C300: Mk II C log 2 waveform 92 IRE cap

G

GabrielGely

New member
I posted this question on another thread, but thought that the topic calls for its n thread.

Something that I've been trying to understand better.

With the new C-log2, we get the 15 stops of DR. And the way the wavewform reads it, it clips the highlights at 92 IRE. At first glance I get it, and the overall response that I've been hearing around is to continue lighting the same way, just treat 90 as the new 100. But if I think about it in detail I honestly don't fully understand it. What is it doing exactly?
- Is it bumping everything down on the waveform?
- Is it doing some sort of higher compression at the highlights?

For example, if I was shooting a scene with the c300mkI and wanted skin tones to land on 45 IRE and I want my highlights to end at 82 IRE, how would this translate to the mkII? And why?

Any of this making sense?
 
Some has to either know this or be wondering about this. Or at least tell me it's a silly question. I honestly don't know, and I am extremely curious
 
Not exactly sure the science behind it but Slog3 does a similar thing and tops out at 93 IRE. I think it has to do with the curve being engineered to fit more stops of DR through later sensors (16 stops or more)
 
GabrielGely said:
With the new C-log2, we get the 15 stops of DR. And the way the wavewform reads it, it clips the highlights at 92 IRE. At first glance I get it, and the overall response that I've been hearing around is to continue lighting the same way, just treat 90 as the new 100. But if I think about it in detail I honestly don't fully understand it. What is it doing exactly?
Click to expand...

Canon's documentation confirms full well saturation of the sensor to represent IRE 92.17% in Canon Log 2. They then say that they make an "important extension" of 2 stops of the Log curve up to IRE 109.13% Canon says this is to allow for push-pull in the color grading process.

I've been thinking about this too for a while, but feel I'm missing information. But I'm thinking that it might be beneficial when shooting Log2 and adding luts to have some headroom when doing certain color conversions. The C300 mkII seems very flexible in terms of color spaces and profiles it supports.

In DaVinci Resolve, when you push the highlights into clipping in one node, you can still recover them later because Resolve keeps track of the true values in 32bit floating point. I've been wondering if it might be beneficial to a baked in Log curve to offer some headroom above actual clipping to allow for calculations later in the chain, even if still 'in camera'.

Anyway, the way Canon describes it, it sounds more like the idea is flexibility in post.
 
I'm sorry, I'm still lost. Maybe it's me that I'm still not understanding. I still don't know what that means in regards to where everything would lands on the waveform, for example, comparing it with the mkI.
 
Log curves are a form of mathematical value scale compression. Without seeing a graph curve of C-log2 it is hard to know exactly what the high end compression curve shape is, if any. Most of the modern cine curves pull about 4 stops compression on the low end of the curve to bring linear values into video DR range but the top end stays pretty linear. Video DR unless one is working in REC2020 is about 9.5 stops max as represented by the 0-110 IRE scale. We are talking about the value scale of a display monitor, not the camera.
If as was stated above Canon says 92 indicates full white clip point, you should expose to keep scope values under that to protect highlights when shooting. But when grading a log profile, one can and should grade to use the maximum range for full contrast from black to white depending on aesthetic intent.
The power of shooting log lies in giving you the creative choice of where to place critical midtone values for optimum detail recovery while controlling the high and low range cutoff points for contrast.
 
Last edited:
I have the same problem as Gabriel. When I am working with my C300, I use my waveform for exposing, especially when using it on the M10, where it's hard to see the image in bright sunlight, I 'monitor' the waveform and dial in at what I think is right, however, I do it on instinct, not on pure knowledge. Is there a platform where there's a proper practical explanation on how to interpret the IRE in real life when working. thnx folks, that's all.
 
The details released so far, to my knowledge, on Canon Log 2 aren't as details as the information we have on Canon Log. But you can compare the corresponding IRE values for 0%, 18% and 90% reflectance here:

ProfileCanon Log IRE value %Canon Log 2 IRE value %
0% reflection7.33.54
18% reflection32.839.2
90% reflection62.758.36

We can see that the original Canon Log, which was designed for 8bit sits a bit higher in the low end due to noise of that sensor generation. With new image processing in the C300 mkII and better noise performance it can reach deeper and expand the lower-mid regions. Between 18-90% reflection CLog2 is more compressed, but remember this was designed for 10bit.

Use the 18% gray value for reference and then put your skin tones according to your own preference relative to that.
 
Andree,

Thank you. It's true that there isn't as much info on the Clog2 as we have on the Clog. It's a brand new camera so kinda makes sense. But still makes it overall difficult to gauge. But your explanation helps A LOT. Essentially all the values go down a bit. Granted, it's for 10bit, so there is more information. But as Alain says, instinctively I look at the waveform to understand where everything is landing, and with this I can see how to translate it. Thank you.

I'd still like more information on how Clog2 works. After years of shooting with the mkI, I'm used to understanding the image and the tech of it naturally. Would like to get just as capable with the mkII.
 
You must log in or register to reply here.
Back
Top