How do you know what wattage light to select?

[John Milich]

Im assuming "based on experience" is a big factor. Anyway, I have a bedroom set, thats approximately 150 sq/ft. Its supposed to be primarily lit by moonlight. The scene has the actor sitting at the edge of a bed, with a window obscured by some tree branches to the right, parallel to her face. I wanted the moonlight to serve as the key here, with a bounce to add some fill on the shadowed side of the face. This is a small set piece so its ground level and I have total control over it. I was figuring on maybe an HMI with a tungsten WB on the camera, (tungsten end table lights) but I've no idea what size/wattage light I should be looking at.

[Phil Rhodes]

Arri have a calculator on their site which gives an output in both lux and as a photographic exposure. I have used this several times in order to appear to be much more of a director of photography than I really am.

[Travis Gray]

There are a bunch of iPhone apps as well that'll help out with general wattages, etc. I like cinecalc pro. Has a bunch of brands built in, and you can set distances, get Lux/footcandles, find exposure. Then just figure out your ratios from there if needed. Has a gel calculator too to help figure out what you're starting with and need to get to and what gels to use (at least for Rosco, Gam, or Lee).

[John Milich]

Thanks for the answers and pointing me to that calculator. What a great tool! It looks like the ARRI Fresnel D5 would do the trick. At 20ft it puts out 803 Fc. Gives some play room for a bounce and some latitude with f-stops. Now this is a 6000 Kelvin light so it's definitely going to be on the blue side. Would this generally be ok, as is color wise, with no filtration for what I'm after? Moonlight that is. This is really my first foray into the realm of night shooting. Or at least making the scene look like it's night.

Edited August 19, 2013 by John Milich

[David Mullen ASC]

Yes, I think an uncorrected HMI would be fine for a blue color if you set your camera to 3200K. If you want less blue, you can select something closer to 4300K. You have some flexibility if the moonlight is the only color and source in the scene. If you have some tungsten table lamps, they will look warmer as you raise the color temp of the camera setting, but that may look nice too.

 

Remember that the moonlight would not be at full key or else it will look too bright, and as you underexpose the HMI color, it will look more saturated, a deeper blue.

 

If you are worried, carry some 1/4 and 1/2 CTO to warm up the HMI.

 

There is an old rule that says you need 100 foot-candles to get an f/2.8 at 100 ASA (assuming 24 fps, 1/48th shutter). So if your camera is more like 400 ASA, that's only 25 foot-candles needed for an f/2.8, and if the moonlight needs to be 2-stops underexposed, then you only need 6 foot-candles. This is one reason why the HMI's are good for the broad moonlight over the whole space but much lower-wattage, dimmer lights that need to be blue can be done with gelled tungsten or daylight LED's or fluorescents because you don't need much output for a fill light, let's say.

[John Milich]

Amazing information. Thank you all. With regards to the moonlight being underexposed. Is this the norm for that type of scenario/lighting? If I underexpose the Key by 2 stops, what does that do to the fills and tungstens? Would that just bring the ratio closer together? Sorry if thats a little confusing. Just want to make sure I have a good grasp of the concept.

[David Mullen ASC]

It would look odd for moonlight to be at full exposure, but it is also a relative thing, by which I mean, in real life, you would perceive moonlight as brighter if it were your only source of light, but compared to other light sources near you, the moonlight would feel dimmer.

 

So if I am doing a scene that is entirely lit by moonlight, I assume that the characters' and the viewers' eyes have gotten adjusted to the moonlight, so I don't underexpose as much, maybe by 1.5-stops if the moonlight is frontal (which I rarely do). It's trickier when it's a backlit moonlight, I would tend to expose the backlight as if it were the key light, not underexpose it if I were pointing the incident meter at it, but then I'd have to add some soft fill light if I didn't want a silhouette effect, and then I'd have to decide how dark the shadow side facing camera should be -- maybe 2-stops under, maybe more depending on the mood and contrast I wanted.

 

Now in a room with a few table lamps on, the moonlight would look dimmer in comparison to the room lights. If the moonlight were just some glow on the windows and raking across the wall, I may expose it to be quite dim, but perhaps the table lamps are creating a backlight or edge light and the light on the face is from the moonlight, then it can't be too underexposed in order to read the face, again, maybe 2-stops under. Also, how bright the moonlight looks depends on the tone of the object the light is hitting, you'd need less light in a white room with white curtain sheers than in a dark wood-paneled room.

 

If you are shooting digitally, you can balance a lot of these levels by monitor, knock things down until they look about right.

 

Whenever I begin a feature, I often shoot an under and overexposure test on a person so I see what 1-stop, 2-stop, etc. over and under looks.

[Guy Holt]

Whenever I begin a feature, I often shoot an under and overexposure test on a person so I see what 1-stop, 2-stop, etc. over and under looks.

 

A camera test, like the one David suggests, is an invaluable tool when it comes accomplishing the look you are after for a project. The standard approach is to systematically test the effect of Key, Fill, Back Light, Kickers, Hair Lights, and Liners that are over and under exposure. For example, to test the effect of your key light on flesh tones, set your exposure with two doubles and a single in your key light. Then remove them a half stop at a time (without changing your camera exposure setting or exposure of the chip chart), and systematically note on a slate in the frame what you are doing. Once you have removed all the scrims, your flesh tone will be two and a half stops over exposed (since you have not changed the camera setting.)

 

 

Put all the scrims back in and now, using single and double nets, systematically under expose the flesh tone in half stop increments (remember rotating a net relative to the light source will make it "fatter" or "thinner", which will enable you to "dial in" the exact level you want from the light.) Since you want to play on the lower register, continue to under expose the flesh tone until it becomes a pure silhouette. Do the same for Fill, Back Light, Kickers, Hair Lights, and Liners in isolation and in specific combinations that you plan to use them in. Having systematically tested each light, you can now see the effect that different levels of each has on the scene and can even use the test as a reference on set when lighting the scene.

 

 

So that your eye does not compensate for the low light levels, you should put a fully exposed white reference in the frame (the white foam-core in the background of the pictures.) If you use a chip chart with variable gray steps form white to black, you will actually be able to see how tonal values are compressed and (block up or burn out) as you push them onto the “knee” or “toe” of your film’s characteristic curve.

 

 

For a good explanation on how to light a dark scene, see David Mullen’s excellent post at http://www.cinematography.com/index.php?showtopic=55891. In it he warns not to “make the classic mistake of assuming that a dark image involved working in low light levels.”

 

Guy Holt, Gaffer, ScreenLight & Grip, Lighting and Grip Equipment Sales and Rentals in Boston.

[John Milich]

That's great info. Thanks everyone

[John Milich]

The above information is fantastic. I'm confused about one thing though. Does something in the scene, be it the corner of a cabinet or whatever, not have to be fully and properly exposed? I'm under the impression that shooting the whole scene underexposed would lead to a potentially noisy image.

[Phil Rhodes]

It'll only look noisy if you grade it back up to normal brightness, which is not what you wanted anyway. Noise is not something that magically happens at low exposures; it's always there, just invisible (or not very visible) in the shadows. Even if your entire image is made of shadows, it won't abruptly become more visible for any reason other than that the audience's eyes will eventually become dark-adapted.

 

That said, shooting things where the entire frame is underexposed can make things look unintentionally murky; personally I'd prefer to have something in shot that's motivated somehow to be better exposed, even if only in the deep background or out of focus.

[David Mullen ASC]

You have to mentally separate the overall exposure of the shot that allows you to color-correct or print it at a level that minimizes noise, and the exposure levels of portions of the frame.

 

In real life, and in decent lighting designs, you generally would not have every corner of the space lit to the same level. That would look very flat. And it's not necessary to control noise.

 

First there is the base noise level of the overall image, which depends on the camera, the ISO rating, etc. Then there is the noise that comes from brightening dark images up to normal brightness, either by using higher ISO ratings using cameras that apply gain of some sort to get those higher ratings, or by color-correcting in post to compensate for underexposure (or a high ISO rating for a camera that records raw) whether you brighten the whole frame or parts of the frame.

 

But if you select an ISO rating where the noise is low, and you don't try to brighten all or parts of the frame in post, the noise shouldn't change.

 

For example, imagine a brightly-lit room shot at a low-ish ISO rating, exposed properly, with low noise characteristics. In that shot, the main character turns off most of the room lights, plunging most of the room into darkness or creating deep shadows.

 

Well, if you don't make any corrections to that in post to brighten the dark room, the noise isn't going to increase just because a large percentage of the room went dark, anymore than just covering the lens with the lens cap is going to make the noise increase (unless your camera has auto-gain on). Or if the character, well-lit against a bright wall, crosses the room and stands in a narrow spotlight against a black velvet curtain -- he's still well-exposed but now the background is many, many stops underexposed. But the noise doesn't change just because the background is now very underexposed... unless you try to lift that black curtain up to a medium gray in post.

[joshua gallegos]

Yes, I think an uncorrected HMI would be fine for a blue color if you set your camera to 3200K. If you want less blue, you can select something closer to 4300K. You have some flexibility if the moonlight is the only color and source in the scene. If you have some tungsten table lamps, they will look warmer as you raise the color temp of the camera setting, but that may look nice too.

 

Remember that the moonlight would not be at full key or else it will look too bright, and as you underexpose the HMI color, it will look more saturated, a deeper blue.

 

If you are worried, carry some 1/4 and 1/2 CTO to warm up the HMI.

 

There is an old rule that says you need 100 foot-candles to get an f/2.8 at 100 ASA (assuming 24 fps, 1/48th shutter). So if your camera is more like 400 ASA, that's only 25 foot-candles needed for an f/2.8, and if the moonlight needs to be 2-stops underexposed, then you only need 6 foot-candles. This is one reason why the HMI's are good for the broad moonlight over the whole space but much lower-wattage, dimmer lights that need to be blue can be done with gelled tungsten or daylight LED's or fluorescents because you don't need much output for a fill light, let's say.

 

Wow, that's great! Everything is perfectly understandable when it comes to footcandles. So if I wanted the fill light in a scene to be 3stops under the Key exposure, then I can merely look into it in terms of footcandles. If I'm filming at 400ASA at F/5.6 for the Key, that would require 100 FCs, so I could set my Fill light the same way at 400 ASA at f/5.6 and stop down the light to f/2, which would be 13 footcandles. Something I would've never figured out, I suppose the cinematographers with experience can do it by eye, but I can just do that with a light meter and see how it looks whilst knowing the exact number of stops its set from the key.

Edited October 15, 2013 by joshua gallegos