Robert Richardson lighting style: how to?

[Alexander Ibrahim]

The first picture is exactly what i meant by it.

so your saying its achieved using a powerful source behind the actor? Because to me it appears a large blast of light is coming from above.

I see this as a point in a continuum of the same technique as in the second still. I threw that second picture in merely as a clue, but not a direct map to creating the exact result you are after.

The rim light is created by a light on the far side of the subject and above the subject.

(My convention is that elevation 90º is directly above the subject. -90º would be directly below the subject, and 0º would be on the plane with the camera and subject.. In all cases the source is aimed at the subject. I wonder if there is a more universal convention for these sorts of discussions.)

In the Inglorious Basterds still I posted the rim light is coming from behind and above the actors at roughly 50º.

In the JFK shots David posted its coming directly down. I'd probably have set it very slightly tilted, perhaps 85-89º ... but its the same idea.

What I don't know is if I'd be bold enough to let that much of the image blow out. I think it would be very dangerous in digital, you'd have to control it much more. The Inglorious Basterds lighting seems much more like what I might aim for with digital imaging.

 

[Alexander Ibrahim]

The main thing is that you want a very hot center that you basically don't even bother metering because the idea is to let things burn out in the center of the beam -- so you meter the fringes of the beam or the bounce back into the shadows. And a little diffusion on the lens will extend the halation further. And making the top light come in from more of a backlit angle helps keep the burn-out effect from being too obnoxious because it's mostly the hair and shoulders that get burned up, not the face, which is more lit by the bounce of the backlight up into the faces.

First off ... yet another awesome contribution David! Thank You!

You know I've used this metering technique before, but somehow reading it makes it more concrete in my mind. It also raises some questions.

I usually meter a scene like this with incident to set my light's positions and levels, but I use reflective metering in the faces to get a sense of my exposure.

Do you have any comments on specific use of the meter here? How do you do it?

 

[David Mullen ASC]

If a face is mostly backlit, I use an incident meter to read the face and decide how dark I want it to be, since it's being lit by the ambient bounce back into the shadows. Depending on the mood, I may decide it should be one or two stops underexposed, or 1.5-stops under. With film, there is such tremendous overexposure latitude I generally don't meter the backlight unless I have a good reason to.

Now if the light is directly overhead and thus top-lighting the face, then I meter that light and decide how much I want it to be overexposed. But in the case of that Richardson style where you use very narrow spot globes and the actors are moving around, I base the exposure more on the fringes of the top light so that when the actor momentarily passes under the hottest spot, they go nuclear. There isn't much point in metering that, you're just going to be shocked at how many stops over it is anyway. Robert Primes once said "Your meter makes you afraid -- your eyes make you brave" or something to that effect.

I don't really use a meter to set exposure with a digital camera so in this case, I'd just play around with the stop until I got the balance of shadow detail versus clipping I wanted.

 

[David Mullen ASC]

What I don't know is if I'd be bold enough to let that much of the image blow out. I think it would be very dangerous in digital, you'd have to control it much more. The Inglorious Basterds lighting seems much more like what I might aim for with digital imaging.

Keep in mind that in "JFK" many of the more extreme shots with hot toplight were flashbacks that were on-screen for a short time, so the point was to make them "pop" -- sometimes with flash-frames connecting the shots. The other thing is that the actors move in and out of the hottest spot but I picked the most extreme moment to grab a frame.

I don't think one should generalize too much about what sort of lighting to do with digital versus film -- good lighting is good lighting and you can find examples of almost any style done in both digital and film movies. And with modern digital cameras approaching the same DR as film, there really isn't an excuse anymore to avoid the same lighting effects that you like in film.

In fact, back in 2000 when I shot one of the first 24P F900 movies, "Jackpot", a Deluxe lab person came into the screening room and saw the film-out and said "you lit it just like film!" to which I replied "I don't know how to light any other way." Bits of overexposure that burn out parts of the frame can be beautiful and shouldn't be avoided just because you are shooting digitally, if that's what you want to happen. "Bad" overexposure in digital movies generally looks like a mistake, but it can be done in a way where it looks intentional and attractive. But it's a lot easier today than it was ten years ago.

 

[Alexander Ibrahim]

If a face is mostly backlit, I use an incident meter to read the face and decide how dark I want it to be, since it's being lit by the ambient bounce back into the shadows.

An incident meter? Why not a reflective meter? I thought the "right" way to do things was to use the incident meter to measure the light falling in a space, and to use the reflective meter to measure a particular object, say a face.

Depending on the mood, I may decide it should be one or two stops underexposed, or 1.5-stops under. With film, there is such tremendous overexposure latitude I generally don't meter the backlight unless I have a good reason to.

I would almost certainly set my initial look by eye, but then I would have metered the backlight to keep it consistent from set up to set up. I might let the amount of the frame that gets the over exposure vary shot to shot, but I try to keep the overexposure levels consistent.

I know, consistency is the hobgoblin of little minds, but isn't that a valid consideration for the production? Am I missing something else that should be an over riding concern?

Now if the light is directly overhead and thus top-lighting the face, then I meter that light and decide how much I want it to be overexposed. But in the case of that Richardson style where you use very narrow spot globes and the actors are moving around, I base the exposure more on the fringes of the top light so that when the actor momentarily passes under the hottest spot, they go nuclear. There isn't much point in metering that, you're just going to be shocked at how many stops over it is anyway. Robert Primes once said "Your meter makes you afraid -- your eyes make you brave" or something to that effect.

Its a funny thing. I heard that quote before, but I had a backwards reaction to it. I've shot film, but most of my work has been in video. So, I've been pretty conservative about controlling highlights on set. I find that consistently using the meter makes me more aggressive. Especially after I have experience with the camera system and have learned its limits.

I don't really use a meter to set exposure with a digital camera so in this case, I'd just play around with the stop until I got the balance of shadow detail versus clipping I wanted.

Why not?

I use the meter mostly for shot to shot control, usually working from a plan.

This way, I sort of know what I am after, and if I have to come back to a set up, say in pick ups, then I can match it very quickly and accurately.

I know that is a technical way of working, but it helps me keep very consistent within a scene (as appropriate) and I think it benefits my productions by lightening the color correction workload. (and hence expense.)

 

[Alexander Ibrahim]

I don't think one should generalize too much about what sort of lighting to do with digital versus film -- good lighting is good lighting and you can find examples of almost any style done in both digital and film movies. And with modern digital cameras approaching the same DR as film, there really isn't an excuse anymore to avoid the same lighting effects that you like in film.

In fact, back in 2000 when I shot one of the first 24P F900 movies, "Jackpot", a Deluxe lab person came into the screening room and saw the film-out and said "you lit it just like film!" to which I replied "I don't know how to light any other way." Bits of overexposure that burn out parts of the frame can be beautiful and shouldn't be avoided just because you are shooting digitally, if that's what you want to happen. "Bad" overexposure in digital movies generally looks like a mistake, but it can be done in a way where it looks intentional and attractive. But it's a lot easier today than it was ten years ago.

I agree that with pretty much any imaging format your lighting is much more important, and that you use the same techniques regardless of the medium.

Where I think I disagree, and I may not understand, is in highlights. I think that the clipping of video systems is harsh and rather ugly ... and that its gotten worse in digital media where there is simply no information.

I think that its better looking to control the peak highlight levels on set and push the overexposed look in post. With Red, I can apply looks in camera and get what I am after in my monitors while retaining data in the RAW files, but with most camera systems its better to keep exposure set so that data gets recorded, even when what you want is to blow it out.

Have I made an error in my reasoning? Are you seeing highlight roll-offs that are more appealing in digital systems now?

 

[David Mullen ASC]

If the face is lit by the bounce from the backlight up into the face and you are metering the face, then you are going to be consistent generally even if you don't meter the backlight. Plus, as Caleb Deschanel once pointed out, once a backlight burns-out, it looks the same to some degree within a range of overexposure, the only thing that would changed is the degree of halation (which to some degree can be seen in the viewfinder or the monitor anyway) and the amount of ambience back into the shadows (which you are metering).

And generally, unless you plan on coming back and recreating the lighting set-up, it's not hard to be consistent even if you don't meter everything because within all the coverage looking at the same direction, you are going to be working with much of the same lighting used in the master, you aren't starting from scratch on every set-up.

But the main truth is that simply I prefer to get away from metering, I find it a very technical way of working and the whole point is to get away from that over time as you gain experience.

As for using a spot meter versus an incident meter, there's no right or wrong answer unless you need to meter objects that create their own light source (like a sunset sky or a TV screen), it's just a matter of what you are more comfortable with, and I like using an incident meter -- I'd rather know the amount of light falling onto an object rather than the reflectance of an object. You mentioned consistency -- I find it easier to be consistent if I know the levels of lighting falling in a room rather than deal with the individual reflectance of various objects in the room. In other words, with an incident meter, I know that the light falling from above on the desk is f/4, let's say, regardless of what objects are placed on the desk or who sits at it, whereas with a spot meter, I can get all sorts of confusing readings just by what is placed on the desk, and if I end up instead looking for an 18% grey object to spot meter, then the readings should be close to what an incident meter gives you.

But all meter readings have to be interpreted anyway, which is why you should use what you are most comfortable with, whatever gives you the results you like. I never got into spot meter readings, so I only pull out that meter when I'm forced to, like to read how bright a TV screen is when doing an insert, or a sunset sky.

Just because I don't take a lot of meter readings, don't assume that my exposures are all over the map -- in fact, one timer commented that my entire feature printed very consistently. It matters more what you decide to meter and how you interpret it than the volume of readings you take. And you have to train your eye anyway so metering starts to become more of a formality over time.

As for digital, I use a meter to prelight a set (so I don't have to drag out the movie camera), to light a series of keys that have to be at the same level (like for a long Steadicam move), and to scout locations. But I don't use a meter to set the final exposure, I use the monitor and various in-camera tools like histograms, zebras, waveforms, etc. to double-check my clip point, etc. But in a sense those tools are meters - in fact, a digital camera is a form of meter.

I do try to make sure that my keys are within a consistent range so I'm not shooting one person at f/2.8 but the reverse angle at f/5.6 (unless I have a reason for that.) But when I'm going for a nuclear backlight, I don't see the point of sticking my meter in the beam and seeing f/64 on the meter, or whatever -- instead I read what that backlight does to the ambience, the shadow side of the face, etc.

Sure, if I were shooting efx plates and greeenscreen stuff, I'd have to do a lot more metering, partly just for documentation. But for lighting a little drama, you want to light by eye, get a few quick readings, tweak, and shoot, and move on -- it's not a science project. There's nothing fun about taking meter readings.

 

[David Mullen ASC]

I agree that with pretty much any imaging format your lighting is much more important, and that you use the same techniques regardless of the medium.

Where I think I disagree, and I may not understand, is in highlights. I think that the clipping of video systems is harsh and rather ugly ... and that its gotten worse in digital media where there is simply no information.

I think that its better looking to control the peak highlight levels on set and push the overexposed look in post. With Red, I can apply looks in camera and get what I am after in my monitors while retaining data in the RAW files, but with most camera systems its better to keep exposure set so that data gets recorded, even when what you want is to blow it out.

Have I made an error in my reasoning? Are you seeing highlight roll-offs that are more appealing in digital systems now?

Your reasoning is sound, but it matters more for Rec.709 broadcast cameras with only 9-stops of DR -- when you start talking about 12, 13, 14-stops of DR, the way overexposure starts to look is more like how it looks on film, so a similar attitude can be applied within reason.

Yes, you want to go into post with more data, more information, than you ultimately need, but there are reasonable exceptions, you don't need to always shoot flat and add contrast later because sometimes that doesn't look right either. And if you want something clipped, then there's no real reason not to clip it if all you are going to do is immediately clip it when you get into post. If someone points a flashlight into the camera lens and it flares the lens, you aren't going to expose for the filament in the bulb of the flashlight just to hold detail that you then are going to discard anyway. Nor if you shoot in a white limbo cyc set do you want to see every scuff mark and brushstroke in the white cyc -- it's meant to be a featureless white field. Same goes for a halo around a head that is heavily backlit -- if you exposed for the backlight to avoid any clipping, the net result is less shadow detail -- and in that shot, maybe 98% of the shot is in the shadow, so are you really going to base the exposure on something that only takes up 2% of the image? So there are reasonable limits to how much you have to avoid clipping of any kind.

Certainly with the Epic and the Alexa, objects roll off into overexposure a lot more gracefully. But even with cameras like the Genesis or Red One where clipping happens a bit sooner, you just have to judge the individual image and decide how obnoxious any clipping might be -- a tiny edge or spot, etc. it doesn't matter as much as a whole shirt or wall, for example. If a car drives by in the background in full sun and the sun glances off of the windshield momentarily, well, the clipping isn't going to be film-like but on the other hand, it's going to be very brief and it's not the subject of the shot. It starts to fall into that "well, if you wanted it to absolutely behave like film, you should have shot on film" sort of moment -- to some degree, once I commit to using a particular camera, I have to embrace its natural tendencies within reason. If I shoot Fuji, I don't demand that it behave exactly like Kodak, if I shoot with a Cooke lens, I don't demand that it behaves like a Zeiss, you know what I mean? Otherwise the whole production becomes an exercise in frustration.

Interesting cinematography, such as Richardson's, often pushes elements to the extremes, that's what makes it interesting. So you want to caution against overly correcting and controlling the image. Caleb Deschanel once said that there should be one element in the frame that looks out of control, otherwise the image has no life, no reality to it. So if you work at the extreme, occasionally you fall over the edge, but that's better than playing it safe all the time.

 

[Alexander Ibrahim]

If the face is lit by the bounce from the backlight up into the face and you are metering the face, then you are going to be consistent generally even if you don't meter the backlight.

Duh ... why didn't I notice that! Thanks David.

And generally, unless you plan on coming back and recreating the lighting set-up, it's not hard to be consistent even if you don't meter everything because within all the coverage looking at the same direction, you are going to be working with much of the same lighting used in the master, you aren't starting from scratch on every set-up.

Sadly, I often have to do exactly that .. and VFX which you mentioned later.

As for using a spot meter versus an incident meter, there's no right or wrong answer unless you need to meter objects that create their own light source (like a sunset sky or a TV screen), it's just a matter of what you are more comfortable with, and I like using an incident meter -- I'd rather know the amount of light falling onto an object rather than the reflectance of an object. You mentioned consistency -- I find it easier to be consistent if I know the levels of lighting falling in a room rather than deal with the individual reflectance of various objects in the room. In other words, with an incident meter, I know that the light falling from above on the desk is f/4, let's say, regardless of what objects are placed on the desk or who sits at it, whereas with an incident meter, I can get all sorts of confusing readings just by what is placed on the desk, and if I end up instead looking for an 18% grey object to spot meter, then the readings should be close to what an incident meter gives you.

Ah, I see. I typically use the spot meter on the same object. Say the same person's face when I move shot to shot or set up to set up. This helps me stay consistent.

But all meter readings have to be interpreted anyway, which is why you should use what you are most comfortable with, whatever gives you the results you like. I never got into spot meter readings, so I only pull out that meter when I'm forced to, like to read how bright a TV screen is when doing an insert, or a sunset sky.

Oh ... it may partly be that I have a unit that has both spot and reflective meters built in. (Sekonics L758Cine) There is nothing for me to pull out.

Just because I don't take a lot of meter readings, don't assume that my exposures are all over the map -- in fact, one timer commented that my entire feature printed very consistently.

I've been an admirer of your work since Jackpot David, so I never assumed you were inconsistent. In fact you wrote elsewhere about the timer telling you your feature printed consistently. Its from that comment I learned that consistency was important!! (Then I only had to figure out for myself why ... )

Sure, if I were shooting efx plates and greeenscreen stuff, I'd have to do a lot more metering, partly just for documentation. But for lighting a little drama, you want to light by eye, get a few quick readings, tweak, and shoot, and move on -- it's not a science project. There's nothing fun about taking meter readings.

Part of the problem with me is that I'm trained as a scientist ... I actually like taking readings and such.

Once when I was working on Star Trek one of the actors was watching me run around and meter this and that. They said I ought to play Spock the way I was running around with my "tricorder."

That just about made my week. I am such a geek.

 

[Alexander Ibrahim]

First off thanks again ... this is helpful!

Your reasoning is sound, but it matters more for Rec.709 broadcast cameras with only 9-stops of DR -- when you start talking about 12, 13, 14-stops of DR, the way overexposure starts to look is more like how it looks on film, so a similar attitude can be applied within reason.

Is it just that there is latitude to handle more light ... or that a softer rolloff to clipping is an emergent property of systems with higher DR?

If someone points a flashlight into the camera lens and it flares the lens, you aren't going to expose for the filament in the bulb of the flashlight just to hold detail that you then are going to discard anyway. Nor if you shoot in a white limbo cyc set do you want to see every scuff mark and brushstroke in the white cyc -- it's meant to be a featureless white field.

I may have overstated my position. Of course I wouldn't try to preserve all highlights in those situations!

Interesting cinematography, such as Richardson's, often pushes elements to the extremes, that's what makes it interesting. So you want to caution against overly correcting and controlling the image. Caleb Deschanel once said that there should be one element in the frame that looks out of control, otherwise the image has no life, no reality to it. So if you work at the extreme, occasionally you fall over the edge, but that's better than playing it safe all the time.

Well ... you've written that before. I'm trying to embrace it. Honest.

 

[Scott Crawley]

When spot metering flesh tones the tendency is to homogenize and the subtle differences between faces that add richness and character to the image can be lost. That is why I've learned to avoid it.

My film work has all been in still photography. All of my moving pictures have been video acquisition with very little vfx. For the most part I spot metered when I had an object that I really wanted to place in particular zone, and I wasn't printing it myself.

 

[paulherrin]

other than the situations where you are forced to allow clipping (hopefully these will continue to decrease), i tend to agree with alex that i generally do not like to clip data to white (other than background light sources and flashlights and all that stuff). i like to be able to control how the highlights roll off. but that's just me. and there's obviously situations where it's not going to make much of a difference and situations where you may actually want to let the camera clip something, or as you mention, david, you don't want to start losing detail you may want in the shadows just to be silly about highlights. it's always a give-take, push-pull... but it's already becoming less of an issue.

i think the point you're trying to make david, is that it is totally acceptable to let your highlights clip. you're like me, and don't like all these "rules" that seem to have been developed. let's do what we will and make beautiful images in whatever way we think best.

on the other end, there are other things to keep in mind, like the good ol' boys doing post-production... don't slouch off and give me a headache please.

for the record i don't use a light meter unless someone puts it in my hand and expects me to look busy with it. not discounting them, i just use the "ole eyeballin" method.

 

[Alexander Ibrahim]

When spot metering flesh tones the tendency is to homogenize and the subtle differences between faces that add richness and character to the image can be lost. That is why I've learned to avoid it.

I think that would be true if you were spot metering everyone's face with the same criteria. If I tried to meter everyone's face and light them to f/2.8 exactly, that would eliminate some of the variation in people's faces.

What I am suggesting is that you meter a particular individual's face, and keep that consistent shot to shot. So my leading lady may be f/2.8.1 (My meter is set to read tenths of a stop), and my leading man may be f/2.8.5 under the same light lighting.

I aim to keep those values consistent shot to shot in the scene.

Now, understand I'm lighting with the incident meter. If 32 candles are falling in the area I want 32 candles, then I should be good right ...

I then use the spot meter from the camera position to make sure things turn out as intended, when I think I'm done.

I agree with David's statement that a video camera is a fancy light meter ... but for me at least you can forget little details of how a shot looked as the day goes on, while its easy to remember a string of numbers. If I do it by eye, I may let the look drift a bit during the day, but by meter I stay precisely consistent.

The scopes in cameras aren't typically detailed enough for my taste, but if I have a display with scopes, or better yet standalone scopes that's another matter. With those at hand I sometimes eschew the spot meter completely.

 

[David Mullen ASC]

Everyone develops their own system of judging exposure, so whatever works for them is fine by me. I remember years ago reading an AC article where a DP said that he rated 500T stock normally, i.e. at 500 ASA, and always printed in the 40's... which just doesn't make sense. So clearly his metering method, whatever it was, favored exposing for the shadows and printing down because 500T rated at 500 ASA should be printing in the high 20's, low 30's in general. But who am I to tell this Oscar-winner that his metering method was "incorrect" -- clearly it was working for him.

Same goes for spot meters versus incident meters -- I know that Steven Poster uses a spot meter almost exclusively, whereas someone like Conrad Hall used an old Sekonic, and Douglas Slocombe didn't use a meter at all, amazingly. Whatever works.

Now if you are doing still photography and employing the Zone System, you need to spot meter because that system was all about assigning reflectance values to certain Zones.

 

[Scott Crawley]

I think that would be true if you were spot metering everyone's face with the same criteria. If I tried to meter everyone's face and light them to f/2.8 exactly, that would eliminate some of the variation in people's faces.

What I am suggesting is that you meter a particular individual's face, and keep that consistent shot to shot. So my leading lady may be f/2.8.1 (My meter is set to read tenths of a stop), and my leading man may be f/2.8.5 under the same light lighting.

I aim to keep those values consistent shot to shot in the scene.

Most people I've known who do this standardize within ethnic groups, not with every face. I can't fault your method, though your AD might ;-)... It sounds a little time consuming and tedious to me when it is easy enough to fine tune that balance in DI. For speed sake I tend to set general levels by histogram and use my meter mostly for contrast ratios, so I meter almost exclusively by incidence. It's a matter of personal taste, I guess.

 

[Alexander Ibrahim]

Most people I've known who do this standardize within ethnic groups, not with every face. I can't fault your method, though your AD might ;-)... It sounds a little time consuming and tedious to me when it is easy enough to fine tune that balance in DI. For speed sake I tend to set general levels by histogram and use my meter mostly for contrast ratios, so I meter almost exclusively by incidence. It's a matter of personal taste, I guess.

I know you are poking fun at me a bit, but this is a really fast system.

Once I've memorized the values I set in the master for each subject of interest (which takes very little time) It takes no more than 15 seconds to meter every person in a scene.

Now the key here is subject of interest. If I have a scene where a woman, her best friend and six other characters are shopping for dresses, and only the woman best friend and sales person speak ... I only have three values to memorize, and frankly I only care about two of them.

It gets cumbersome when I have a scene in the conference room on the Enterprise, and I have the gang of seven plus the alien of the week in their makeup, all of them speak in singles, two shots, three shots and the master. Then the scene takes some more analysis. Who is the scene really about ... that's usually just 3-4 characters. Add that you always make Uhura and Kirk look good, and alien make up may need special attention to look its best.

(Of course its Star Trek, so sometimes the alien of the week just has a funny hairdo or outfit but is otherwise human ...)

Even that's not so complicated that you can't keep it in your head and work fast. I may have to remember six individual values.

Compared to the "ethnicity system" you mention which would require me to track 5 values. (vulcan, alien of the week, human caucasian, human asian, human african.) Its not that much more mental effort, and it lets me give more finely grained individualized attention as needed on a scene by scene basis.

Now that I've written it down though ... I have to say its a lot harder to explain than it is to do on set!

 

[Scott Crawley]

I know you are poking fun at me a bit, but this is a really fast system.

...

Compared to the "ethnicity system" you mention which would require me to track 5 values. (vulcan, alien of the week, human caucasian, human asian, human african.) Its not that much more mental effort, and it lets me give more finely grained individualized attention as needed on a scene by scene basis.

Haha! That's funny. Kudos.

 

[Alexander Ibrahim]

Haha! That's funny. Kudos.

Yes, well my life often takes a turn for the ridiculous, my work even more so. Frankly, I wouldn't have it any other way.

In any case, I think I'll try the "ethnicity system" on a more normal production.

 

[Scott Crawley]

Yes, well my life often takes a turn for the ridiculous, my work even more so.

Well I will take your word on that. I'm just contemplating the process of determining what zone this week's alien should be exposed in, and it gives me a chuckle.

Sorry for running off topic.

 

[Bill Anderson]

When spot metering flesh tones the tendency is to homogenize and the subtle differences between faces that add richness and character to the image can be lost. That is why I've learned to avoid it.


This makes no sense whatsoever. None. Unless of course you don't know how to interpret different faces, and simply take a reading and apply.


"... what zone this week's alien should be exposed in"

That is solely up to you, you're the artist.
However, If you're implying that the zone system subverts innovation and all that that connotes, you couldn't be farther from the truth. You'd be as well to ridicule the little squiggly lines on a music sheet.


And there is a fair bit of iffy info throughout this thread, when it comes to spot-meter pros and cons.