Arri's New Camera

  • Thread starter Thread starter Guest
  • Start date Start date
combatentropy said:
But isn't highlight roll-off a grading thing? Aren't all electronic sensors natively linear?

The Arri does this part of the grading in camera, but could you not do the same thing on computer later, as long as you recorded raw or lightly compressed?

And I keep hearing dual gain as if it were exclusive to Arri. But I remember researching it and finding several sensors that did it. I can't remember all of them, but I remember that at least the Blackmagic Pocket had it (yes, the $1,000 HD model).
Click to expand...

Yes this is true. Fairchild which makes all of the blackmagic sensors other than the Pocket 4k and 6k, does dual gain. The URSA mini 4.6k which can be had used for around 2k now uses dual gain. That said the sensor is still much lower dynamic range than an ARRI. Still great cameras though.

Otherwise yeah as long as you don't mind grading, any camera can emulate Arri's roll off. Its the dynamic range that really makes it shine not the roll off though IMHO. Hard to clip an Alexa. I think a lot of newbies grade log without adding in any roll off which ends up giving you a piss poor looking image. A lot of default REC709 camera manufacturer luts also don't have great roll off. LUTs like this show that any camera can get nice roll off though.

https://www.emotivecolor.com/

Arri's color makes it hard to get a bad image out of it which is the other strong point.
 
combatentropy said:
But isn't highlight roll-off a grading thing? Aren't all electronic sensors natively linear?

The Arri does this part of the grading in camera, but could you not do the same thing on computer later, as long as you recorded raw or lightly compressed?

And I keep hearing dual gain as if it were exclusive to Arri. But I remember researching it and finding several sensors that did it. I can't remember all of them, but I remember that at least the Blackmagic Pocket had it (yes, the $1,000 HD model).
Click to expand...

It's a mix of things, in addition to their dual-gain architecture (which is NOT the same as the dual-native-ISO we're now seeing on most new cameras) which gives them such an edge in highlight handling, Arri do a particularly good job of rolling off the saturation of all three colour channels into white as they clip out - this massively helps hide the seams when a digital sensor clips out.

The only other camera that sports the dual-gain architecture of the Arri is the upcoming C300iii (which interestingly, appears to have used the technology to improve their shadows, rather than their highlights).

From the few competent preliminary tests that have been posted so far. I suspect the C300iii will be able to be rated a stop under (so 800 ISO becomes 1600 ISO) to gain an extra stop in the highlights (without the shadows giving way). Which will make it the first camera other than Sony's much more expensive Venice, that can keep pace with Arri's highlights.

Ideally, Canon will develop a new CLOG4 curve to push middle grey down a stop on their new DGO sensors (and probably the C500ii since it's high-res makes the shadows pretty good). So that the extra stop of highlight range can be accessed without any fuss.
 
Grug said:
which interestingly, appears to have used the technology to improve their shadows, rather than their highlights).
.
Click to expand...

How is there a difference?

Have a lut that forces you to stop down and you have robust highlights. Differen LUT that encourages you to open up and you have clean shadows. All with the same base image. Im not even talking a user LUT as every camera has some form of internal lut to turn the 0s and 1s into a displayed image.
 
morgan_moore said:
How is there a difference?

Have a lut that forces you to stop down and you have robust highlights. Differen LUT that encourages you to open up and you have clean shadows. All with the same base image. Im not even talking a user LUT as every camera has some form of internal lut to turn the 0s and 1s into a displayed image.
Click to expand...

I don't know if this is relevant, but the bottom few stops of the C300mk3 seem to be overlapping the noise floor significantly. Which I think would either condemn them to staying in the shadows (or even being crushed), or else you have to be happy with some noise in your image or a lot of noise reduction.
 
morgan_moore said:
I just dont buy 'space in the highlights' as a camera feature. Close down the iris and you have space in the highlights. End of. Yes cameras have wider DR than other cameras.. but space in the highlights is not a thing - its an operator choice. (which may be driven by LUTs)
Click to expand...

No. Not even close. This goes way beyond LUT's or closing down the iris. It's not just about shifting up or down where you place middle grey. I am wondering if you have ever worked with an Alexa or an Amira. It sounds like you haven't. There is literally far more information captured in the highlights which you cannot even see on an Arri cam. But if you decide to pull it in, it is there, and it is often astounding how "normal" it appears - as opposed to the flat grayish kind of mess you might normally find from most cams, (with a ton of color distortions to boot).

On a certain level from an analog point of view, dual gain would be the equivalent of flashing film before exposing. The point is to be able to capture more info in the highlights as a result of bringing down that portion of the curve. However, even that would appear as flat and grey in the old days. But with an Alexa all that invisibly captured information in the highlights can have virtually normal contrast and separation if you choose to pull it back in. I have never, ever seen that kind of tolerance which leads to virtually the same looking image whether it is 3-4 stops underexposed, or 3-4 stops overexposed as from an Alexa. The same image. No other camera touches it. What is astounding with the Venice though, is that they don't use dual gain, but the camera definitely holds good separation in terms of DR and highlights far longer than most cams and begins even so to approach Arri territory without dual gain. That is testimony to Sony's sensor capacities. One can only imagine if they used Arri's architecture with that kind of modern sensor...

Which leads me to my next point. That is precisely what Arri is doing right now.
 
Last edited:
I honestly think there is a bit of conflation going on in the thread, regarding highlight information.

There are a couple of different variables manufactures have to work with. Obviously, the end result is the sum of the parts. I don't think there is any synergy going on, because everything rests on a physical reality and mathematics, let's say. But it's clear to me, and this is more of an opinion, that different manufactures are guided by different sets of rules or ideals. So even if they are manipulating the same variables in the image chain, their decisions will be different along the way.

As has been pointed out, light capture is linear on the sensor level: photons hit the the light sensitive diode and ups the charge. This response is linear.

Already here, depending on sensor architecture and components, as a camera manufacturer you will notice that everyone doesn't start with the same data. But once we're past this stage, I think this is where sleeves are being rolled up and hands are spit.

Now the time has come to take linear information and turn it into something useful. A manufacturer will use their own Color Science and OETFs (log curves) to eek out what they feel is possible. Some will be conservative and some more aggressive. The stronger the desire to post a large DR number–in relation to the raw data, the more likely you'll end up with 'magenta this', 'neon clipped that' in post.

I think we're seeing a field where some players are over reaching because they simply don't have the data budget to spend in the first place, while others—due to technology or different strategies (diode size/resolution)—can afford to 'go easy'.

Ideally, you want to start with a really robust SNR in capture and then let conservative, sound principles guide your software development team so that you end up with a strong, clean image.

RED was probably one of the early offenders regarding official DR and they have always had a very forgiving and on cheering community. But it let to a race that hasn't been only positive.

Nowadays RED has several cameras with more than good enough DR to produce beautiful images and an efficient codec and easy to work with color science.

Anyway: you can always save any highlight with any camera by just stopping down or by filtering, but then you pay at the shadow end because everything comes back to linear DR and signal quality. I think there is some truth to manufacturer log curves and recommendations, or community shooting style, leading users to expose in a certain way perhaps. But regardless of that, you can't escape the underlying physical reality.
 
In my seeing experience the human eye completely blows away the dynamic range capabilities of the best cameras, which means cameras still have a lot of catching up to do which will likely take awhile to achieve.
 
yoclay said:
On a certain level from an analog point of view, dual gain would be the equivalent of flashing film before exposing.
Click to expand...

More like exposure bracketing, because flashing changes the contrast.

yoclay said:
Which leads me to my next point. That is precisely what Arri is doing right now.
Click to expand...


Arri have always understated their DR, the only company to do so.

They also have an unusually large pixel pitch. Everyone else is rushing to pack in more pixel density and Arri are the exact opposite. Which is also why they haven't had a 4K friendly camera in S35.

And dual gain.

And some pretty cool stuff to do with noise processing.

So funny when people bemoan a 2 year old sensor and it's clearly, not JUST the sensor...

https://www.arri.com/en/camera-systems/cameras/alexa-10th-anniversary

JB
 
Arri Log C White Paper said:
The Log C curve actually is a set of curves for different exposure indices.
Logarithmic encoding means that the relation between exposure measured in stops and the encoded signal
is linear (straight) over a wide range. Each stop of exposure increases the signal by the same amount.
Click to expand...

The Log C encoding has a linear exposure, grayscale characteristic similar to a scan from negative film. There is no highlight rolloff in Log C except at 3200. Below 1600, maximum exposure does not fill the 10 bit encoded bucket at the clipping point. At 200, exposure clips at 858/1024. It is exactly the same process for BMD Film gamma and Sony S-Log3 except the latter is a single EI curve.

morgan_moore said:
I just dont buy 'space in the highlights' as a camera feature. Close down the iris and you have space in the highlights. End of. Yes cameras have wider DR than other cameras.. but space in the highlights is not a thing - its an operator choice.
Click to expand...

An opinion with support in Arri's Log C white paper.

https://www.arri.com/resource/blob/...483e/2017-03-alexa-logc-curve-in-vfx-data.pdf
 
John Brawley said:
https://www.arri.com/en/camera-systems/cameras/alexa-10th-anniversary

JB
Click to expand...

Amusingly, big-name DPs quoted in that Arri press release imply disagreement with dvxuser opinions on highlight rolloff and large sensor advantages. (I'm not saying I agree with these DPs.)

Roger Deakins: "I was also drawn to the subtle fall-off to highlights" of the Alexa.

On the Alex 65, Greig Fraser ASC, ACS, commented: “I looked in the lens port at the sensor and was just completely gobsmacked. It was one of those moments where the dominoes instantly start falling in front of you; I remember thinking, ‘Right, here we go—this is the beginning of a new era in cinematography.’”

Remi Adefarasin OBE, BSC used the ALEXA LF on “The Last Vermeer,” commenting: “Large format isn’t just about resolution, it’s a different relationship between the audience and the image. With the ALEXA LF we’re able to capture all the beauty of longer focal lengths, but with a wider field of view. This in turn lends itself to capturing images of truly cinematic quality.”
 
Nothing really all that surprising. For years I've skimmed threads like these on message boards and am fascinated at the apparent technical depth of the discussion. If you tried to engage the majority of working DP's over some of these details, their eyes would glaze over (up to a point I'm admittedly one of those). Perhaps the point being, the visceral reaction one has to the results from a piece of filmmaking equipment (camera, lens etc) may not correlate with how it performs on paper. For instance, I have experienced some real-world issues relating to skin tones while working in a wider color gamut, resulting in green or magenta tones showing up that I have to work harder to eliminate. More accurate rendition doesn't always equate to a more pleasing image.
 
yoclay said:
No. Not even close. This goes way beyond LUT's or closing down the iris. It's not just about shifting up or down where you place middle grey. I am wondering if you have ever worked with an Alexa or an Amira. It sounds like you haven't. There is literally far more information captured in the highlights which you cannot even see on an Arri cam. But if you decide to pull it in, it is there, and it is often astounding how "normal" it appears - as opposed to the flat grayish kind of mess you might normally find from most cams, (with a ton of color distortions to boot).

On a certain level from an analog point of view, dual gain would be the equivalent of flashing film before exposing. The point is to be able to capture more info in the highlights as a result of bringing down that portion of the curve. However, even that would appear as flat and grey in the old days. But with an Alexa all that invisibly captured information in the highlights can have virtually normal contrast and separation if you choose to pull it back in. I have never, ever seen that kind of tolerance which leads to virtually the same looking image whether it is 3-4 stops underexposed, or 3-4 stops overexposed as from an Alexa. The same image. No other camera touches it. What is astounding with the Venice though, is that they don't use dual gain, but the camera definitely holds good separation in terms of DR and highlights far longer than most cams and begins even so to approach Arri territory without dual gain. That is testimony to Sony's sensor capacities. One can only imagine if they used Arri's architecture with that kind of modern sensor...

Which leads me to my next point. That is precisely what Arri is doing right now.
Click to expand...

Ive only gripped Arris, not DPd them although I have played.

Im talking about the no lut lut.. the base in the camera that turns numbers to some sort of image.

Ive plenty of experience of big pixel low density pro DSLRs (D3 1dx) and my ol 16bit Sinar and even Kodak DCS to have a good concept of the flex in the grade and recovery. Not to mention a decade of film processing printing and scanning :)

I think beyond moving the grey point and LUT choice there are still elements of highlight handling to be done well. ie all colours should clip at the same time. My first sonys were horror stories of blue or yellow rings around specular highlights cause by inconsistent clippiing. Arri has that nailed too :)

And the allocation of bits per stop. You will be aware of the huge amount of levels per stop(in linear) at the top end of 13 stops of 12,14 or 16bit, and how the majority of this data must be thrown to avoid monster files without a payback worth the data. Doing that well can also make a camera excel.

Im convinced this is where my fs7 falls over, one is not getting enough bits per stop at the high end. Partly why I changed from Slog3 to Slog2 , one gets less DR but the file is no so thin per stop.
 
Piszkos Fred said:
So, the new s35 4K camera will be about the same size as the Mini LF, same viewfinder and capture media.

https://youtu.be/gUZLP-1YPEM?t=2081
Click to expand...

I was listening to an Arri video today, and they were saying that the Mini LF frame rates were reduced because of heat. The body was too small to adequately dissipate the heat for higher frame rates.

Most likely they won't have that issue with their newer tech.

I am interested in:
- if they have an instant on or will they have more modular hard ware that requires booting up.
- Is battery life improved, and will it have an onboard mount? The Red Ranger is actually smaller than the Alexa mini's in part because of this.
- iso3200 good? 6400 usable?
- improved dynamic range? I know the Alexa was good, but would still be nice to have a little more feather in the roll offs
- integration with modern cine tools and systems

I'd assume a lot of it is going to be rather straight forward.
 
ahalpert said:
Amusingly, big-name DPs quoted in that Arri press release imply disagreement with dvxuser opinions on highlight rolloff and large sensor advantages.
Click to expand...

I'm not sure about the disagreement. Well, I guess it kind of depends on what part of the thread you mean... it's 24 pages after all... =)

CharlesPapert said:
Nothing really all that surprising. For years I've skimmed threads like these on message boards and am fascinated at the apparent technical depth of the discussion. If you tried to engage the majority of working DP's over some of these details, their eyes would glaze over.
Click to expand...

I agree.

But I don't think that "not knowing something" or "not being interested in something" is a strength—and to be clear: you're not saying it is. A top end DP doesn't have to know the technical details of a camera system. It won't harm them if they do, but it simply isn't necessary. There is a reason that DIT is job title today. Then again, depending on Director/DoP team, the influence of the DIT can vary a lot.

I think the reason for all the technical forum discussions is simple: a lot of people here are buying gear themselves. Typically, even though I don't feel this way myself, people love to compare specs. And to compare properly you have to know something about what the specs mean. And the the discussions start and the rabbit hole goes deep....
Plus, everyone knows what gear is being used on professional sets. And we know what that gear costs. So the question becomes: "what is the difference between expensive camera x and cheap camera y?" And the discussions, often technical, continue.

If you're a good enough DP you just ask for your established, favourite high end cinema package. What's to discuss about some technicality? If there was something better around, you would surely hear about it from colleagues or maybe directly from the manufacturer in some cases.

It's probably good if they can verbalise how they interpret images though. I'm pretty sure Deakins would be a reasonable guy to talk to. And then we have DoPs like Yedlin who seem pretty pedantic with a strong sense of intent (not saying this is bad). But I'm also kinda sure it wouldn't be all that hard to find some DoP who was put in front of the camera in some marketing segment for some gear, and you realise their technical prowess isn't exactly rock solid. Especially if they came up on film and didn't quite make the digital transition their mission in life—and they didn't have to, since they always had support on hand whose job it was to bridge the gap.
 
combatentropy said:
For me, it's 6 pages :)
Click to expand...

Very interesting...

...because I did the same thing on reduser.net a long time ago. I don't know why I haven't done it here. I think it has to do with the style people post and the typical page count. But now that you've brought it up I will change it here too. I also generally prefer fewer pages and less reloading.
 
If you're a good enough DP you just ask for your established, favourite high end cinema package. What's to discuss about some technicality? If there was something better around, you would surely hear about it from colleagues or maybe directly from the manufacturer in some cases.
Click to expand...

Everyone does it differently. I do learn about new gear from colleagues and manufacturers, sure, but also one of the reasons I hang out around here is to learn about things from this angle. As I said though, I do find some of the discussions here drive so deep into the technical that I'm fairly sure they would go over the head of not only most DP's I know, probably even most of the DIT's also.


I'm pretty sure Deakins would be a reasonable guy to talk to.
Click to expand...

He is! Very straightforward in approach on pretty much everything. The last movie I did with him was back around 2007, and we used to sit at lunch and discuss techniques from previous films like the bleach bypass process in Jarhead (often punctuated with "Andy, what did we do on that one?")! Having just worked with the RED One for the first time, I gave him a briefing on my experiences and even though this was a good four years before he engaged in digital acquisition, he was quite interested.

But I'm also kinda sure it wouldn't be all that hard to find some DoP who was put in front of the camera in some marketing segment for some gear, and you realise their technical prowess isn't exactly rock solid. Especially if they came up on film and didn't quite make the digital transition their mission in life—and they didn't have to, since they always had support on hand whose job it was to bridge the gap.
Click to expand...

This is essentially my point. I have a number of colleagues who think of me as being more "techy" than them, yet I can't keep up with the deeper pixel-peeping type conversations here...but even without a DIT on set, I still find my way!
 
If the new sensor surpasses the noise performance of the older large photosite design, it will be interesting to see how Arri eventually incorporates the new tech into LF and the Alexa 65. Will they go with the smaller photosites or use larger versions for even better low light performance in the large format cameras.
 
You must log in or register to reply here.
Back
Top