f-stop, what I see versus what the camera sees.

[Jay Young]

I've been shooting a lot of little pickups and one-day shoots with my BlackMagic.

 

Recently, I tried to setup a lighting scheme where the background was at least 2 stops under the talent.

The talent was lit beautifully at a 4/5.6 split. Let's call it 5.6 for the sake of argument.

 

The "sun coming through window" read 2.8 on the meter, and the shadow area outside that read at most 1.4.

 

I've also been reading a lot about what others are doing in their lighting setups and I've either confused myself, or learned something wrong.

 

What is a full stop?

 

I guess I should have tried to light the background SIX stops down from the talent, as on the camera sensor it never did read how I wanted it, even though it looked good to my eye.

 

I keep hearing master cinematographers say things like f2.8 > f4 ' is a full stop ' , Except when I read about the theory and practice it is not. ISO and Shutter speed double, which is easy.

And even mathmatically, 2.8 is not double or half of 4. Is it just semantics?

 

I recently watched a video where the DP set the lens to f4, liked the look but wanted to open the lens up one stop so he threw in a ND.3, and opened to f2.8 - It looked good on the monitor, and he was happy, but I was not. Did he not just cut 1/2 a stop?

 

Why is it that some DP's call each half number a "full stop" and some call each doubling (f2, f4, f8) a "full stop" and who is correct?

I can tell you that it sure doesn't look like I let twice as much light in when I open up from 5.6 to 4, the math doesn't seem to support that either, but math is hard.

 

 

[Phil Rhodes]

I can tell you that it sure doesn't look like I let twice as much light in when I open up from 5.6 to 4

 

That is, however, how it works.

 

One stop represents a doubling or halving of the light from the next stop up or down. The halving or doubling is chosen because successive halves or doubles looks, visually, to humans, like equal steps in brightness. The actual difference between (say) f/5.6 and f/8 is much smaller in terms of actual photon count than the difference between f/2 and f/2.8.

 

Mathematically, the f number of a lens is the ratio between the focal length and the diameter of the entrance pupil (broadly, the hole through the lens). The reason they're weird numbers is that each stop allows double the light of the previous one, and you're talking about the area of a circle. The area of the circle roughly doubles between a circle of diameter 2.8 and diameter 4. Try it: pi * (0.5*2.8) is 6.16, whereas pi * (0.5*4) = 12.57.

 

The ND 0.3 filter absorbs 50% of the light going through it, so it's a one-stop ND filter. This equates to an optical density of 0.3, which is where the name comes from. I don't particularly like this numbering system, but it's what we've got. It's derived from logarithmic mathematics. The fractional transmittance can be calculated by raising 10 to the negative power of the optical density. 10 to the power -0.3 is 0.501, or about 50%.

 

A background exposed six stops down from the foreground would be very much in deep shadow, and possibly completely crushed to black even on a modern camera - depending, of course, on where you'd put the foreground. That may be what you wanted - sometimes, with less-than-ideal locations, it can be desirable to do that, and it can sometimes be quite hard to keep all spill and bounce off the background.

 

That said, it wasn't that long ago that cameras only had a range of seven or eight stops to begin with.

 

P

[Jay Young]

Thanks Phil!

 

I guess I just needed reassurance that I wasn't going crazy.

[Aapo Lettinen]

 

A background exposed six stops down from the foreground would be very much in deep shadow, and possibly completely crushed to black even on a modern camera - depending, of course, on where you'd put the foreground. That may be what you wanted - sometimes, with less-than-ideal locations, it can be desirable to do that, and it can sometimes be quite hard to keep all spill and bounce off the background.

 

Yep, 6 stops is 2/4/8/16/32/64 times darker than the foreground so it is indeed quite dimly exposed. 1 stop change is always double the light level/half of the light level. using a light meter when lighting is generally easier and faster than looking through camera if you are familiar with contrast ratios and your camera's dynamic range and are used to light by eye. you can use the camera/monitor scopes to ensure the right exposure and to fine tune lighting if needed

[Aapo Lettinen]

btw were all your meter reading incident or spot readings or were they mixed, if you are using zone system the readings have to be spot readings from camera direction to get it right. I got a little confused about the "sun" reading which did not seem right compared to the shadow area next to it, that's why I asked

[Jay Young]

Appo,

 

All my meter readings are incident. I only use the zone system when I shoot 8x10 - and even then I'm not sure I quite understand it.

[John E Clark]

Here is a table of f-stop to 'amount of light'... These are in 'full stop' increments... some tables will give 1/3 stop, and often a meter will give 1/3 increments as well.

 

Since 1 stop difference is either doubling or halving... the 'number of stops' difference results in a power of 2... so 6 stop difference is 2^6 in terms of amount of light.

 

 

[Aapo Lettinen]

(to Jay)

oh then it may get a little confusing. because if your background is not middle grey you should usually also take spot readings of it to know where it falls on the curve if you want to have it to look like for example two stops under key and it is darker/brighter (lower/higher albedo) than middle grey.

 

the all incident readings works best if the whole set is lit by same light, say, key light and ambience. I use it a lot when shooting documentary stuff on film with natural light.

 

usually I take more readings though, usually a reading of key light, say, sunlight at T8 incident, then the ambience reading, say, T2.8 incident, then the brightest area which needs to have details, say, the sky at T16 -22 spot, and the darkest area which needs to have details, say, a forest line reading at T2 spot.

 

if the film records well from 3.5 stops under to 7 stops over, I can set the forest line to about -3 stops which sets the aperture to T5.6 which sets the sunlit middle grey objects to one stop over, the sky 3 stops over, ambience lit middle grey object to 2 stops under.

= Forest from -3.5 to -3 > middle grey shadow -2 > -1 > camera aperture 0 > sunlit middle grey +1 > +2 > sky +3 - +4 stops.

Considering it's film and with best light transfer with good scanner and if the shadow areas are important I would probably expose closer to T4 to get:

Forest from -3.5 to -2 (more contrast and detail) > middle grey shadow -1 > camera aperture 0 > sunlit middle grey +2 > sky +4 - +5 stops. then pulling down the highlight in scanning about one stop to get the sky detail from +5 highlights.

 

But most of the time I'm just interested in ambience readings and how much they fall under base exposure, the idea being to expose the shadows between 1 and 2.5 stops under the incident ambience light depending on content and then correcting the highlights afterwards. film is easier for this being an "open top" format contrast wise, with video it is usually easier to correct the shadows in post than with film but the highlights are an issue.

but you can use the same principle with video also, by determining how much noise you can tolerate in image to set the usable range in stops which the video camera can capture. then it would be shadow spot, fill incident, key incident, highlight spot like with the film example and quite reliable end result

Edited March 22, 2016 by aapo lettinen

[Aapo Lettinen]

sorry for the lengthy posts. the point was that I use zone system usually only for the highlights and deeper shadows and use incident metering for the middle range and in low contrast situations.

 

in your example that would be sun lit area with spot metering, for example T2.0 (if T2.8 incident) if the floor is darker colour than middle gray. Thus starting with the readings of the least controllable light. That would set the camera aperture to T4 if the sunspot needs to be 2stops under.

 

Then I would lit the talent key to about T4 incident, and now I could see where the ambience level goes in the set so I could add ambience to the scene and could finally see how much extra fill is needed for the talent.

 

Lets say that the ambience reads T0.7 incident and the background is middle grey, which is three stops under the sunspot and five stops under the ""middle grey"" (base exposure middle grey T4) in scene. If it is OK that the background contrast is lower I would bring the ambience to about T1.0/1.4 split (about 1.5 stops under the sunspot reading). Then one could add fill to the talent if needed, to get the fill to maybe two stops below key light incident.

With Caucasian skin colour the spot reading would be something like one stop brighter than middle grey incident ? so it would be:

 

background middle grey -3.5 (T1.0/1.4 split spot reading)

> background sunspot -2 (T2 spot reading)

> talent shadow middle grey -2 (T2 spot reading)

> talent shadow skin tone -1 (spot reading)

> talent key middle grey 0 (T4 spot)

> talent skin tone +1 (T5.6 spot) , talent skin & hair reflections + X stops, maybe +3 ? (T11) .

 

That would be 6.5 stops of total contrast which the camera can easily handle and you can fine tune the contrast ratios from there to the taste.

Maybe the ambience could be on lower level if camera noise is not a problem and you can tweak the aperture up and down to get more highlight headroom/less noise to the image when pulling the overexposed image in colour correction.

But you can see that you need both incident AND spot readings when lighting a scene, especially when working in shadow or highlight end of the scene

Edited March 22, 2016 by aapo lettinen