Jump to content

How the Alexa really works


Tom Yanowitz

Recommended Posts

Hi guys!

 

After some research, I made a little chart one every rich man's favorite camera, the Alexa, which digs into how it really works. So I thought I might as well share it.

The content holds true for every Alexa I believe, since they all have pretty much the same sensors and algorithms (even the Alexa 65 I assume, who consists of three Alexa sensors merged).

 

On the right of the table are the only available curves to my knowledge, the LogC ones.

Every other curves isn't, but should be available, so here they are.

 

Where did I find this stuff ?

A combination of gathering bits of informations here and there, and the ability to think like an Arri R&D guy I guess (they always seem to make the most logical choices in sensor architecture).

 

CLICK and enjoy for zoomable chart.

 

VKaX0e.jpg

Link to comment
Share on other sites

By the way if someone finds something that is or might be inaccurate, please tell me and I'll make an update.

 

Why do you think Arri doesn't share more that kind of infos ?

I think either they think we don't care about the in-depth stuff, or because it makes EI 800 looks bad.

 

They have this popular "dynamic range" image where we see how you get +5/-9 stops at 200, +7/-7 at 800 etc etc.

This is sort of true, but they forget to mention that every stop isn't created equal, in terms of how many values each has going for him.

A highlight stop might have hundreds of values for it (arriraw), where a low-light stop ony has a couple.

 

Out of curiosity, I tried to see at what ISO would a middle gray have as many values below and above him : it's ISO 64.

And in terms of stops, it depends where you decided that the dynamic range ends.

The Arriraw has a high-contrast part of the curve that is 8 stops long so if you want high contrast low lights in your file, you'd go for ISO 64 or 80 with a +4/-4 middle gray.

 

 

By the way if Art Adams passes by on the topic : Thank you, I learned a lot with your Alexa articles back in the day, and I hope the stuff I shared will be useful to you as well.

Edited by Tom Yanowitz
Link to comment
Share on other sites

  • Premium Member

This does remind me of some of the stuff that's been tried by the Magic Lantern people on things like the 5D. It was discovered that there are two readout amplifiers on the sensor of that camera, and that with a bit of trickery they could be set to different gains. These two amplifiers read out alternate lines, so that the net result is alternate lines of the image with different exposures.

 

Some work was done on this with the idea of interpolating the alternate pairs of lines out, so that it would approximate the Alexa dual-gain approach, but between the difficulties of the interpolation and the preexisting issues with the 5D Mk. II, it wasn't that successful. That's not really the point, though: what all of this implies is that the limits on at least some modern sensors aren't actually full well capacity, they're amplifier dynamic range and bandwidth.

 

P

Link to comment
Share on other sites

(you mean that the 5D has this with odd and even lines instead of for each pixel ?)

 

Well, Full Well Capacity is still more important than these electrical trickeries I think, as it influences both sensitivity and dynamic range.

Link to comment
Share on other sites

  • Premium Member

Yes, but it doesn't matter if you don't have amplifiers with enough bandwidth and dynamic range (simultanously, of course) to offload that data with adequate noise performance.

 

The 5D experiments suggest that, at least at the level of a (when new) £2k DSLR, compromises are being made.

 

P

Link to comment
Share on other sites

  • Premium Member

The trouble with low ISO ratings on the Alexa is that the loss of extended overexposure detail when combined with the hyper clean image tends to make everything feel more digital, and one reason people like shooting with the Alexa is how much Log-C at 800 ISO is similar to a Cineon log scan of film, even if it means a little bit of noise (and it's still less noise than 500 ISO film stock anyway.)

Link to comment
Share on other sites

The trouble with low ISO ratings on the Alexa is that the loss of extended overexposure detail when combined with the hyper clean image tends to make everything feel more digital, and one reason people like shooting with the Alexa is how much Log-C at 800 ISO is similar to a Cineon log scan of film, even if it means a little bit of noise (and it's still less noise than 500 ISO film stock anyway.)

 

Sure there are reasons both for and against working at ISO 100 or 800 or 1600.

What I'm suggesting is this :

 

- be aware that by working at ISO 800 you essentially are placing you midtones in the tow of the arriraw curve.

If you don't view it as dark on a monitor, it's because it's pushed by a couple of stops by the "LogC LUT"

- the lower the ISO you work at, the cleanest and highest contrast the image you'll get.

 

Why working at a higher ISO ?

- you want low-contrast shadows and noise.

- you don't have enough light available and you are at a wide aperture already.

- deep focus is more important than noise and IQ

- you have some extreme highlights in the frame (practicals, lamp shades you want detailed) AND they are the keylight as well so you can't lower their output. (hello Birdman)

(so highlights to mid gray contrast ratio is above 5.5 stops)

(Actually does any knows the contrast ratio between a bright white cloud and a backlit actor's face ?)

 

So essentially it works exactly the same as when you choose your own ISO/EI with a given film stock.

 

 

So overall I'm more for the dynamic use of ISO than the static use.

Cause if for example your actor's face is one of the brightest area of the shot and you're working at ISO 800, the amount of info you'll lose uselessly is crazy.

You will record the whole contrast on a lot less gray values than you'd otherwise have with a lower ISO.

(By the way every time I talk about working at a given ISO, I'm implying setting the f-stop to what the light meter with that ISO tells you too, not the camera's ISO settings, which doesn't do anything to the actual file, just switches monitoring LUTs if you record raw)

 

Actually ideally you can forget about working with ISO all together and use the ETTR technique but that's an other story.

Edited by Tom Yanowitz
Link to comment
Share on other sites

I will say this about the Alexa, its low light shooting capability is superb! What I was able to do on both Against The Wild films with fire light only was astounding. And not even a hint of noise in the picture. If I had tried the same thing with 35mm I would of needed extra light and I know I would of been applying lots of noise reduction in post. Even using the very best 500ASA from Kodak or Fuji.

 

Reliability is another factor.....hundreds of stories out there about REDS overheating and dying on set, endless stories. Not so the Alexa. On two wilderness shoots, one in Canada and one in Africa, that camera didn't even utter a peep. It was flawless everyday all day.

 

Its a serious piece of gear and was well thought out by a company that knows cameras. If you are in the market, I can recommend the Alexa.

 

R,

Link to comment
Share on other sites

  • Premium Member

Tom, you're thinking like an engineer -- what matters is how it looks on the big screen, and you can't tell me that all these Alexa movies shot at 800 ISO by Roger Deakins and Emmanuel Lubezki look bad because they are "losing too much information".

 

Also, consistency of exposure and noise within every shot in a sequence matters in narrative fiction, whether or not you are placing information in the optimal part of the curve. You don't expose shots in a scene like you would in still photography where you optimize the exposure and ISO setting for that single image.

  • Upvote 1
Link to comment
Share on other sites

I am actually convinced they wouldn't keep shooting everything at a high ISO if they knew all that.

 

It's not about how bad Chivo and Deakins images look at high ISO (they do look great), it's about how different (better or worse) they'd be if they were using ISO dynamically.

Their images might be noticeably or un-noticeably better, but at least equal and never worse.

 

Lubezki would be the one benefiting the most from it because of his extensive work in post for roto work and power windows. Most of his shots are heavily graded.

And he said they took the Alexa over film cameras for The Revenant because they wanted the least possible noise.

 

For Birdman they might have often fallen in the "no enough light output" reason for high (1280) ISO.

 

The most curious thing is : most DPs knew about all this in film days (like the film stock says its 500 but I'm going to work at 250 for deeper shadows and less grain for example)

 

Some other manufacturers (like Red and Blackmagic) are talking about how you shouldn't underexpose the sensor, but Arri doesn't, either by fear of DPs screwing up and ranting about their clipped footage, or to keep the idea that they are natively ISO 800.

Edited by Tom Yanowitz
Link to comment
Share on other sites

  • Premium Member

Sure but they were relying on film's extended overexposure capability when they did that and some were doing it in order to get better blacks due to higher printer lights, plus grain at 500 ISO is higher than the noise of the Alexa at 800 ISO so there is less desire or need to overexpose just to reduce noise. Plus you are assuming that getting the cleanest signal possible is always a priority over extended detail in bright areas. You are also making the false assumption that Deakins, Lubezki, and their DIT's don't already know these issues. And finally, if you have a style that incorporates a lot of available light then you are not going to start out picking a low ISO setting and lighting from there.

  • Upvote 1
Link to comment
Share on other sites

  • Premium Member

Blackmagic and Red cameras do have more noise problems that crop up due to underexposure, hence their warnings. The Red camera in particular has smaller photosites in order to achieve higher pixel counts so it is not surprising that it is a bit less sensitive.

 

None of this is new, Tom -- most of us understand the trade-offs between noise and clipping, etc. I understand why Art Adams rates the Alexa at 400 ISO and I've shot plenty of Alexa footage over the years at different ISO's to know how it will look. But this isn't a technical exercise, it isn't a science experiment, we are creating artistic images and one person's image priorities are not another's. If some cinematographer likes what he is getting at 800 ISO, then more power to him, I'm not going to judge anymore than when Harris Savides would underexpose a movie by three stops.

Link to comment
Share on other sites

I'd agree with David. I shoot with Alexa all the time, and I use 800 ISO when I have highlights I need to control, and at 400 ISO otherwise. 200 ISO and lower just doesn't look as good, regardless of whether it is technically 'better'.

 

This isn't some new camera that needs a critique, it's an industry standard that just about everyone is very happy with.

Link to comment
Share on other sites

Sorry Bill, doesn't you browser automatically fit it to the width of your screen ?

 

Let me re-post a crop of the chart with the most important graphic

 

2DyxOU.jpg

 

 

I actually don't care that much about what ISO this or that DP uses, as long as they understand the ins and out, and that each ISO has clearly defined advantages.

Just don't call 800 the "base ISO" unless you're a marketing guy at Arri, there's no "base ISO" as long as the people at ISO don't make a norm for raw capture.

 

You seem to be very afraid of white clipping, as everyone should, but black clipping is actually the more common problem isn't it. With footage having no details whatsoever in the shadows.

It's a problem we see over and over.

An other huge reason for overly underexposed footage (I'm talking to the point where the DIT or DP are fired once people see the dailies) is trusting the monitor for exposure.

 

And don't get me started on underexposing the raw footage to add noise. That's just nostalgia. If people are so fond of film grain, they could try and experiment with different types of noise in post instead of degrading in capture.

 

Wanting to be an "artist of light and darkness" shouldn't make one skip the nitty gritty about how digital cameras work (like, for example, that half of a sensor's values are dedicated to the last stop of highlights before clipping).

 

You guys trust your eye way too much. I did some experiment showing footage at various ISO, but lying about which was which and guess what, what people want to see based on their limited knowledge of the camera takes over what they actually see.

 

So please tell me in what way the Alexa looks worse at 200.

Edited by Tom Yanowitz
Link to comment
Share on other sites

  • Premium Member

 

 

there's no "base ISO"

 

This is of course completely true.

 

The effective sensitivity of digital (and indeed photochemical, to some extent) cameras is based on a compromise between dynamic range and noise and is thus a matter of opinion.

 

What's interesting about the Alexa approach is that it emphasies that simply recording the light on the surface of the sensor is one thing; offloading it elsewhere is another, and the best current designs need to be careful about both.

 

P

  • Upvote 1
Link to comment
Share on other sites

If people are so fond of film grain, they could try and experiment with different types of noise in post instead of degrading in capture.

 

The best way to obtain film grain is to shoot film.

The resulting image is completely different from adding noise in post, or otherwise exploiting the noise floor of digital sensors.

Noise is not degradation. Finite bit depth is degradation and one which is physically alleviated by adding noise in advance of quantisation.

This is not nostalgia unless a technically better signal is nostalgia.

 

But what you are otherwise suggesting is correct.

 

However I wouldn't hold this against cinematographers. They are trying to use digital technology that's framed in terms that don't actually belong to what digital technology is actually doing.

It's marketed as magic, and so cinematographers will treat it like such. As magic. After all it's supposed to be ultra convenient in that way, ie. where you don't need to use one's slide ruler.

Where you only need your creative eye/brain switched on.

 

In reality, of course, one can squeeze a lot more out of anything if one does use one's technical brain, or otherwise have an assistant (and corresponding post operator) handle that side of things.

 

C

Edited by Carl Looper
Link to comment
Share on other sites

"I actually don't care that much about what ISO this or that DP uses, as long as they understand the ins and out, and that each ISO has clearly defined advantages."

 

"Wanting to be an "artist of light and darkness" shouldn't make one skip the nitty gritty about how digital cameras work (like, for example, that half of a sensor's values are dedicated to the last stop of highlights before clipping)."

 

​Do you not think this comes over as a bit pompous though Tom.. from your Id pic you seem to be a youngish guy.. where do you get the idea that the average DP doesn't know about shifting the grey level .. let alone Deakins/Chivo et el.. is there something in the water students drink these days :)..

Link to comment
Share on other sites

​Do you not think this comes over as a bit pompous though Tom

 

It was intended, but why focus on the way I talk instead of the actual content of the charts. (and the light and darkness part was me making fun of anti-technique DPs).

 

you seem to be a youngish guy

 

What does it have to do with anything ?

 

where do you get the idea that the average DP doesn't know about shifting the grey level .. let alone Deakins/Chivo etc

 

Cause I'm a huge fan of these guys and have seen every single interview or appearance they made on video, audio, magazines articles, internet articles and (for Deakins) forum.

So I have a reasonable estimate of what they know about exposure in the digital age.

What is great about a good DP is the way he lights, the way he frames, the choice of focal lengths and the camera movements.

Exposure is just a tiny detail in all that, so you don't need to be an expert to get a great image (especially if you have DITs following you)

That said, being an expert on digital exposure will give you that tiny extra quality as a DP : slightly faster, slightly more confident and daring while still being in control, and perhaps give slightly better images (maybe 0.1% better, maybe 20% better).

 

 

is there something in the water students drink these days :)..

 

triuqs s'rehtom ruoy

Edited by Tom Yanowitz
Link to comment
Share on other sites

Thanks Phil and Carl !

I didn't reply cause I mostly agree with what you said.

 

There is a great beauty in how all these tools for capturing images are designed, crafted, optimized, and it's sad when the people paid to use them are looking down on it.

Edited by Tom Yanowitz
Link to comment
Share on other sites

  • Premium Member

Yeah, although quite often the grey-haired old crumblies can be quite breathtakingly ignorant of how the modern stuff behaves, even when they're hugely experienced and respected. Many very big names rely very heavily on DITs and gaffers. As with anything like this, there's a lot of BS involved. There has to be a give and take on it.

 

P

  • Upvote 1
Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
×
×
  • Create New...