Question about when Gordon Willis says he underexposes half a stop

[Nick Thompson]

Hi all,

A lot has been said about Gordon Willis keeping things dark, of course, but I have a question about what he means in a couple interviews about The Godfather when he says he underexposed the whole negative half a stop. 

Excerpt from first article:

"Do you underexpose just to keep from getting detail in the shadows and then print up for the release print? Because I would think it would make your flesh tones go too dark.

No. As I said, I pick a printing light for the movie and I work to that light. I don’t print up or down.' - https://ascmag.com/articles/flashback-gordon-willis-asc-interview-at-afi-part-i

Excerpt from second article (film he was using was rated at 100):

'I’m exposing at ASA 250, and having the film pushed one stop in the lab, which means that, theoretically, I’m underexposing half a stop. The truth of the matter is that I like the look of that result. It appeals to me because the film becomes more translucent. You can see through the colors, rather than having them just sticking on the screen. Because the fog level is raised slightly in the pushing, the material tends to have a kind of foggy, not-quite-there look which, at times, is quite nice. 

Doesn’t this method leave very little latitude in printing?

Exactly. It means that once I’ve got the basic quality I want on the negative, the lab can do very little to jerk it around...They can’t print it up, for example — and that’s exactly why I expose it that way. Film material is designed to be printed in one range and one range only. I don’t like giving a lab the flexibility to print it up and down, because sooner or later someone, somewhere, in some little room will decide that it should be a little more this way or that. He’ll straighten it out for you. Everything will be just perfect — and it will be a disaster.' https://ascmag.com/articles/on-loacatiom-with-the-godfather-a-discussion-with-gordon-willis

I'm wondering what he really means by saying it's all underexposed by half a stop. Is it really all underexposed, or is it that something one would usually base exposure off of, like faces, are exposed half a stop lower than most DPs would expose them, and he's just explaining his aesthetic preference in a weird way? While Willis often liked things dark, he'll still have bright highlights in some of his shots, and as he says himself in various interviews, he uses the whole dynamic range of the film, so how is it really underexposed? I don't understand why he would be rating the film at 250, metering to expose it normally, and pushing it only to 200 rather than rating it at 200 but just metering for everything to be half a stop lower. If anyone can help me understand what he's saying, I would appreciate it. I shoot still film sometimes but I don't develop myself and know very little about the development process. Thank you very much.

 

Nick

[Stuart Brereton]

By underexposing by one and a third stops, he's pushing all his shadow detail down onto the toe of the characteristic curve, where it's only just above black. Then he pushes by one stop. A one stop push is really only one stop in the highlights, it's less in the shadows, so that may be what he means when he says it's still half a stop under even after the push, rather than one third. The end result is a slightly thin negative, with shadows that are almost impossible to print up. A thin neg has a different look to "normally" exposed one, which he evidently liked. Ultimately, it's about creating a look, and maintaining control over how the neg is printed.

[David Mullen ASC]

Technically 250 ASA is 1 1/3-stop underexposed for a 100 ASA film stock -- but it's hard to set a light meter rating in 1/2-stop increments.

He's underexposing by 1 1/3-stops by setting his meter to 250 ASA for a 100 ASA film stock (5254).

He's pushing development by 1-stop to compensate so the negative ends up only 1/3-stop under normal density.

He doesn't "print up" that 1/3-stop to compensate so that overall, the printed image is left a 1/3-stop dark.
 

As for why he doesn't set his meter to 200 ASA and then remember to underexpose everything by a 1/3-stop after taking a reading, by setting his meter to 250 ASA, he doesn't have to make that extra mental calculation. This is normal operating procedure, you input a value in your meter based on how you want to expose the stock (and taking any filter factors into account) so you don't have to keep remembering to adjust your meter reading for these things.

You could imagine a DP back then just rating the 100 ASA stock at 125 ASA with normal development for example. In fact, by 1980, Kodak had re-rated 5247 from 100 ASA to 125 ASA.

One thing that is hardly ever mentioned is that "Godfather I and II" were printed originally using Technicolor's dye transfer / imbibation system, which was known for its excellent blacks.  In the 1980s and 1990s when higher-speed film existed and Technicolor I.B. prints were a thing of the past, Willis was rating his stocks slower than normal, not faster, and printing down.

[Nick Thompson]

Thanks Stuart and David. 

So I take it from what you said Stuart that if your shadow details are all right above total black, there are details there, but printing them up won't look good/right (lots of grain, lifted blacks, the contrast will be off)? This makes sense to me from experience but I want to confirm that's what you meant?

Thanks for the clarification that it's actually a 1/3 stop underexposed, this makes sense. 

So in the later years, was part of the reason Willis rated his stocks slower than normal and printing down to achieve better blacks since the I.B. prints were no longer made? Assuming he's rating lower than normal and developing normal rather than pulling, why develop normal and print down rather than pulling? I assume that there could be a bunch of different reasons for this but I wonder what it might be in this case, just as an example. Would developing normal but printing down result in more contrast than pulling?

Nick

[Stuart Brereton]

36 minutes ago, Nick Thompson said:

So I take it from what you said Stuart that if your shadow details are all right above total black, there are details there, but printing them up won't look good/right (lots of grain, lifted blacks, the contrast will be off)? This makes sense to me from experience but I want to confirm that's what you meant?

Yes, that's what I meant. As Willis himself says "the lab can do very little to jerk it around...They can’t print it up, for example — and that’s exactly why I expose it that way."

Willis used to rigorously test his stock and printing lights before shooting a movie, so he knew exactly where to place his exposures, and as he says, once the exposure/printing routine was established, he didn't deviate from it.

[Phil Rhodes]

32 minutes ago, Stuart Brereton said:

Yes, that's what I meant. As Willis himself says "the lab can do very little to jerk it around...They can’t print it up, for example — and that’s exactly why I expose it that way."

I talk to quite a lot of DPs on projects of all levels these days, and the number of times this is said - often strictly off the record, on the quiet - is quite startling. People are very concerned about this.

[Stuart Brereton]

1 hour ago, Phil Rhodes said:

People are very concerned about this.

And they are right to be. In the digital world, it has become incredibly easy to alter imagery. Back in the "baked in" REC709 days, there was only so much that could be done to the footage, so if you exposed or lit it a certain way, you could be reasonably sure that it would remain fairly close to that. With the advent of RAW and Log Gamma, combined with the inexpensive ubiquity of Resolve, it's become the wild west in post production. Editors can do a color pass, slapping on filters of dubious provenance. Colorists can completely change the look and feel of material. Producers have seen just how much flexibility there is, and now often actively discourage DPs from attending the grade. They like having the control taken away from the DP. A few years back, I had a producer specifically tell me to light flat, expose in the middle, and they would "create" the look of the film in post (I ignored him). I've even had directors who I trust lose their nerve in the grading suite and brighten up material that we had agreed should be dark, just to play it safe. With underexposure, my approach has always been to do most of the work in camera, and then finish the last 10-15% in post, but these days the temptation is to go the whole way in camera, and damn the consequences.

[David Mullen ASC]

2 hours ago, Nick Thompson said:

Thanks Stuart and David. 

So I take it from what you said Stuart that if your shadow details are all right above total black, there are details there, but printing them up won't look good/right (lots of grain, lifted blacks, the contrast will be off)? This makes sense to me from experience but I want to confirm that's what you meant?

Thanks for the clarification that it's actually a 1/3 stop underexposed, this makes sense. 

So in the later years, was part of the reason Willis rated his stocks slower than normal and printing down to achieve better blacks since the I.B. prints were no longer made? Assuming he's rating lower than normal and developing normal rather than pulling, why develop normal and print down rather than pulling? I assume that there could be a bunch of different reasons for this but I wonder what it might be in this case, just as an example. Would developing normal but printing down result in more contrast than pulling?

Nick

Black level in a print is a function of the printer light values needed. If you sent in a totally unexposed roll of film to the lab, developed it (and ended up with close to a clear strip of film other than the base fog density) and then asked the lab to print it in the 10's, 20's, 30's, 40's (the scale is 0-to-50) with in theory 25-25-25 being the numbers for a color negative of normal density (rarely in reality though), you find that the black level of the print increases as the printer light values increase (i.e. printing "down").  So a "thin" negative printed "up" has less deep blacks than a "dense" negative printed "down".  But also keep in mind that "The Godfather" was made in the days of 100 ASA film, so it wasn't easy to consistently light a movie to the high levels for a dense negative and have it look naturalistic, except for day work (Conrad Hall often overexposed his negative for day exteriors, like in "Butch Cassidy and the Sundance Kid").

Mainly Willis overexposed and printed down in the 1990's because high speed film got grainy if it didn't get enough exposure but also to get better blacks. He never really expressed an opinion on the black level of I.B. prints.

Overexposing reduces the appearance of graininess by exposing the small (slower) grains in between the large (faster) grains -- the large grains are always there as part of the speed of the stock and collect light the easiest. This is another reason why high-speed stocks respond better (get less grainy) to overexposure, they are a mix of slow and fast layers of emulsion.  Slow film doesn't really need two layers of speed for each color, they only have the small grains.

So the reduction in graininess comes from extra exposure. Pull-processing to compensate reduces density build-up so an overexposed negative once pulled has close to normal density (if you overexposed by 1-stop, for example, and pulled by 1-stop) but then you don't have the printing benefits of a denser negative, you can't "print down" to normal brightness but get deeper blacks.  Generally pull-processing also softens contrast as well.  

[Satsuki Murashige]

22 minutes ago, Stuart Brereton said:

With the advent of RAW and Log Gamma, combined with the inexpensive ubiquity of Resolve, it's become the wild west in post production. Editors can do a color pass, slapping on filters of dubious provenance. Colorists can completely change the look and feel of material. Producers have seen just how much flexibility there is, and now often actively discourage DPs from attending the grade. They like having the control taken away from the DP.

I’ve found that sometimes this can happen in the edit, and it’s not really anyone’s fault (well, not the director’s, anyway) - a scene which was supposed to take place in cold dreary London (thanks Phil!) now has been repurposed to warm sunny California, so the shot which was exposed and white-balanced for one location now has to be graded to match the other location. 

I had a director with whom I have a good relationship ask me to do those color shifts with a LUT change instead of in-camera white balance on a project last year, and I’m glad I did. Of course, I did get to supervise the grading session in the end, but these things do happen.

But in general, I’m in agreement with you. If I’m going to put my name on something, I’d rather any photographic choices (good or bad) to be mine rather than someone else’s. 

[Takoda Porembski]

So say I meter 500asa film at 800 then have it pulled in lab half a stop (roughly speaking), would this increase density in blacks and get rid of some grain? Given I get my film developed with the cheaper "best-lighting" option (I cant afford fotokem), this would then take flexibility away from the person scanning forcing the look closer to what the shooter (me) intended? ?

[David Mullen ASC]

No, making a print of a denser negative at higher printer lights will lead to deeper blacks in the print.

Overexposing and pull-processing reduces contrast (from the pulling) and smooths out the grain (from the overexposure filling in the gaps between the large grains with smaller slower grains).  If you pull-process by the same amount more or less that you overexposed, the net density of the negative will be normal.

Black level in a video transfer or color-correction from a log scan is controlled digitally, you can set the blacks to "0" basically, the only issue for you is whether this looks natural in terms of shadow detail.

To repeat, black level is a video or digital value. Some video transfers set black a little above "0", like at 10 IRE, to show you all the shadow detail on the film, expecting you to later color-correct it back down. A log gamma scan will also have lifted blacks, which is normal -- they should go to black once you correct the image to Rec.709 display gamma (or add a Rec.709 LUT.)

If you're just looking to get less grain, then overexpose a little and develop normally. Even better, use slower-speed film.
If you overexpose to reduce grain, shoot a grey card or grey scale at the head of the roll with the same amount of overexposure so that the colorist doing the video transfer sees a neutral reference and should time the roll so that the overexposed grey card looks normally exposed and thus the footage that follows will also look normally exposed.

[Satsuki Murashige]

19 hours ago, Takoda Porembski said:

So say I meter 500asa film at 800 then have it pulled in lab half a stop (roughly speaking), would this increase density in blacks and get rid of some grain? Given I get my film developed with the cheaper "best-lighting" option (I cant afford fotokem), this would then take flexibility away from the person scanning forcing the look closer to what the shooter (me) intended? ?

If you did this, your negative would be 1.3 stops underexposed so the image would less dense and have more grain. For a denser negative with less apparent grain, you would want to set the meter to a lower ASA value and avoid pulling development.

[Nick Thompson]

Thank you all for the all the helpful info.

I definitely think that, just from experience as a viewer, shooting in LOG has influenced a trend in the look of video to be very flat, low-contrast, often with lifted blacks. It seems that even though people shoot in LOG knowing that they have to then correct it in post, they get used to the look and softer, flatter looks become more and more common. I can't tell you the amount of low budget music videos or corporate work I've seen where it almost seems like they shot it in LOG and then forgot to correct it, or barely corrected it at all. 

Obviously any sort of look can work sometimes, and going for something flat, low-contrast is sometimes the right thing to do. But it seems very ubiquitous right now. The colorist for my movie sent me back initial stills with slightly lifted blacks because he said that almost every client he works with asks for that. As I said, sometimes this is a cool look, but not to this extent, in my opinion. 

It'll probably shift in the next few years, though. About ten years ago, when film look presets started to become widely available for consumers or low-budget work, there seemed to be more of a trend for overly crushed blacks and high contrast/saturation in an effort to get that 'film look.'

[David Mullen ASC]

Even before digital and even before log scans, dailies of film material was often transferred with the blacks at 10% and the whites at 90% as standard operating procedure so that the client saw everything usable in the shadows -- it used to drive me nuts on features that I couldn't get dailies with the blacks at 0% but the post house always said "we've never had a studio complain when dailies were too light but almost always got complaints when dailies were too dark."