The surprisingly simple difference of large-format cinematography
There’s always been an interesting sort of mysticism that’s surrounded the idea of large-format cinematography. We associate it with big pictures projected on even bigger screens. The medium of the grandest epics of all time; Lawrence of Arabia, Ben-Hur, 2001: A Space Odyssey etc.
It’s bigger, so it must be better. Right?
For a long time now the language we’ve used to describe the impact of larger formats has been pretty vague and subjective. We’ve talked of the longer lenses (needed to achieve an equivalent field of view on a larger format) providing “increased magnification”, and “greater compression” to the image. We’ve talked of “rounder” images with “greater dimensionality”.
Now, in the interest of full disclosure, I’ve been as guilty of proselytising these vague descriptors as anyone else. I’d always heard other cinematographers use the same hazy phrases (and there was clearly something special about the photographs I took on medium format 120 film compared to regular 35mm) so I suppose it seemed the obvious choice to simply continue parroting the party line.
Combine this with the added benefit of not being able to get anywhere near a real large-format motion picture camera, and I was able to comfortably and confidently trust in my natural lack of scientific rigour, to keep me from worrying much about aesthetic qualities that were so obviously intangible, for years.
That’s not so easy to do today.
With the arrival of a whole crop of readily accessible larger format digital cinema cameras to the market (Red’s Monstro, Sony’s Venice and now Arri’s new Alexa LF) those same vague, but tried-and-true descriptors leapt to the fore again in people’s re-invigorated discussion of larger formats. But this time something lab-coated and clipboard-carrying stirred deep within me, and I found myself suddenly wanting to try and quantify what the actual, objective differences in aesthetics might be.
And at Arri Australia’s Melbourne launch of the new Alexa LF, I got my chance. I was able to get an hour alone with the new camera and some Signature Primes, and with the help of the lads from Arri, threw together an impromptu little test, to investigate an existing theory I had as to the objective aesthetic differences between a conventional spherical S35mm image (23.76mm x 17.82mm on a regular Alexa), and the exact same image when captured on the larger Vistavision format (36.7mm x 25.54mm on Alexa LF).
My theory is that once you match field of view and depth of field between the two formats, there is no discernible difference, only an improvement in image quality as your sensor size gets larger.
Fortunately, this is quite easy to test when comparing S35mm to Vistavision, as the difference in depth of field between the two formats (in most cases) is approximately 1 stop, and the difference in field of view is approximately 1.5x. And these parameters are relatively simple to replicate.
For the test we had the Alexa LF, two Signature Primes, a 25mm T/1.8 (for the S35mm image) and a 35mm T/1.8 (for the Vistavision image), and Arri’s new Full Spectrum NDs (FSNDs) which we used to match our exposures at varying ISOs.
Now the 25mm and 35mm match their respective fields of view very closely, but they are a fraction off being exact, so there is a minute difference in optical compression between the two. But for the purposes of this test, and by the real world measure of a cinematographer looking for a specific field of view/level of optical compression from a lens, for a specific image format - they’re close enough as to make no difference.
So with the images matched for field of view, the elements left to consider are depth of field, and image quality. As far as depth of field is concerned, with the larger format stopped down one stop to T/5.6 (to match the T/4 on S35mm) I can discern no apparent difference in amount or quality of focus falloff, and subject isolation appears pretty much identical between the two formats once both FoV and DoF are matched.
Now some people might want to argue that at identical apertures, the larger format will have less depth of field and greater subject isolation than the smaller one (which is true). However, due to the one stop difference in depth of field, if you were to shoot an Alexa LF wide-open at T/1.8 on a Signature Prime, you’re going to have basically an identical image to an equivalent S35mm Alexa with a Master Prime wide-open at T/1.3 - which renders it a bit of a moot point.
Obviously you can take the aperture arms race to it’s ultimate conclusion and compare an Alexa LF with a Sigma Cine Prime or Tokina Vista Prime wide open at T/1.5, to a S35mm Alexa at T/1.3 on a Master Prime and the LF will (at that point) beat the smaller format for shallow depth of field, but it’s just not that big of a deal otherwise.
Which brings us to our last obvious point of comparison, image quality.
And this is where we finally see the simple difference of large format photography - it’s BETTER.
With literally twice the imaging area and double the pixels, the improvement in image quality is substantial (and immediately apparent), resolution is significantly improved (see how much clearer our model Evan’s eyes are, and how much more detail is captured in that magnificent beard of his), and noise is half the size compared the smaller format (which provides the added benefit of allowing you to push the camera’s sensitivity almost a stop higher than the S35mm format, and still get equivalent results).
Given how great the standard S35mm Alexa image already looks on the big screen (and small ones too), to see such a profound jump in quality is pretty exciting.
Based on this test, and how well the smaller noise pattern cleans up in Davinci Resolve, I’d be comfortable shooting the Alexa LF at 3200 ISO if I had to (something I wouldn’t dare to do on the S35mm model, where I generally top out at 1600 ISO). Which, conveniently, should help assuage the fears of focus pullers everywhere, who are worried about the shallower depth of field of the new format - simply up the ISO a stop, close down the aperture a stop and you’re back to where you were with S35mm, but with greater image quality.
The other optical benefit of the larger format, should be that (as with the noise pattern) it will proportionally shrink the appearance of any optical aberrations like CA, or fringing around out of focus highlights. Now this is unlikely to be much of an issue on the shiny new Signature Primes, but could prove useful on more mortal glass.
But don’t take my word for it, I’ve uploaded the test in its full 4.4k ProResHQ glory to Vimeo, so download the full version, pull it into your grading software and have a look for yourself. I’d love to hear your thoughts. The only processing that I’ve applied to the images is the standard Arri Rec709 LUT.
**A big thanks to the team at Arri Australia for making this possible (and for supplying the cheese and the delightful Pinot Noir that sustained me through these tests). GM Brett Smith, Sales Manager Sean Dooley, our beautiful model Evan Andrews, and Product Manager Optical Systems Thorsten Meywald for a wonderfully informative chat about glass.
**A shout out too, to the Cinematography Mailing List for kicking off the discussion that’s led to this. And in particular to DP Stuart Brereton and techwizard Adam Wilt, whose calm and measured approaches to their own tests and analysis convinced me that I was on the right path with my own.