Black Swan (The DP's decision to use a 7D)

[gibbo]

Hi guys,

Was just trying to figure out the DP's decision to go with a 7D for the subway scenes on Black Swan over a 5D, in the interview he states:

The 7D has more depth of field than the 5D, but I needed that because I didn’t have a follow-focus unit and needed to work really fast.

Now I know that the 5D has shallower depth of field than the 7D because it's full frame.

But he goes on to say:

I ended up shooting on a Canon 24mm lens at 1,600 ASA to get as much depth of field as possible at a stop of T81⁄2.

First of all, is a T-Stop equivalent to an F-Stop? So T-81/2 would be approx F-8/12? If so, he's basically that he used the 24mm at approx F8 to get a deeper DOF so he wouldn't need use extra gear like an external monitor etc (deeper DOF = more in focus), they just wanted to get in and out quickly, point and shoot style.

I may be confusing myself here but why not go for the 5D? At f8 on the 5D, wouldn't more light come through than the same lens at f8 on the 7D? Hence he could have used a lower ISO?

Or would f8 on the 5D also come with a shallower DOF too? Which is probably the case...

On another note: How the hell did he manage to use f8 and ISO 1600 indoors at night? my living room is a lot brighter than the subways are and I have to use f2.8 and ISO 1600... I must be missing something here?

FYI, if you haven't read the interview, it's pretty interesting:

http://www.alexandrosmaragos.com/2010/12/black-swan-canon-7d.html

Thanks....

 

[chilipie]

Hi guys,

Was just trying to figure out the DP's decision to go with a 7D for the subway scenes on Black Swan over a 5D, in the interview he states:

The 7D has more depth of field than the 5D, but I needed that because I didn’t have a follow-focus unit and needed to work really fast.

Now I know that the 5D has shallower depth of field than the 7D because it's full frame.

Black Swan was shot on 16mm - I suspect shooting on a 5D would have been too much of a mis-match in terms of sensor size.

But he goes on to say:

I ended up shooting on a Canon 24mm lens at 1,600 ASA to get as much depth of field as possible at a stop of T81⁄2.

First of all, is a T-Stop equivalent to an F-Stop? So T-81/2 would be approx F-8/12? If so, he's basically that he used the 24mm at approx F8 to get a deeper DOF so he wouldn't need use extra gear like an external monitor etc (deeper DOF = more in focus), they just wanted to get in and out quickly, point and shoot style.

An f-stop is an entirely geometrical calculation - it is the diameter of the aperture in relation to the focal length of the lens, with no allowance made for the number of elements in a lens and the rate at which some of them absorb light. A T-stop is a transmission stop; what the f-stop would be in a perfect world where every element of the lens transmitted 100% of the light.

The difference between the f-stop and T-stop of a given lens are usually very slight, particularly on prime lenses - a T1.3 lens may have an f-stop of f/1.2; a T2.8 zoom lens may have an f-stop of f/2.6. Depth of field calculations etc. are done with f/stops; T-stops only matter when it comes to exposure.

I may be confusing myself here but why not go for the 5D? At f8 on the 5D, wouldn't more light come through than the same lens at f8 on the 7D? Hence he could have used a lower ISO?

Or would f8 on the 5D also come with a shallower DOF too? Which is probably the case...

That's the beauty (and entire point) of the f/stop. If you used an absolute measurement - say you had a 50mm lens, setting the aperture diameter to 25mm would be the equivalent of f/2. But when you moved to a 100mm lens for the next shot, setting the aperture to 25mm would give you an aperture of f/4. For an aperture of f/2 on a 100mm lens, you'd need to set the aperture diameter to 50mm.

I admit these are very simple calculations, but in practice it wouldn't be quite so simple. It's much easier to just turn a dial to f/5.6 than to try and calculate 37mm divided by 5.6, for example.

While f/8 on the 5D would have let the same amount of light in, to get the same angle of view as on the 7D you would need to use a longer lens… and so the background would be more out of focus.

On another note: How the hell did he manage to use f8 and ISO 1600 indoors at night? my living room is a lot brighter than the subways are and I have to use f2.8 and ISO 1600... I must be missing something here?

I've seen at least one photo of Matty Libatique holding up an LED panel a few feet away from Natalie Portman's face. Though I suppose the obvious answer is that if you're getting f/2.8 & ISO 1600 then your living room just isn't brighter than the NY subway they shot on

 

[directorik]

Nice answer chilipie.

 

[David.rhsc]

I may be confusing myself here but why not go for the 5D? At f8 on the 5D, wouldn't more light come through than the same lens at f8 on the 7D? Hence he could have used a lower ISO?

Not exactly. Changing camera bodies does not change the amount of light passing through a lens. The difference is the larger photo-sites on the sensor. For example, if you have a 1"x1" square divided into 24 million pixels, and a similar square divided into 16 million pixels, the pixels on the later square are generated by lighting hitting photo-sites that are larger in physical size, and thus more sensitive to light. One could, in theory use a different ISO on the two bodies. I would be surprised if the difference was more than a stop or so between the two sensors. In the end though, the choice was probably as much about getting closer to the 16mm image as it was about technical factors.

Also, the point that he brought the footage to a colorist at Technicolor is not to be missed.

 

[Phil_UK]

I think another relevant answer, as to why he chose the 7D over the 5D, is first of all the sensor size, as the 5D is a Vistavision sensor and would differ to much to what they shot on (Stated by chilipie), but the other reason is that the 7D has a HDMI output of 1080p, even when recording, and this can be a very very big reason, as to why they favoured that over a 5D, which switches to 480p!

Focus really matters on these huge productions and in 480p, spotting focus can be tough!

 

[gibbo]

Thanks for the replies guys!

Black Swan was shot on 16mm - I suspect shooting on a 5D would have been too much of a mis-match in terms of sensor size.

I don't know much about digital and even less about film. I had thought that all film was "full frame" so to speak. Is there any relation between canon's 1.6 crop sensor and the 16mm film size?

An f-stop is an entirely geometrical calculation - it is the diameter of the aperture in relation to the focal length of the lens, with no allowance made for the number of elements in a lens and the rate at which some of them absorb light. A T-stop is a transmission stop; what the f-stop would be in a perfect world where every element of the lens transmitted 100% of the light.

The difference between the f-stop and T-stop of a given lens are usually very slight, particularly on prime lenses - a T1.3 lens may have an f-stop of f/1.2; a T2.8 zoom lens may have an f-stop of f/2.6. Depth of field calculations etc. are done with f/stops; T-stops only matter when it comes to exposure.

Still pretty confused about this T-stop business! You'll have to excuse me...

So basically if you just through a T-stop out there without any other info (lens used etc) it's meaningless? Whereas with an F-stop it's not.

So, the only way to find the equivalent F-stop is to know the exact lens someone used? In this case, T8 on a 24mm lens would be equivalent to F3? (Actually, he says T81/2 so it would really be F2.8)

That's the beauty (and entire point) of the f/stop. If you used an absolute measurement - say you had a 50mm lens, setting the aperture diameter to 25mm would be the equivalent of f/2. But when you moved to a 100mm lens for the next shot, setting the aperture to 25mm would give you an aperture of f/4. For an aperture of f/2 on a 100mm lens, you'd need to set the aperture diameter to 50mm.

I admit these are very simple calculations, but in practice it wouldn't be quite so simple. It's much easier to just turn a dial to f/5.6 than to try and calculate 37mm divided by 5.6, for example.

While f/8 on the 5D would have let the same amount of light in, to get the same angle of view as on the 7D you would need to use a longer lens… and so the background would be more out of focus.

That makes sense, would that mean that the DOF is shallower too? Or the additional blur would be strictly because of the focal length and nothing to do with DOF?

I've seen at least one photo of Matty Libatique holding up an LED panel a few feet away from Natalie Portman's face. Though I suppose the obvious answer is that if you're getting f/2.8 & ISO 1600 then your living room just isn't brighter than the NY subway they shot on

Hehe, that could be true! But, if my above calculations are correct (24mm/T8.5 = F2.8), then his settings are that much different to mine (please let them be true!)

Not exactly. Changing camera bodies does not change the amount of light passing through a lens. The difference is the larger photo-sites on the sensor. For example, if you have a 1"x1" square divided into 24 million pixels, and a similar square divided into 16 million pixels, the pixels on the later square are generated by lighting hitting photo-sites that are larger in physical size, and thus more sensitive to light. One could, in theory use a different ISO on the two bodies. I would be surprised if the difference was more than a stop or so between the two sensors. In the end though, the choice was probably as much about getting closer to the 16mm image as it was about technical factors.

It's just that in the interview he didn't mention anything about the sensor size as a reason, just the DOF. Theoretically, couldn't they have just used a short focal length on the 5D to achieve the same angle of view as the 7D?

Also, the point that he brought the footage to a colorist at Technicolor is not to be missed.

Can you elaborate please?

I think another relevant answer, as to why he chose the 7D over the 5D, is first of all the sensor size, as the 5D is a Vistavision sensor and would differ to much to what they shot on (Stated by chilipie), but the other reason is that the 7D has a HDMI output of 1080p, even when recording, and this can be a very very big reason, as to why they favoured that over a 5D, which switches to 480p!

Focus really matters on these huge productions and in 480p, spotting focus can be tough!

I thought that the 7D did 480p in live mode and while recording but for HDMI out playback it was 1080p?

 

[harmonica44]

But there are a lot of movies shot on 35mm then put on 16 for theater showings. So couldn't you do the same with putting full frame digital on 16mm film? Some movies are shot with both 35 and 16 and it all looks the same in the end, so I think that they could do something, but I am far from an expert of course, it just doesn't seem like a good reason, since they have done stuff like this before.

 

[chilipie]

I don't know much about digital and even less about film. I had thought that all film was "full frame" so to speak. Is there any relation between canon's 1.6 crop sensor and the 16mm film size?

The 5D is 35mm full frame; motion picture 35mm is close to Canon APS-C (the size of the sensor in the 7D, 60D, 550D etc.). Super 16mm is slightly larger than 2/3".

Still pretty confused about this T-stop business! You'll have to excuse me...

So basically if you just through a T-stop out there without any other info (lens used etc) it's meaningless? Whereas with an F-stop it's not.

So, the only way to find the equivalent F-stop is to know the exact lens someone used? In this case, T8 on a 24mm lens would be equivalent to F3? (Actually, he says T81/2 so it would really be F2.8)

I think you've got the wrong end of the stick - you can't calculate T-stops by dividing unconnected numbers. For the purposes of this discussion, let's just say that f-stops and T-stops are the same thing.

Spoiler

If you really want to know more, read on… if we have a lens with a maximum aperture of f/2, then when we shine 100 units of light through it, the sensor should receive 50 units (f/1 would let all the light through, f/4 would let a quarter of the light through etc.). However, because none of the glass elements in the lens are capable of transmitting 100% of the light and some of it is absorbed, only 40 units of light are received by the sensor. This means that 1/3 of a stop has been lost (a whole stop loss would mean 25 units of light reached the sensor).

If we used a light meter to calculate an exposure for a shot, and it told us to set the aperture to f/2, when we took the shot it would be underexposed by 1/3 of a stop. However, because we calibrated the lens in the lab, we made a marking on the aperture ring at f/2 that says T2.2 - one third of a stop slower. By doing this we can be sure that a shot is exposed correctly.

That makes sense, would that mean that the DOF is shallower too? Or the additional blur would be strictly because of the focal length and nothing to do with DOF?

A bit of both… a longer focal length reduces depth of field, but you'd also have to move the camera further away from the plane of focus, which would increase depth of field. However, longer focal lengths make the background appear to be more out of focus - when actually they're just magnifying the background more than what's sharp - so it appears that the depth of field is shallower.

It's just that in the interview he didn't mention anything about the sensor size as a reason, just the DOF. Theoretically, couldn't they have just used a short focal length on the 5D to achieve the same angle of view as the 7D?

Sensor size is intrinsically linked to depth of field. No - the equivalent angle of view is often discussed in terms of crop factor. To get the same angle of view on a 5D as on a 7D, you need to use a lens that is 1.6 times longer. There's no way around that, it's a physical impossibility.

Can you elaborate please?

Having your post-production work done at a high-end facility by experienced technicians will help enormously when it comes to matching shots between cameras and getting the most information from an image.

I thought that the 7D did 480p in live mode and while recording but for HDMI out playback it was 1080p?

The 7D will do 1080 out (I can't remember if it's p or i) while recording. As it says in the article though, they only used the camera monitor, so I don't know how much the 7D's HDMI capabilities factored into their final choice.

But there are a lot of movies shot on 35mm then put on 16 for theater showings. So couldn't you do the same with putting full frame digital on 16mm film? Some movies are shot with both 35 and 16 and it all looks the same in the end, so I think that they could do something, but I am far from an expert of course, it just doesn't seem like a good reason, since they have done stuff like this before.

I have very little idea what you're trying to say. Black Swan was shot in 16mm but released on 35mm; cinemas that project in 16mm are few and far between.

 

[David.rhsc]

Can you elaborate please?

Well, Chili already nailed the technical details that you're asking about so that work is done.

Experience level and tools aside, the objective was likely always to create a desired look. Being able to work with a colorist at a pro facility aids that process and if I can risk some conjecture - the decision to use the 7d was as much an aesthetic one as a technical one. Possibly more so. Once the decision to go that route on the subway was made, it was a matter of finding the tool and parameters that gave the desired look.

 

[CRAYONSEED]

Spoiler

If you really want to know more, read on… if we have a lens with a maximum aperture of f/2, then when we shine 100 units of light through it, the sensor should receive 50 units (f/1 would let all the light through, f/4 would let a quarter of the light through etc.). However, because none of the glass elements in the lens are capable of transmitting 100% of the light and some of it is absorbed, only 40 units of light are received by the sensor. This means that 1/3 of a stop has been lost (a whole stop loss would mean 25 units of light reached the sensor).

If we used a light meter to calculate an exposure for a shot, and it told us to set the aperture to f/2, when we took the shot it would be underexposed by 1/3 of a stop. However, because we calibrated the lens in the lab, we made a marking on the aperture ring at f/2 that says T2.2 - one third of a stop slower. By doing this we can be sure that a shot is exposed correctly.

Thanks Chili! That was the first explanation of the difference between f/stops and T-stops that I was able to actually visualize.

 

[NickClapper]

chilipie=

 

[gibbo]

Thanks for the replies guys, all that is a bit over my head at the moment, I will have to revisit this thread when I know more about cinematography and cameras.

In the meantime can you guys recommend a few Cinematography books and or resource that are personal favourites of yours?

Thanks!

 

[Ernest Worthing]

Start with The Five C's of Cinematography.

 

[IndieBudget]

https://www.youtube.com/watch?v=W4qL-x1wTl8

 

[jax_rox]

There seems to be a lot caught up in talk of T-stops versus F-stops when in reality it doesn't have much to do with why you'd choose the 7D over the 5D. Yes, it has to do with sensor size, but not in terms of a 'better match' - the 7D still has a ~S35mm sized sensor and is therefore ~double that of S16mm anyway. It also isn't to do with the 'look' of it, as the 7D and the 5D look very similar. It's simply to do with the fact that as he says, the 7D's smaller sensor than the 5D affords it a deeper DOF. Sensor size is directly related to Depth of Field - the bigger the sensor, the shallower the DOF. This is why a 7D is a better choice than a 5D unless you want super, super shallow DOF - and tbh, only hardcore DSLR fanboys want that...
Take for example a classroom scene. Because of the 5Ds larger sensor than the 7D, you'd actually need to pump in quite a bit more light into the scene to be able to stop down enough to keep everything in focus than you would on the 7D. This is why it was used in those scenes - no FF, no Focus Puller, stop down to f/8 and keep things in acceptable focus. On the 5D you would have had to have either bring in extra lighting or bump the ISO more.

 

[fernando97echeverria]

Black Swan, for me....Man, what an experience. What a great trip in terms of everything, but mostly the cinematography. Even just the opening shot was amazing.

I also love the choice of having reflective surfaces in almost every shot. Genius way to convey her obsession over her image.

 

[Anim8]

Start with The Five C's of Cinematography.

Is one of those ”Chilipie”?

 

[sonnyboo]

I think it is a simple misunderstanding about the term "depth of field". The DP states he wanted MORE depth of field, meaning more in focus, which is harder to do with the 5D. If he wanted a SHALLOWER, meaning less depth of field, he would have gone 5D...