax]

I don't believe a controlled study is necessary. If FF pictures are so universally better than APS-c, they should be distinguishable on an individual basis. If not individually, they certainly should be able to be picked out of a group of randomly selected*, mixed FF and APS-c photos.

*(No one-eye-in-focus portraits though. That's a dead giveaway.)

[max_pyne]

i can do one-eyed portraits on my K7/5/3 :-)