How the Alexa really works

[David Mullen ASC]

It's still a bad idea in general to use the Alexa at 200 ISO... this theoretical discussion ignores basically how the images look, which is dangerous. And it's even more dangerous when you are talking about cameras other than the Alexa which do weird things near the clip point.

[Alexandros Angelopoulos Apostolos]

How much of semiconductor physics and optics does a cinematographer’s education include?

[Phil Rhodes]

On one hand, it doesn't matter, because we should be concerned by what the pictures look like.

 

On the other hand, knowing why things behave the way they do is sometimes useful in predicting or understanding the behaviour of a camera.

 

I think a "pilot" approach is usually appropriate. Pilots aren't expected to know every nut, bolt and washer on the aircraft, but they are expected to maintain a reasonably detailed working knowledge of how it operates, so that if anything goes wrong it can be reacted-to appropriately and with an understanding of what's happening.

 

P

[Alexandros Angelopoulos Apostolos]

On one hand, it doesn't matter, because we should be concerned by what the pictures look like.

 

Let me then rephrase: can you know what a picture will look like if you know zilch about the optics of a lens or some basics of digital imagery or how to calculate the difference in exposure between two T-stops?

 

Which is this, I guess.

 

 

On the other hand, knowing why things behave the way they do is sometimes useful in predicting or understanding the behaviour of a camera.

 

 

In any case, whether it matters or not, the question is do they. Or do they not.

[David Mullen ASC]

Some cinematographers who have gone to certain film schools like the one in Lodz in Poland get some education in optics, but most cinematographers do not get that training... And yet we've managed to shoot some pretty good-looking movies for over 100 years. Do painters need a degree in chemistry? Doesn't hurt, but it's not necessary.

 

How do you know what a picture will look like with a certain lens on it? You shoot something with that lens and find out. Or you shoot digitally and see what the lens image looks like immediately while you are shooting.

 

Cinematography is both an art and science, but the level of scientific knowledge needed to shoot good images is not on the level that a scientist or engineer needs. There are many levels between "zilch" and being a technical expert on the subject.

 

Before the days of digital, I once was at a lab demo of the ENR process and I discovered that some of the other professional cinematographers there did not even understand how color film normally works in order to understand what a bleach bypass process would do. But I'm sure they all could shoot a good-looking movie. Deep technical knowledge is useful for problem-solving but then again, there are experts to help you out when needed, there are many levels of technical support out there.

[Tyler Purcell]

It's still a bad idea in general to use the Alexa at 200 ISO... this theoretical discussion ignores basically how the images look, which is dangerous. And it's even more dangerous when you are talking about cameras other than the Alexa which do weird things near the clip point.

Well, when you're on location in direct sunlight and you're using 1.2 stops of ND, the lens is all the way closed and it's still peaking, you've got no other option. You can protect highlights if you're very careful, I've done it, it's not rocket science. But you can't be stuck running the camera all the way closed and have nowhere else to go, that's a worse scenario in my book.

[David Mullen ASC]

You could also close down the shutter angle, but I think the key thing is to recognize you need ND filters that go to 1.5 or 1.8 if you are going to use the Alexa. Besides, lowering the ISO isn't going to make it an easier to protect highlights in bright sunlight because you're not actually changing the amount of light hitting the sensor -- and if you lower the ISO but then protect the highlights by stopping down some more, then you're right back where you started if you were using the lower ISO in order to not stop down the lens.

 

Besides, you of all people should be wanting more of a film look from the Alexa, which means that highlight latitude should be higher than shadow detail recorded. The Alexa at 200 ISO starts to look hyper-clean and video-ish to me.

[Tom Yanowitz]

How often do you end up in a situation were 1.2 ND, ~F/8 and 1/50 second still let too much light in though ?

[Tyler Purcell]

Yea, I always turn the shutter angle down, usually to 45. Much below that and it becomes too staccato.

 

If you run 1.2 stops of ND, 200 ASA and 45 degree shutter, you can usually get around f8 sure.

 

But yes David, wholeheartedly agree the camera doesn't look as filmic in that situation. I've also been running a grad more recently on shoots where the sky is so overly bright, I can't control the highlights. So I protect them by using the grad and lift them in post production. So far that trick has worked well, but I haven't used it with actors yet. This way, you're not using as much ND (which is harder to correct), you're protecting those highlights AND you've got a cleaner, less noisy image to work with.

[John E Clark]

How much of semiconductor physics and optics does a cinematographer’s education include?

 

A photographer need not know the basic science behind a technology... photographers in the past did not have to take physics or chemistry courses to shoot film and process it... (although perhaps basic chem lab safety was in order, if one processed one's own film...)

 

But by the same token, the should test ever claim made by a manufacturer... ISO being a popular buzz marketing element. Resolution, and what it 'reallly' means for one's own work.

 

Optics... being able to understand why lens X is $200 and lens Y is $50K... and that 'knowledge' may be limited to the $50K is far more precisely made... and 'I don't need that precision for my work'...

 

And of course... there is what I call Camera Knobology... just how to access the functionality of the camera one has at hand.

 

At the Wife's still photo seminars, I'd often spend many minutes getting people to know what their camera could be setup for... many I've never seen before till that moment, and fortunately the owners did have their manuals with (pre ubiquitous Internet access... now I just look up the manual online...).

[John E Clark]

How often do you end up in a situation were 1.2 ND, ~F/8 and 1/50 second still let too much light in though ?

 

Typical So. Cal bright sun day... The other day for a test of a lens (which has vignetting... an ebay oops... but I digress...) I was at f/16, @ ISO 800 @ 180 @ 24 fps, for my BMPCC, in the shade... my meter indicated f/22. So, sure ND 1.2 would have just worked.

[David Mullen ASC]

It's the old Sunny 16 rule. In direct sunlight on a clear day, the exposure is f/16 at 50 ASA if the shutter time is 1/50th. And an ND1.2 (4-stops) would reduce 800 ASA to 50 ASA.

[Alexandros Angelopoulos Apostolos]

Some cinematographers who have gone to certain film schools like the one in Lodz in Poland get some education in optics, but most cinematographers do not get that training... And yet we've managed to shoot some pretty good-looking movies for over 100 years. Do painters need a degree in chemistry? Doesn't hurt, but it's not necessary.

 

How do you know what a picture will look like with a certain lens on it? You shoot something with that lens and find out. Or you shoot digitally and see what the lens image looks like immediately while you are shooting.

 

Cinematography is both an art and science, but the level of scientific knowledge needed to shoot good images is not on the level that a scientist or engineer needs. There are many levels between "zilch" and being a technical expert on the subject.

 

Before the days of digital, I once was at a lab demo of the ENR process and I discovered that some of the other professional cinematographers there did not even understand how color film normally works in order to understand what a bleach bypass process would do. But I'm sure they all could shoot a good-looking movie. Deep technical knowledge is useful for problem-solving but then again, there are experts to help you out when needed, there are many levels of technical support out there.

 

I understand.

 

But look at this.

 

Just a few minutes ago, I searched for discussions about Ansel Adams and his zone system on cinematography.com. I knew that surely there are at least five threads about it. So why ask when it’s already been covered? Bam!, there pops a thread, in which you reply like this:

 

http://www.cinematography.com/index.php?showtopic=70275&p=447291

 

I haven’t understood a word of it. I ran as if pursued by a bunch of wild wolves. Yes, I could badger you about it, but there is no point now for it, for various reasons.

 

Everywhere you look, a thread eventually, in quite a bit of cases, ends up being somewhat “technical”. Which all makes me think that you need to be pretty damn versed into everything from lamp filaments in the 19th century to semiconductor physics, de-Bayering, lens optics and all the rest of it to be able to record an image.

 

When you say that how else will you know how an image will look unless you go out with a camera, mount that lens, and record an image, I keep thinking how I was under the impression that a cinematographer needs to know something like all the lenses available today and what sort of an image those lenses would give coupled with a particular sensor or film stock.

 

For example, here’s a question: You shot with Alexa 65, correct? I think you did. So how did you know what it will all look like before you had the camera in your hand? Does renting that camera start much before the film starts shooting in order for a cinematographer to do a couple of tests to know what he or she will end up with?

 

Sometimes the discussions are downright intimidating. You feel like you need a whole lot to even think about stepping into this whole business.

[Alexandros Angelopoulos Apostolos]

Then there’s that thing that if you just go there and press “Record” and don’t know much about what you’re doing, it feels like you aren’t really doing anything. Anybody can do that. You need to know what you’re doing to get the image that you want to get. If I do this, this, this, I’ll get that, that that. And I know it. It’s fact. I know it because I know theory and how theory becomes practice. Which seems banal, but I just had to write it down. So in my view, I just think, or it just seems to me, that you need to know a whole lot of stuff about a lot of stuff to know what you can do and how you can achieve it.

[Stuart Brereton]

So in my view, I just think, or it just seems to me, that you need to know a whole lot of stuff about a lot of stuff to know what you can do and how you can achieve it.

There is a base level of knowledge which is required about photography in general, and camera systems in particular, but it's not so exhaustive that a DP needs to know intricate engineering detail.

 

There are many DPs who choose to delve into the detail, but it doesn't necessarily make them 'better'. As David has said, cinematography is as much art as science. You could make an argument that the 'Art' part of the equation was the DP's responsibility, and the science was down to others. Either way, remember this knowledge is not acquired overnight, it's the result of experience, and working through many different situations and problems.

[Bill DiPietra]

So in my view, I just think, or it just seems to me, that you need to know a whole lot of stuff about a lot of stuff to know what you can do and how you can achieve it.

 

I agree, but you don't need to know everything. I think everything can appear rather daunting when you are using only one method of learning.

 

In my case, I never attended a dedicated film program. My major (in both undergraduate & graduate schools) was Media Arts. I've always loved the theoretical aspects of film, but the production classes in both programs were extremely limited. After undergrad, I began reading a lot on my own. The first book I read was Cinematography by Kris Malkiewicz (before David had made his contribution to that book) and I really learned a lot.

 

So I found reading, combined with practical use to be the best of both worlds. When I purchased my ARRI-S - as well as some lights & grip equipment - that's when things really started to click. So I guess I learned a lot on my own. Mind you, you always need to be reading, practicing & learning.

 

And of course, this forum has been an invaluable source of knowledge over the past 16 years...

[Satsuki Murashige]

It's always intimidating to work with more experienced people at first, but there comes a point when you simply have to jump in and start doing it - if that's really what you want to do. This is how we all learned what we know and continue to learn every day. Baby steps.

 

A good analogy would be musicianship. There are a lot of very talented guitar players out there. There are also a lot of technicians who know all the minutia of guitar history, how to adjust and modify the instrument, all the scales, modes, and chords. Imagine a Venn diagram representing these two types of people. There is some overlap in the middle, but less than you might think. One can be an incredible musician and only know a bare handful of chords, like some of the early blues men. One can also be a master luthier and be a mediocre player. Ultimately, if what you want to do is sit in with a band and play, you've got to learn how to play well. The other stuff may be helpful but it can't replace actual musicianship.

[David Mullen ASC]

There are people here on this forum with greater technical knowledge than I have in a particular area, that's not surprising on the Internet, you can always find an expert on any field if you search deep enough.

 

And people who have that knowledge often aren't afraid to express it, which can give you the false impression that you need their level of knowledge before you can shoot anything. You don't, you just need to know who to ask when the time comes... In fact, I'd say that the greater skill is the ability to find information.

 

Plus some us go into technical minutiae just for the fun of it, beyond most practical application.

 

Yes, cinematographers shoot tests in pre-production, few people, if they are wise, shoot on a particular camera for the first time when they are actually on set or location, they demand a chance to get to know the camera and establish a workflow in prep. The test may be very basic however.

[Tom Yanowitz]

I don't really like to separate the artistic side and the technical side for any job that blends both, like directing and DPing, but I'll do it for the sake of the argument.

(And I also don't like saying this guy is better than this one but will do as well)

 

I think all the following statements are true :

 

1/ Someone extremely knowledgeable on the technical stuff might never reach the level of someone else that is "just good enough" as a technician.

2/ The more someone learn about the technique, the better his work will be, no matter how deep he delves into it.

 

 

example for 1/ GreatDP admitted he was far from being an expert on digital cameras. Doesn't matter if some averageDP is, he probably wont shoot pictures with the same overall quality as GreatDP , cause other things like composition and lighting are far more important.

example for 2/ However, GreatDP suddenly passionate about digital imaging will shoot better pictures than before.

 

If someone decides to and has the ability to master "both", you end up with Stanley Kubrick. This guy has wayyyy more tools at his disposal to tell his story precisely how he wants, than, say an actor recently turned director, who will rely extensively on his collaborators.

[David Mullen ASC]

But there's a difference between technique and technology -- a cinematographer can get better at his craft through practice like any other artist, but to master his craft he doesn't necessarily have to keep delving deeper and deeper into the technology that he uses, to the point where he can, for example, actually design a camera sensor. Kubrick did not know everything about film technology either; I doubt he could run an FCP print processing line or design a new color emulsion for Kodak. Kubrick was smart enough to hire experts, that is, he knew that he didn't know everything. He also once (jokingly) said to John Boorman that a director could learn everything he needed to know to make a movie in a week. He understood the limits of technique, that at some point taste, aesthetics, knowing what you want to say, what story to tell, become more important than technique.

 

But again, there is a craft aspect to this and you need to know your tools, but only to a degree that is practical and applicable to achieving your artistic ideas -- you can go further into the technology out of intellectual curiosity but it's not going to improve your cinematography.

 

I don't think a cinematographer who is already great wil develop a passion for technology suddenly that will make even better, though of course keeping up with new tools might prolong his career. But "passion" is not the right word here, most visual artists as they mature develop less passion for technology, not more, because they have a solid technical base already and what they are searching for are new artist ideas or they want to refine existing ones.

[Anatole Sloan]

Tom - It seems to me that there are a few issues which you haven't considered.

 

First, given that a colour grading process includes contrast adjustment, often by means of a Log C to rec709 output LUT, the contrast in the original log footage isn't really important. ISO 200 is higher contrast if your final delivery is in log (heaven forbid), but it is perhaps lower contrast after a transform / LUT.

 

What perhaps matter more is the fact that there may be fewer tonal gradations assigned to shadows than to highlights due to this log curve, and thus under exposing can create muddy images.

 

This leads me to the second point. This should be pretty evident anyhow when monitoring with a LUT. Additionally, we have a 12-bit codec which has far more tonal gradations than an 8-bit or 10-bit codec, so the issue is less important with the Alexa. The bias towards highlight gradations is something we have in our own vision as well; it's preferable from that perspective.

 

Third, the key situation where this analysis makes a differences is where you have a low key scene where you are exposing just the bottom half of the waveform. I have to say, personally I would tend to overexpose in such situations anyhow, to get a better spread across the waveform.

 

I think a lot of cinematographers do that. This is in effect doing exactly as you're suggesting - rating the camera at 200 or 400, without touching the menu EI settings. So in effect, DPs already do what you're suggesting - you're just presenting that same, relatively intuitive process in complex lingo.

Edited June 10, 2016 by Anatole Sloan

[David Mullen ASC]

I don't think many cinematographers are using the Alexa at 200 ISO, certainly not for anything other than day exteriors. 400 ISO is not uncommon though for people shooting outdoors or for interiors with a decent light level, or for commercial work, or for chroma key work.

 

I've written about this before, but the problem with ETTR is that it ignores the reality of narrative production and the fact the most shoots do not color-correct dailies and even fewer will do live color-correction on set. Plus there is a principle in narrative cinematography of continuity within a scene, which is why you tend to want a consistent noise level from shot to shot within a sequence rather than have some shots cleaner and other shots noisier. So you can't approach exposure like with still photography where every shot is exposed optimally for itself isolated from every other set-up in a scene. For one thing, unless you can color-correct live on the set, you'd have to explain why every shot doesn't look exposed correctly for the look desired and why shots don't match each other even when they contain the same actor, same action, under the same lighting, but the exposure changes on every shot size because of what shadow or highlight information is contained in the frame.

 

So while you could certain rate the Alexa at 200 ISO, the reduction in highlight latitude will mean closer monitoring on a waveform and adjusting every set-up in order to avoid clipping, whereas at 800 ISO, you have a decent overexposure range that you can concentrate on how you want to expose the overall scene and the actor in it for the mood you want. Also, many of us shoot very quickly with multiple cameras where we don't want to have to worry so much about clipping. Plus that brings up another continuity issue, which is when you have two cameras pointed at the same thing but one is wider than the other, you'd generally expose them the same so the two angles intercut, you don't expose them individually based on the highlight and shadow detail in both.

 

I say "generally" because, yes, you do sometimes have to tweak the exposure to hold something on the extreme edges where you want more information to play with in post, but the general principle is to expose for how you want the scene to look, not to expose each shot with maximum exposure possible just below the clipping point unless for some visual effects reason.

[David Mullen ASC]

Keep in mind that if the cinematographer is very good at lighting and exposure so that there is minimal shot-to-shot adjustment in post, then the noise levels should remain consistent for the ISO level they have chosen to work at. It's just like in film when someone like Gordon Willis could consistently underexpose a movie so that it printed at the same printer lights throughout, so the grain level was where he wanted it to be with minimal variation.

 

So if a cinematographer chooses 1600 ISO on the Alexa for that look, based on testing and viewing the images on a large screen, if they are consistent and get the look and colors right in camera, then it doesn't matter if theoretically they would have more latitude at a lower ISO because they don't need the latitude, they aren't stretching anything in post in terms of luminance or color.

[Tyler Purcell]

Good points David. There are so many tricks one needs to learn about making a consistent image through production. That's quite a big challenge on film, with digital and monitoring on set, it's a lot easier.

[Tom Yanowitz]

I agree with you, once again, one most points David, and Anatole.

First I'd like to point out I don't advise changing the aperture for every setup like a madman.

 

One of the reasons you keep mentioning though is the need for a constant noise level.

 

The reason I'm not 100% sold on this specific point is : on the one hand, noise level is almost not noticeable when you give a lot of light to the sensor. On the other hand, DPs seem to often get away with far more important changes from shot to shot. Like going from a wide to a closer shot on an actor : softening the light and/or the lens, adding fill etc...

 

And ignoring this, it feels in the end the rating is a matter of taste : I found that the 7 stops of overexposure at 800 was mostly useless to me, I'm fine with 5, and that the shadow stops were not as good as I wanted them (and I also prefer having the least possible noise), while most DPs want to be safe highlight wise and have the supposedly film-like level of noise.

 

Anyway, I don't know why I keep rambling on about this like a grandpa, my studies are over and I probably won't see an Alexa on a set for the next 5 years.