Log/Rec.709, Headroom, and Dynamic Range....

[Eryc Tramonn]

If you change ASA in camera you simply change the LUT of the monitor-out but not the exposure of the recorded RAW file. It stays at roughly 320 ASA.

Hans

That's right for the RED, sorry to have left that out. If you're shooting film - different story.

 

[Dj Joofa]

Last week I shot some simple followup Red test footage for Stu to see just how far down we can push 18% gray to preserve highlights and have reasonable flesh-tones (noise wise) when they are brought back up.

Paul

Paul, if you have a relatively smaller range to work in (digital latitude) and you try to fit in a larger range in that (film latitude), then there is no way you can preserve both shadows and highlights. I think there is limited value to underexposing and trying to resurrect things by bringing them back up.

You have only a "window" into a larger range and you can move you window up and down, but not capture all of the larger range.

The temporal noise is eating information away at the lower stops and if you provide more of your precious information to it by underexposing it shall oblige by having a go at larger chunks of it and you can't bring them back.

In the film world you have less noise in the darker areas vs. highlights and you can loose some information by crushing as provided by underexposing, but since it is an analog medium you can bring some information back later.

 

[Bill Anderson]

Adding to what joofa says, "underexposing" habitually is not going to increase your exposure latitude; the more accurate the exposure, the more latitude you'll be able to pull from. But I'll admit, there is a point when "accurate" and diminished high values are so close together, that one is, from time to time, quite right to back away from the abyss a notch or so.

 

[dVonAllmen]

To make sure I'm not misunderstanding what Mr. Mullen and Mr. Paulsen are saying, it basically boils down to

"Don't blow out the highlights, then if your mid-tones are too dark adjust that part of the curve up in post."

Yes? No?

And, Hans, to answer your question, my ancestors emigrated from Germany down to Switzerland in the 16th Century, settling in a town 30k south of the border.

 

[Paul Nordin]

Adding to what joofa says, "underexposing" habitually is not going to increase your exposure latitude; the more accurate the exposure, the more latitude you'll be able to pull from.

Bill & Joofa, I think we are saying the same thing. Of course there is no reason to underexpose habitually. If you have well balanced lighting in your frame, and it all is looking good on the histogram, and your talent's faces are falling where you want them for the mood you are creating, then I would not underexpose. No need to, because even if I miss something small, there's an additional .4 stop headroom. Or if I -want- clippy-blown windows as a style for the film, or am willing to put up with the challenges that will introduce in post for film-out, or for any other reason that I've thought about then I let 'em go.

However, when shooting in contrasty situations (car interiors, interiors with sunny windows, in shade with hot backgrounds) and I want to balance inside-outside so there is minimal clipping (say 2 stops over key), I need to know what will happen if in post if I underexpose the -main thing- in the frame (talent) to preserve detail in the background, when a powerwindow is placed on their faces to bring them up to where they should be? That's all I'm exploring. What I've found is that with build 13 & 14, for me I would never underexpose those faces more than 1.5 stops below where I want them to be as a strategy to keep detail in the highlights. And I'd think hard about going below 1 stop under. So that's the current answer for me barring new info from testing build 15 and on: I have a little more than one stop of latitude on underexposed fleshtones before I get into dutch. That is a key factor to know in designing lighting plans, ordering lighting packages, thinking about time of day, etc.

It's not a habit, but a data point that I can use to know what I can and cant get away with.

 

[Bill Anderson]

Exactly, Paul.

 

[jimhare]

Okay. We know that setting the ASA in camera is really just for monitoring and a meta-data starting point for post and doesn't affect the way the sensor generates the image. Clear, got it.

So how does this affect how you measure light on set? If you dial up 100ASA in post is it truly 100ASA or still 320 lowered? In other words, are you better off just shooting EVERYTHING at 320 and compensating with iris and NDs?

Sorry if this is a stupid question, every time I think I understand it a new piece of info appears and confuses me all over again!

 

[Dj Joofa]

Agree with you Paul. Thanks for the explanation.

 

[Paul Nordin]

In other words, are you better off just shooting EVERYTHING at 320 and compensating with iris and NDs?

Its not a stupid question, and you are right. its for monitoring only. If you are used to shooting film, consider your Red camera loaded with an ASA320 film stock. ND, Iris, Shutter...standard ways, and the safest ways to control exposure.

 

[Michael Brennan]

Keep it simple.

Keep it simple.

The primary reason to underexpose HD or RED one or two stops is if you can't control the dynamic range of the scene via normal means, ie lighting or filters.

Those who regularly underexpose as a matter of course are losing one or two stops of shadow detail into the noise floor.




Mike Brennan

 

[NateWeaver]

In other words, are you better off just shooting EVERYTHING at 320 and compensating with iris and NDs?

Bingo.

If you shoot 100 and you have blown highlights, they're going to look pretty clippy, because the clip point is lower.

ASA on this camera as it is right now is a LUT, whether you want to call it that or not (I don't).

The more I become familiar with the camera, the simpler it becomes for me. All the ASA/LUT/curves stuff in post is just obfuscation of the fact that this is a ASA320 camera, with a 5000K balanced sensor. Make what's in front of the camera look good to those specs, and you'll have the best image possible.

 

[jimhare]

Good to know, makes it much simpler.

Thanks for not jumping on my ignorance. There is some scary talent/experience on this list and it makes it intimidating to speak up sometimes!

Cheers,

Jim

 

[Dj Joofa]

its for monitoring only. If you are used to shooting film, consider your Red camera loaded with an ASA320 film stock. ND, Iris, Shutter...standard ways, and the safest ways to control exposure.

So if you only increase ASA from 320 to say 640, in your estimation, which ones of the following is Red doing in software (could be more than one):

(1) Boosting signal values using digital gain
(2) Compensating for the aperture size
(3) Compensating for shutter angle
(4) Anything else.

 

[Joe Carney]

Bingo.

If you shoot 100 and you have blown highlights, they're going to look pretty clippy, because the clip point is lower.

ASA on this camera as it is right now is a LUT, whether you want to call it that or not (I don't).

The more I become familiar with the camera, the simpler it becomes for me. All the ASA/LUT/curves stuff in post is just obfuscation of the fact that this is a ASA320 camera, with a 5000K balanced sensor. Make what's in front of the camera look good to those specs, and you'll have the best image possible.

Thanks everyone, thats what I thought.

 

[Paul Nordin]

So if you only increase ASA from 320 to say 640, in your estimation, which ones of the following is Red doing in software (could be more than one):

(1) Boosting signal values using digital gain
(2) Compensating for the aperture size
(3) Compensating for shutter angle
(4) Anything else.

:nerd: Multiple choice test!? What do I get if I answer correctly?

I probably shouldn't venture a guess as that's all it could be, but I will anyway. Though this is a question better suited to Graham who probably had a hand in writing the software. And by software are you talking about the monitoring software in camera or in RedCine? Either way, my ill informed guess would be:
in what probably is a very complicated way, changing ASA results in boosting gain and maybe monkeying with gamma. I suppose it could also effect some detail and smoothing to deal with increased noise if you increase ASA, but that may be flying in the face of the Raw paradym. The red doesn't know about aperture size since its not getting a signal from the lens yet. And ASA has nothing to do with the shutter angle.

Graham, did I pass?

 

[Blair S. Paulsen]

What choo lookin' at...

What choo lookin' at...

Hmmm. Did you pass? Well you might start by spelling Graeme correctly

In any case I am operating under the assumption that changing the ASA in the monitor path does not in fact change the image data in any way. If your query was simply curiosity about the technique employed for making the monitor image brighter I have always assumed it was simply a value shift - essentially digital gain but I hesitate to use that term with clients since they might jump to the wrong conclusion that the RedOne is adding gain to their image.

FWIW I take the approach of referring to the monitor output as a high quality video tap but not actually a live signal output. Somebody on the RedTeam can correct me if I have this wrong but my understanding is that in order to provide a real time output the de-bayering is a bit crude compared to full processing in RedCine. It looks amazingly good, but it is just a tap.

On a related note I do not believe that the video tap is software driven, I think it is done in hardware with a FPGA - but that is just a guess.

 

[Paul Nordin]

Hmmm. Did you pass? Well you might start by spelling Graeme correctly

:blush: Oh crap, I'll sit quietly in the back of the room now.

Sorry Graeme.

Cheers,
Paul (can't tell you how many times people have misspelled my name) Nordin

 

[Dj Joofa]

:nerd: Multiple choice test!? What do I get if I answer correctly?

Thanks Paul.

You get a certification in video technology if you pass

 

[Joshua Provost]

"Rating" the camera higher in order to underexpose (say from 640 instead of 320) has very little to do with setting the ASA on the camera. The important factor is setting your light meter to your desired rating (again, say 640), and then exposing based on the meter. Regardless of the on-camera ASA and preview you are seeing, it's how you use your light meter to set your iris that makes the difference. Changing the ASA on the camera does very little, as we have found out.

That being said, if you do rate at 640, you might as well set the ASA on the camera to 640, but it's the lightmeter and iris, not the setting, that are important.

 

[Paul Nordin]

it's the lightmeter and iris, not the setting, that are important.

Yep, exactly right