Canon goes 8K60p and "20+ stops DR" in 4K by NAB

I don't think so. Gain does not affect full-well capacity. Neither does it affect the native noise level. Gain amplifies signal and noise together. In fact it is more like a pre-grade, from what I understand it. Either way, even if it did, they would be welcome to give Alexa the best measurement (take the noise at low gain and full well at high gain, or whatever).

By the way, Alexa is not the only camera with dual gain. I have seen it in several specs. For instance, the Blackmagic Pocket had dual-gain read-out. I think any of the sensors that Blackmagic got from Fairchild may have had it. Arri did do an extra-fine job of combining the two images in camera, complete with subtley desaturating the highlights to mimic film (but you could do that in post).
 
drboffa said:
I'm asking this in all seriousness: has any camera—other than the Alexa—actually delivered more than 14 stops of independently verified dynamic range?
Click to expand...

No, not a single one.

And in particular, in 11 years no one's managed to best the Alexa's highlight latitude specifically (which is, frankly, the element of dynamic range that's most important for a smooth, "filmic" rolloff in the highlights). A couple of recent cameras are now (finally) coming pretty close (the Sony Venice and Canon C500ii). But no one's managed to match (let alone best) it yet.

The gains most manufacturers have laid claim to, lie primarily in shadow latitude - but the detail that close to the noise floor is up for a fair bit of debate as to how much of it is truly "usable" (hint: it's never quite as usable, as manufacturers DR claims would have you believe).

A properly exposed Xyla 21-step DR Chart, when combined with a waveform - will paint a plenty clear picture of how much DR you really have on tap. It allows you to clearly see the camera's noise floor, and you can make your own judgements about how far into it, you're willing to use the shadow detail.

But the best, and most useful tests of latitude, remain the classic "Over/Under" test - where you set up a scene (and most importantly an 18% grey card) and then both under and overexpose it, until the grey card clips at both ends of the spectrum (clipping to white, and clipping to black). This will give you a definite count on exactly how much detail you can capture in both overexposed and underexposed situations. So you can know for a certainty, how far is too far, when you're exposing a scene.
 
Really difficult for me to understand why only ARRIs do what they do.

I know there are some patents involved, and there aren't many making the best cameras in the world, and I know we have some great-looking REDs, and you know I love my conspiracy theories, but ARRI's technology is mysterious to me.
 
It isn't that ARRI is the only one but they put a lot of processing power and quality wiring into the signal path. And then they use OLPF to smooth out the details. A lot of people could probably do it but not many can manufacture, then sell an $60,000-$100,000 package.

That said, that Phil Holland 8K clip looked really nice and, at some point, it just doesn't matter that much. All top cams look great and all can be massaged to produce spectacular images.
 
Grug said:
no one's managed to best the Alexa's highlight latitude [...] The gains most manufacturers have laid claim to, lie primarily in shadow latitude[...]
Click to expand...

If you shoot RAW, then I think that you can do in post what Arri does in camera. In RAW there is no concept of highlight or shadow latitude, because you have not yet chosen a midpoint. If you're talking about highlight latitude or shadow latitude, then you are judging a camera by its built-in looks, which I guess for whatever reason often aren't matching Arri's?

But the best, and most useful tests of latitude, remain the classic "Over/Under" test - where you set up a scene (and most importantly an 18% grey card) and then both under and overexpose it, until the grey card clips at both ends of the spectrum (clipping to white, and clipping to black).
Click to expand...

Again, this is a test of the camera's built-in processing. If you don't wish to mess with the footage too much in post, then yeah, this is the test you should do. But to get the camera's full latitude you need to shoot something like RAW or almost uncompressed, and basically grade from scratch. For whatever reason it sounds like many cameras are leaving a lot of potential latitude on the table with their built-in looks.

On a different tack, I don't get the opinion that Arri is so far out there, beyond other cameras. I've seen tests for years where it's indistinguishable, including one with the original Blackmagic Pocket.

---

I guess I should disclaim that I am totally armchairing this, having been out of the business for several years. However I dug into this deeply a few years back, read a lot, watched a lot, etc.
 
Choosing the correct camera from a bunch of cameras is next to impossible; they all can look the same in ideal conditions (usually studios where these tests take place).

It's the over/under (as Grug mentioned) you'll see the tremendous difference the ARRIs have.

And RAW helps in post but the sensor still needs to capture the information for that concept to play out well because by now almost every camera shoots internal or external RAW.
 
NorBro said:
Choosing the correct camera from a bunch of cameras is next to impossible; they all can look the same in ideal conditions (usually studios where these tests take place).
Click to expand...

No, I'm talking about tests under severe conditions.

And RAW helps in post but the sensor still needs to capture the information for that concept to play out well because by now almost every camera shoots internal or external RAW.
Click to expand...

Right, that's what I'm saying. I think that these cameras that say they have as many stops as the Alexa, or more, really do at the sensor level.

---

By the way, I think that scenes greater than 11 or 12 stops are extremely rare, even the ones you think of as extreme conditions.

This is because the range, first of all, of color reflected is only 5 stops. If you put the cleanest white shirt you can find beside the darkest black shirt you can find, and they're both under the same light, the difference is 5 stops.

To generate more stops in the scene, then you need to vary the light. For example, put the white shirt in direct sunlight and the black shirt in shadow. But outdoors under a tree, the light is only 2 or 3 stops darker than being out in the sun. So black shirt under the shade of a tree is a total of 7 or 8 stops darker than white shirt in the sun.

If you mix indoor and outdoor, then yeah, the sky is the limit to how much dynamic range you can get, if you don't light the indoors at all.

---

But I think that the reason the Alexa looks so good isn't the number 14, and that no other camera really has the number 14. It's that the Alexa grades its 14 stops so well, that even when you are shooting a 12 stop scene, it still looks better than all other cameras, whether they happen to have 12 stops or 15 stops.

But then again, that's if the Alexa really is still looking better than all these other cameras. I will say it again: as an end viewer, I don't see it. I don't see this massive improvement of the Alexa over all other cameras. I don't even see a minor one.
 
Last edited:
A lot of people would actually say it's its handling of skintones which separates the camera from others. You're able to push and pull like crazy without changing them much. (But DR and color are also related to that here.)

IDK, but to me the ultimate test would be for someone to shoot a major Hollywood blockbuster with a BM or one of the Japanese cameras.

I just want to see the 1% take a Blackmagic or a Canon or something else and make a major film with it.

Has only happened with the Venice in the last 15 years, with others getting small supporting roles here and there (besides RED). And there are a few other very rare examples but I'm talking a major motion picture.

Why can't anyone just use a C500 Mark II?
 
combatentropy said:
But then again, that's if the Alexa really is still looking better than all these other cameras. I will say it again: as an end viewer, I don't see it. I don't see this massive improvement of the Alexa over all other cameras. I don't even see a minor one.
Click to expand...

Maybe they have a hard time getting other cameras to look like this:

 
NorBro said:
IDK, but to me the ultimate test would be for someone to shoot a major Hollywood blockbuster with a BM or one of the Japanese cameras.

I just want to see the 1% take a Blackmagic or a Canon or something else and make a major film with it.
Click to expand...

I don't know if these fit your criterion of "major", but there are a handful (a small handful) of big non-Arri films:

After Earth (2013; Sony F65 and Canon C500)
Ex Machina (2014; Sony F65 and F55)
Tomorrowland (2015; Sony F65 and F55)

Also The Crown started on an F55 before switching to the Venice ("just" a show but a pretty major one).

That said, it is really difficult to find major films not shot on an Alexa variant.
 
They're major enough. I also forgot about the F35 which did a bunch, but the Alexa was pretty fresh at that time.

Now which ones have a Sony Pictures credit is worth thinking about as well.
 
About a month ago, I saw the Cafe Society - online and in 4K. F65 + Vittorio Storaro (+ Woody Allen at the helm). The look was tolerable.

(I hated the film)
 
combatentropy said:
By the way, I think that scenes greater than 11 or 12 stops are extremely rare, even the ones you think of as extreme conditions.

This is because the range, first of all, of color reflected is only 5 stops. If you put the cleanest white shirt you can find beside the darkest black shirt you can find, and they're both under the same light, the difference is 5 stops.

To generate more stops in the scene, then you need to vary the light. For example, put the white shirt in direct sunlight and the black shirt in shadow. But outdoors under a tree, the light is only 2 or 3 stops darker than being out in the sun. So black shirt under the shade of a tree is a total of 7 or 8 stops darker than white shirt in the sun.

If you mix indoor and outdoor, then yeah, the sky is the limit to how much dynamic range you can get, if you don't light the indoors at all.
Click to expand...

What about incident light sources in frame? Lamps, clouds, sky, the sun?

Another thing missing from this discussion is that you typically lose color information before you use luma information. You see this in the over/under tests - that when you pull the exposure back to normal, the color is totally shifted or missing altogether. But I think that Arri does exceptionally well in this respect. So, in other words, not all "stops" of exposure are created equal
 
Last edited:
Many of you may have already seen this, but the Alexa sensor really shines when it comes to overexposure: https://youtu.be/ccc9-zhGPbo

Even at five stops over you still have a usable image. I don't think any other camera can make that claim (yet).
 
drboffa said:
Many of you may have already seen this, but the Alexa sensor really shines when it comes to overexposure: https://youtu.be/ccc9-zhGPbo

Even at five stops over you still have a usable image. I don't think any other camera can make that claim (yet).
Click to expand...

Yeah the Alexa gets a bit over 5 stops over. It quickly gets noisy underexposed but the color remains pretty good and consistent.

I've found my Panasonic S1 gets close to 5 stops overexposure but not quite there. Probably a stop behind. The color combined with dynamic range is why ARRI is still king.
 
You must log in or register to reply here.
Back
Top