ON COMMON MISCONCEPTIONS ABOUT LARGE FORMAT OPTICS
Nov 6, 2019
(click here to skip the intro and jump to the meat)
Hi, -------
My name is Steve Yedlin, and I'm a cinematographer.
I'm writing to you about your recent article on large formats. I greatly appreciate your enthusiasm for the craft of cinema, and [some of the films you mention in your piece] are beautifully shot and amongst my personal faves for cinematography in the last few years.
However, I think you've missed an opportunity, or perhaps there can be a different opportunity in the future, to unwind the differences between the hard science of how optics actually work and some common misconceptions that are (despite their status as debunked superstition) actually driving some of these decisions in the marketplace today.
Although it's true that there exist these prevalent misconceptions that are communally reinforced within the industry (and not corrected by vendors who benefit from the misunderstanding) and that they deserve being reported on for their wide dissemination and their driving force in the market place, I think it's also worth reporting on their objective falsity: their inconsistency with the physics of optics that are well understood and documented by hard science
I think it’s important to include the hard science too, because every report that repeats the fallacies without debunking them perpetuates them.
I explain some technical details and give example images here:
http://yedlin.net/lens_blur.html
and here:
http://www.yedlin.net/190127.html
[And it's also explained pretty well in this thread: https://twitter.com/edmoore/status/1174633351560478730?s=20]
But here's a Cliff's Notes version of the aspects as they apply to what's actually discussed in your article:
First, I agree with your own assertion that [any increased resolving power of large format cameras is not what's driving their growth in the market place]. So, we can take it as stipulated that this community belief that there is a large-format "look" is not related to resolution, but is based on a belief in some optical character inherent to a sensor's size -- that's why proponents of the belief refer to a "large format look" rather than "resolution." But the belief in this supposed optical character or "look" can be demonstrated to be superstition or misunderstanding that doesn't stand up to scrutiny.
The believed-in optical properties often proffered without hard (non-anecdotal) evidence always fall into one or more of 5 categories which can be demonstrated by hard science not to yield an image that has properties unique to sensor size.
1. Depth of Field (or, more accurately, 'blur circles').
Blur circles, which define the amount of blur that out-of-focus objects have, are not subjective -- they are mathematically well defined. And they are determined by the RATIO of the sensor size to the f/stop. Since the audience can see only the final blur circles and neither the f/stop nor the sensor size, they can't see in the final image if blur circles are increased by a larger format size or by a larger aperture. Those two things are interchangeable in the final image, so this is not a "look" that's discernable in the final image. (It may be easier for the the filmmaker to achieve a certain size blur circle in one format or another but the audience also can't see how easy or difficult it is -- they only see the final image, so again, it's not a "look.”).
Even in an unusual edge case where a filmmaker has a specific model of large format lens at an extremely wide aperture and the only lens model available to him/her for a smaller format camera can't open wide enough to get the same size blur circles: anyone who only looks at the final image and wasn't there when the image was captured can't SEE that the aperture was at its endpoint, so that's not a "look" -- it's just something that the filmmakers are aware of during production (that the aperture and not the sensor was the limting factor in this particular case). The opposite could have been true if different gear had been on hand: it might have been the case that only the smaller format camera could get blur circles that large, since real world lenses tend to have larger max apertures if they also have smaller coverage areas. Either scenario could have been the case and anyone who only views the final image can't tell which it was, since both yield the same blur circles.
2. Angle of View.
Just as with blur circles, AoV is determined solely by a RATIO: the ratio of focal length to sensor size. Again, since the audience only sees the final angle of view and not the focal length or sensor size, there is no optical difference in the final image between widening the AoV with a larger sensor and doing so with a smaller focal length. (Or narrowing it by doing the reverse.)
3. Perspective.
Perspective includes vanishing lines, relative size of objects, how foreground objects obscure background object, the angle at which an object is viewed, etc. All of this is determine by one thing only: the position of the entrance pupil (the precise position in space of the point-of-view). It's been well understood since nearly the beginning of optical science that how objects in view stack up with one another is determined by "ray tracing": drawing lines from the points in the scene to converge at the point of view -- this is irrespective not only of a camera's sensor size, but even of the presence of a camera at all. Every point in space has its own perspective on its surroundings -- the laws of physics don't care which type of camera you put at that point; the manner in which light rays from the surroundings converge at a point in space is not altered by which camera happens to be at that point.
4. Uniformity (or lack of uniformity) of magnification across the field.
All "distortion" or lack thereof is based on how magnification is distributed across the field of view. This is determined only by the engineering of the lens and not by the format size. For example, just within lenses designed for regular "super-35mm" target size (which is 24mm across the frame area), I could show you four different lens models that deliver the same angle of view but one is (nearly) rectilinear, one is (nearly) equisolid, one is fisheye, and one is rectilinear near the center and more fish eye near the edges. These are 4 generalized descriptions of magnification across the field, which, again is determined by the engineering of the lens model and is not enforced by the sensor size (in fact each those same 4 lenses of various character can be put on cameras with different sensor sizes).
5. Compression.
Compression isn't actually an optical attribute in itself, but it's a word that's often thrown around while describing things in a confused way. Compression (in the sense that it's meant by filmmakers when discussing optical properties) is a combination of camera placement and angle-of-view -- it's just a combo of two other attributes. It means that your perspective is distant while your AoV is narrow. So, a shot that's 4 meters away from an actor on focal length x is much more compressed than (but the same image size as) a shot that's 1 meter away from the same actor on focal length that's one quarter of x. This is true no matter what the sensor size is. But, in trying to describe "the big format look," people often make a word-salad of focal length, AoV, and compression, like "you can get closer on a longer lens with the same AoV but it's more compressed." But that's just a self-contradictory jumble of words that doesn't really have any meaning. AoV and perspective are each properties that are not unique to one format size as opposed to another and are independent of one another (putting a certain lens on a camera doesn't force you to put that camera in a certain position in space -- a filmmaker is free to choose AoV and camera placement separately). And compression is nothing but how you combine AoV and camera postion when setting up a shot. If you use the same perspective (camera postion) and same AoV on two different sensor sizes, you'll have the same compression. Compression is not an attribute of the sensor size, but a description of how a filmmaker sets up a shot (like: do we back up with a narrower AoV or go closer with a wider one).
I appreciate the article's timeliness and it's accuracy of reporting on the increased use of larger sensors as an actual market trend, but I hope there are opportunities in future reporting for more rigor in differentiating the fallacies that are being perpetuated from the facts of hard science, because it's a very real phenomenon that demonstrably false beliefs are non-trivially impacting decision making and market trends. Without that clarification, the reporting itself becomes both part of the reinforcing feedback loop that propagates the misunderstanding and an unwitting advertising mouthpieces for the vendors or brands who profit from disinformation (whether or not they are themselves propagating the fallacies).
Thanks for your time.
Best.
-Steve Yedlin, ASC