Ryan Patrick O'Hara
Veteran
I was having a conversation with a friend about Stanley Kubrick's use of the F/.7 Nasa lens in the candlelit scenes from Barry Lyndon.
We were originally discussing my article on the future need of super-fast aperture lenses, and he mentioned the lens which I allude to in the article as "with the exception of rare/specialty lenses and film stocks, the 35mm format lens has settled around T/1.3 or T/1.4 as the standard super-fast lens".
We started talking about the lens and I got off on a HUGE tangent. Thought it was interesting and perhaps I'd share it here for others.
Because we were discussing my article which believes the need for T/1.4 or faster lenses are no longer a common necessary and in the future, much at all. So I proposed the following:
A F/.7 lens is two stops faster than a F/1.4.
Barry Lyndon was shot with Kodak Eastman 100T 5254 film... which is three stops slower than a RedOne MX camera rated at 800iso... thus, if Kubrick was alive today, and shooting that scene on a RedOne MX... he could have gotten the SAME exposure using a Cooke S4 at a T/2.
I'll even venture to say that ... one could, just maybe, get the same exposure with a Cooke Panchro at a T/2.8 ... and here's why.
My logic leads me towards this theory because they were using a *modified* F/.07 lens... and F-stops are not a true measure of light transmission. They are simply mathematical calculations of light transmission. F-stops tend to be over-rated slightly on sensitivity, and since this is such a fast lens and modified, I'd be willing to guess it was closer or at a T/1 after physics was all said and done.
The reason F-stops are not truly accurate is because it is a simple formula (N=f/D) which does not account for the light loss due to the reflected light waves bouncing off the glass elements inside a lens. Glass is very clear, but 100% of light does not pass through it. Some of the light will bounce off the glass as it passes through. In fact it's about 4% per element. I don't know what is inside that crazy F/.7 lens, but the average cooke S4 has about 10-14 elements. Let's assume the F/.7 lens has eight elements... this is a cautious number I think we can work with.
Take 100% of light and assuming each element side is a 4% light loss we need to multiply (100 x .96) 16 times.
100
x.96= 96
x.96= 92.16
x.96= 88.47
x.96= 84.93
x.96= 81.53
x.96= 78.27
x.96= 75.14
x.96= 72.13....<- 4 elements = a little over 1/2 stop loss.
x.96= 69.25
x.96= 66.48
x.96= 63.82
x.96= 61.27
x.96= 58.82
x.96= 56.46
x.96= 54.20
x.96= 52.04... <- 8 elements = *almost* 1 stop loss
That means if 100% of light enters the lens... only 52% of that light makes it out. The rest is erroneous fodder bouncing out of or around the inside of the lens.
52% is simply 2% off from a whole stop. Meaning if the lens has more than 8 elements, the lens would be a T/1 or greater... if the lens has 8 or less elements it would be somewhere between a solid T/.7 and a solid T/1.
This, off course, is simply a more accurate math equation, not the real T -stop. A real transmission test would involve having the actual lens, shoving some light through it and measuring the light out of the other side.
In addition, there are other factors to be accounted for, such as the lens designers could have done edge blackening. This is painting the cylindrical edges of the lens elements with a matte black finish. Helps keep light from bouncing around. Also, vignetting can be a factor which can effect the T-stop of a lens, or at least even illumination across the entire screen. This is reliant on the quality of the lens build and the focal length. As mentioned, the average percentage of light that bounces off lens elements is 4%... but wider angle lenses are more susceptible. A telephoto lens has a very narrow field of view. Meaning it sees a tight frame of what is straight ahead. The reflected light it is collecting is originating from directly in front of the lens. When it strikes the lens elements, it is a relatively straight on angle and simply needs to be bent slightly to get it to the film plane.... however very wide angle lenses see a much wider angle / field of view. It collects reflected light waves traveling from a much greater angles. When light strikes glass at an angle of 90*, or straight on, the light is perpendicular. When light strikes glass at an incoming angle of, say, 45*, it is more susceptible to bounce. This is called the Law of Reflection (http://www.physicsclassroom.com/class/refln/u13l1c.cfm)...
Hence the light waves coming in from dramatic angles will bounce more... probably greater than 4%. The light, however, coming in from relatively straight ahead, say 90*, will enter with minimum bounce. This is *one* of the reasons cheap or old lenses have vignetting issues. Because light in the center of the lens will get through with less trouble than light coming from the peripheral. Of course, more expensive lenses counter this by doing a better job with lens coatings, collecting the light, and bending it correctly. I do not know how the Zeiss lens performed. I know it was super expensive back in the day, and probably the best technology could produce... but I'm sure there had to have been some optical sacrifice to reach a stop of f/.7.
I also do not know if the light coming from 90* would be less than the average of 4% light reduction per element surface, but if it is, it could counteract the exposure loss from the additional reflection of the light from the dramatic angles of reflection.
Again, with Kubricks lens, it being a 50mm, they were probably not terribly effected by increased light bounce from the law of reflection. But it is something that should be addressed when determining the true transmission value of the lens.
So my last thought is simply this... I'd bet a good deal that the T-stop of the F/.7 lens is actually T/1... and because the film emulsion was three stops slower than a RedOne MX'd camera, a T/2.8 Cooke Panchro 50mm lens would get the equivalent exposure today!
I was a bit excited about this, until I dug out my American Cinematographer magazine from March 1976 and read the unfortunate fact that Alcott pushed the negative one stop for the entire picture.
This sadly made the film stock the equivalent of a 200iso/asa stock... meaning I lost a stop!
With the film rated at 200iso/asa, the sensitivity of the medium is now only two stops slower than a RedOne MX camera rated at 800iso. Hence, a T/2.8 Panchro would have to be replaced with a Cooke S4 at T/2... or we'd have to bump the RedMX iso to 1600.
Sad day. I guess I can't shoot Kubricks all candle light scene with absolutely NO lighting with my T/2.8 lenses!
I'll just sneak in some practicals or simply add more candles!
Whew.
We were originally discussing my article on the future need of super-fast aperture lenses, and he mentioned the lens which I allude to in the article as "with the exception of rare/specialty lenses and film stocks, the 35mm format lens has settled around T/1.3 or T/1.4 as the standard super-fast lens".
We started talking about the lens and I got off on a HUGE tangent. Thought it was interesting and perhaps I'd share it here for others.
Because we were discussing my article which believes the need for T/1.4 or faster lenses are no longer a common necessary and in the future, much at all. So I proposed the following:
A F/.7 lens is two stops faster than a F/1.4.
Barry Lyndon was shot with Kodak Eastman 100T 5254 film... which is three stops slower than a RedOne MX camera rated at 800iso... thus, if Kubrick was alive today, and shooting that scene on a RedOne MX... he could have gotten the SAME exposure using a Cooke S4 at a T/2.
I'll even venture to say that ... one could, just maybe, get the same exposure with a Cooke Panchro at a T/2.8 ... and here's why.
My logic leads me towards this theory because they were using a *modified* F/.07 lens... and F-stops are not a true measure of light transmission. They are simply mathematical calculations of light transmission. F-stops tend to be over-rated slightly on sensitivity, and since this is such a fast lens and modified, I'd be willing to guess it was closer or at a T/1 after physics was all said and done.
The reason F-stops are not truly accurate is because it is a simple formula (N=f/D) which does not account for the light loss due to the reflected light waves bouncing off the glass elements inside a lens. Glass is very clear, but 100% of light does not pass through it. Some of the light will bounce off the glass as it passes through. In fact it's about 4% per element. I don't know what is inside that crazy F/.7 lens, but the average cooke S4 has about 10-14 elements. Let's assume the F/.7 lens has eight elements... this is a cautious number I think we can work with.
Take 100% of light and assuming each element side is a 4% light loss we need to multiply (100 x .96) 16 times.
100
x.96= 96
x.96= 92.16
x.96= 88.47
x.96= 84.93
x.96= 81.53
x.96= 78.27
x.96= 75.14
x.96= 72.13....<- 4 elements = a little over 1/2 stop loss.
x.96= 69.25
x.96= 66.48
x.96= 63.82
x.96= 61.27
x.96= 58.82
x.96= 56.46
x.96= 54.20
x.96= 52.04... <- 8 elements = *almost* 1 stop loss
That means if 100% of light enters the lens... only 52% of that light makes it out. The rest is erroneous fodder bouncing out of or around the inside of the lens.
52% is simply 2% off from a whole stop. Meaning if the lens has more than 8 elements, the lens would be a T/1 or greater... if the lens has 8 or less elements it would be somewhere between a solid T/.7 and a solid T/1.
This, off course, is simply a more accurate math equation, not the real T -stop. A real transmission test would involve having the actual lens, shoving some light through it and measuring the light out of the other side.
In addition, there are other factors to be accounted for, such as the lens designers could have done edge blackening. This is painting the cylindrical edges of the lens elements with a matte black finish. Helps keep light from bouncing around. Also, vignetting can be a factor which can effect the T-stop of a lens, or at least even illumination across the entire screen. This is reliant on the quality of the lens build and the focal length. As mentioned, the average percentage of light that bounces off lens elements is 4%... but wider angle lenses are more susceptible. A telephoto lens has a very narrow field of view. Meaning it sees a tight frame of what is straight ahead. The reflected light it is collecting is originating from directly in front of the lens. When it strikes the lens elements, it is a relatively straight on angle and simply needs to be bent slightly to get it to the film plane.... however very wide angle lenses see a much wider angle / field of view. It collects reflected light waves traveling from a much greater angles. When light strikes glass at an angle of 90*, or straight on, the light is perpendicular. When light strikes glass at an incoming angle of, say, 45*, it is more susceptible to bounce. This is called the Law of Reflection (http://www.physicsclassroom.com/class/refln/u13l1c.cfm)... Hence the light waves coming in from dramatic angles will bounce more... probably greater than 4%. The light, however, coming in from relatively straight ahead, say 90*, will enter with minimum bounce. This is *one* of the reasons cheap or old lenses have vignetting issues. Because light in the center of the lens will get through with less trouble than light coming from the peripheral. Of course, more expensive lenses counter this by doing a better job with lens coatings, collecting the light, and bending it correctly. I do not know how the Zeiss lens performed. I know it was super expensive back in the day, and probably the best technology could produce... but I'm sure there had to have been some optical sacrifice to reach a stop of f/.7.
I also do not know if the light coming from 90* would be less than the average of 4% light reduction per element surface, but if it is, it could counteract the exposure loss from the additional reflection of the light from the dramatic angles of reflection.
Again, with Kubricks lens, it being a 50mm, they were probably not terribly effected by increased light bounce from the law of reflection. But it is something that should be addressed when determining the true transmission value of the lens.
So my last thought is simply this... I'd bet a good deal that the T-stop of the F/.7 lens is actually T/1... and because the film emulsion was three stops slower than a RedOne MX'd camera, a T/2.8 Cooke Panchro 50mm lens would get the equivalent exposure today!
I was a bit excited about this, until I dug out my American Cinematographer magazine from March 1976 and read the unfortunate fact that Alcott pushed the negative one stop for the entire picture.
This sadly made the film stock the equivalent of a 200iso/asa stock... meaning I lost a stop!
With the film rated at 200iso/asa, the sensitivity of the medium is now only two stops slower than a RedOne MX camera rated at 800iso. Hence, a T/2.8 Panchro would have to be replaced with a Cooke S4 at T/2... or we'd have to bump the RedMX iso to 1600.
Sad day. I guess I can't shoot Kubricks all candle light scene with absolutely NO lighting with my T/2.8 lenses!
I'll just sneak in some practicals or simply add more candles!
Whew.
Last edited:
They would have used telescopes for that.