Format Comparison Results

Last week my friend Dan came over with a couple of medium format cameras. (As a quick aside, Dan and I do a podcast called Circuitous Conversations with Bill & Dan if anyone is interested in that sort of thing) The idea was to take the same or very similar picture with a number of different cameras and formats in a controlled setting to see how much of a difference there actually is.  Now I'm sure a number of other people have done similar comparisons over the years, but often they're of some building outside or a still-life of random objects on a table.  Pixel peeping porn for sure, but not really meaningful to me as a portrait photographer. To that end I asked my other friend Chris Keeting (yes, I only have two) to come and sit very still for a couple hours. Btw, Chris is @analognation on twitter is a painfully witty guy whom you should follow.

In the running were the following:

  • a Canon 5D Mark II with a 50mm/1/2L prime at ISO 100
  • a 1972 Hasselblad 500CM with 80mm lens loaded with Kodak Portra 160 film
  • the same Haselblad 500CM with a Phase One P65 digital back at ISO 100
  • a Phase One 645 body with 80mm lens and a different Phase One P65 back
  • and for good measure, a Cambo 4x5 with 150mm lens loaded with Fuji 160S

The low-res results can be seen here: http://ontakingpictures.billwadman.com/2011/02/format-comparison-teaser/

This is a surprisingly hard thing to do for a number of reasons.  First is having the subject in the same position, and Chris did a wonderful job in this role. Though even subtle changes in the way the light hits his fact can effect how you perceive the images. Film speed is also very difficult.  I wanted to get the light set and shoot them all with the same power of light at the same aperture.  However if you look at the list above you'll see that the film cameras were using ISO 160 film. Now, that's only about a half a stop, and C41 negative film is fine with a little bit of over-exposure. So I let it slide. Plus, I was going to try to work a little curves action to get them in the same ballpark anyway.

Matching It was also really hard to get the white balance of the pictures similar. To get the two MF backs to visually match the Canon at 5600K, they were at 5180 and 5860 respectively.  I'm not sure if that's a matter of terrible calibration at the factory or if that's the way these things always act, but a 600k shift between samples, and neither was correct, is a alot.  Somehow I trust my Canon more than the $40,000 backs on this respect.  I also had to scan two different formulations of film and try to get them to match. Nulling out the orange cast of the film substrate in the process. Ideally what I should have done was to shoot a frame of a color chart which would give me a reference for everything plus neutral grays.  But alas I didn't think of it and therefore had to do it by eye.

The other big problem is that I was shooting at the same aperture, f/8, on all formats so that I didn't have to reset the light value with every switch. The problem here is that f/8 gives a very different depth of field depending on the size of the sensor or film. Especially on the 4x5 where f/8 is just one stop away from wide open.  In retrospect, I could have used a lightmeter to set the light lever for each camera while trying to compensate by stopping down the bigger cameras, but then you get into questions of diffraction effects of the lenses at that aperture and such. The practical upshot being that these are all different beasts and in some ways it's like comparing apples and oranges. All those qualifiers out of the way, I do think I did a relatively good job of weighing and mediating all of the factors.

So what did I find out? In the end, these are the points that I took away from looking at the resulting images.

First off, all of the cameras did a fine job taking the picture. All of them are professional quality. And if you're only going to be looking at them on the web or your average magazine size print, there is no practical differences in my eyes.

You all know what 20MP full frame files look like, and I'm sure that many of you have shot film, so I don't need to go into detail on those.  The 4x5 looked suitably great. However I'd much rather use slide film or b/w in large format for some reason.  In fact when it comes right down to it, I really only enjoyed shooting Polaroid 55 with it.  That leaves us with the two MF backs.

In the 'pro' side for the MF backs is the fact that there is a sick amount of information in the files.  I'd say they have more in common with a 4x5 frame of film than a 120 frame.  This also extrapolates down to smaller digital where I think my 5D2 gives me a file that's almost as good as my Hasselblad using film. Definitely way more than my Leica can.  The question then becomes "Do I need that much information?" and that's a good question.  Like I said, if you're not printing big, then to my eyes there isn't the need. Another thing is that because there is no AA filter in front of the sensor, you also get an incredibly sharp image as well. Like poke your eye out sharp and detailed. See below. MF RAW files are also 16bit (most modern SLRs are now 14bit) which means there's more data in each color channel for you to work with.  This results in extreme plasticity in manipulating the RAW data. Over-exposure of 3 full stops can be pulled back; similar results in pulling up the shadows, which are less grainy and quantized as well. You can do with with Canon and Nikon files, but not 3 stops in the highlights. Maybe 1, or 1.5 stops tops. This can be a lifesaver in very specific situations.

The 'con' side to all the above 'pro's is that your technique must be flawless. They are precision instruments and EVERYTHING you do is magnified.  You get 3 times the data of my 5d, but camera shake and focus errors are magnified along with it. I found it very hard to focus accurately enough on both the Hasselblad (manual) and the Phase (auto/manual). The viewfinders, while huge compared to 35mm, are still no large or fine grained enough to match the sensor that's ultimately catching the photons.

Speed is also a concern as these backs max out at about a frame a second. So if you're used to snap snap snapping away at a clip, you're in for a awakening. The autofocus is also slow and hard to control.  Really there's only the center point, which is fine for me since that's all I usually use anyway, but if you're trying to focus on the eyeball you may get the eyebrow instead half the time. I guess the answer here is to stop down the lens so that more stuff is in focus, but deep focus portraits are often not what I'm looking to take. Honestly, minus the backs, medium format cameras feel like 15 year old consumer film slr bodies. They're also big and heavy.  Not the kind of thing you're going to want to be holding for hours.

Oh and I should mention computers for a second.  Remember that these files are 3 times the size of the Canon's. Which means they're 3 times bigger when you open them up in Photoshop, which means you need at least 3 times the RAM to manipulate them at the same clip.  I've got 12GB of RAM in my desktop and these images are HUGE. The picture of Chris that I posted the other day, with fairly light retouching and editing for me, came in at around 1.6GB as a PSD.  If I were using one of these cameras everyday, I'd upgrade my RAM to 24GB or more without flinching.  There were times when my desktop felt like a laptop while working on them, and I've got a very fast computer.

Finally there's the question of cost.  The Phase One setup was about ten times the cost of the Canon setup. Like in all things it's certainly not a linear progression.  The Phase is not 10 times better than the Canon.  Maybe twice as good or three times as good in image quality, but only if you assume that everything else is perfect. The light, focus, tripod, etc.  On top of the initial cost of the camera is the cost of the lenses which are often thousands of dollars each. More than the price of the Canon setup.  Look, I've got a $8000 stereo, so I know the law of diminishing returns, but even I've got my limits.

Would I want one? Sure. Do I need one? No.  And that's actually a surprising result for me.  I thought the results would make me speechless with awe, but in the end I was just really impressed.  My next big gig where I've got a budget, I'll certainly be renting one. A rig like that would have been great for some of my personal projects, but I can't really complain as I've still got a setup that most people would kill for.

If you've got any questions, feel free to ask in the comments and I'll try to answer as best I can.