Okay. . . even more puzzled by exposure. . . what is noise and what isn't?

[Stephen Pruitt]

Hi all. . .

In reading the manual for Build 30.5, it plainly states that, when using the VIDEO view, purple is areas of noise. However, on our camera, we are able to get all sorts of purple without even beginning to show any noise in the left-most exposure bar. This seems odd. Is purple ALWAYS noise? Or just maybe noise? How would we actually be able to tell the difference?

Thanks.

Stephen

 

[Graeme Nattress]

It's very hard to measure image noise in camera, so we substitute how bright or dark the recorded code values are as a proxy for noise. When you see purple it is "potentially noisy". If there are any critical image tones purple - check. It doesn't mean they'll be noisy - just potentially. Visibility of noise relies on rendering intent, and scene content.

Graeme

 

[Jean Déraps]

If the purple areas are meant to be "very dark" then I find there's generally no problem...it's only if you intend to boost the those "purple areas" because of an error in exposure (or because the producer decides that the scene you shot at dusk would be nicer if it looked like it was shot under the midday sun) that you'll introduce "unacceptable" noise. Now, "unacceptable" is the debatable point here of course....

 

[Stephen Pruitt]

Okay. . . but let me ask this:

It seems as if the left-most exposure graph goalpost is measuring something very different from "purple" areas in the video image. What does a red-line in the left-most goal post mean with respect to noise??? Again, the puzzling aspect is that I was under the impression that noise means random pixel firing (noise floor), and that the goalpost measured the percentage of pixels at the noise floor. Is this not accurate? If that IS accurate, then why wouldn't we have purple show us which pixels are at the noise floor?

It's all very confusing. . . just when I think I have it down. . .and that the goal posts were my friend. . . I find out that they might not be telling me anything really useful at all.

Thanks.

Stephen

 

[Graeme Nattress]

The goalpost gives an indication of the number of pixels (as in area of the screen) that are potentially noisy. The purple tells you where they are on the screen.

Graeme

 

[Stephen Pruitt]

Hi there, Graeme. . .

The question is this: How can I have a lot of purple on the screen but not have any red on the goal posts? They obviously aren't measuring the same things, right?

Stephen

 

[Graeme Nattress]

Do you have an R3D that Stuart and I could look at?

Graeme

 

[Stuart English]

False Color Overlay

False Color Overlay

In reading the manual for Build 30.5, it plainly states that, when using the VIDEO view, purple is areas of noise.

Well one answer to your question is the Ops Guide does NOT say that.

Are you looking at VIDEO or EXPOSURE false color.... ?

 

[Stephen Pruitt]

Sorry for the confusion, Stuart. . . I meant "false color". It states "purple areas represent sensor exposure levels that are likely to be noisy if gained up in post production."

I just shot an R3D and am sending it on to both you and Graeme. It is a one-frame shot that, when shot using exposure, was basically entirely in purple under the exposure false color. However, there was no red on the left goal post at all. Zip. Nada.

I guess I just don't know how that can be. Again, either you are looking at two very different things with these two meters, or something is wrong with Build 30.5 or with my camera (which seems very unlikely).

I'm just confused as all get-out with these various RAW meters. From my reading of the manual, it seems to me that all three meters should be giving me the same information, but they clearly are not. False color (exposure) is clearly not consistent with the goal posts (either high or low, as a matter of fact) or with the RGB histogram.

I'd love to have this confusion lifted in the next day or two.

Thanks.

Stephen

 

[Stephen Pruitt]

Well, I want to do a follow up here.

As it happens, there was some sort of a software issue with my camera (possibly related to a black shading), and after reloading Build 30.5, restoring the factory set black shading, and then doing a new black shading, my exposure meters began working properly.

Very strange. It apparently bemused both Graeme and Stuart for quite a bit this morning, but they kept at it and got me back in the saddle again.

Super, super kudos to both Stuart and Graeme for getting me squared away. I could not possibly have asked for better service.

What an amazing company. . .

Stephen