Welcome to our community

Be a part of something great, join today!

Log in
  • Hey all, just changed over the backend after 15 years I figured time to give it a bit of an update, its probably gonna be a bit weird for most of you and i am sure there is a few bugs to work out but it should kinda work the same as before... hopefully :)

Is RAW Monitor On Camera Really RAW?

R

Russ Lasson

Well-known member
Joined
Nov 9, 2007
Messages
85
Reaction score
0
Points
0
Location
Salt Lake City, Utah
Website
www.colormill.net
My original understanding was that when you choose to monitor in RAW on the camera, that was a representation of what the sensor was capturing.

But when I open up an R3D file in Scratch, it automatically looks darker than what I saw on the camera. After playing around a little, it's clear that the RAW file in Scratch is seeing the R3D file in linear light (gamma 1.0), while the camera output has a different gamma applied (closer to REC709?).

My guess is that the RAW setting in camera is still applying a gamma curve to the sensor data to make it more pleasing to look at. The problem I'm seeing is that several DPs are underexposing in certain environments because of what they're seeing on the camera monitor.

I've seen several DPs shoot footage that they thought was well exposed, only to bring it into Scratch to see that the RAW file is exposed no brighter than 10 IRE. While the R3D file still has enough latitude to work with these files, there is a definite increase in noise when compared to shots that were exposed properly. If the DPs could see the RAW output on the camera as linear (without a gamma adjustment) then it seems they could make better exposure decisions.

So first, is Scratch right with the RAW file or is the RAW monitor on the camera right (or did I not have the camera set up right when I did my testing).

Second, is there any way to have the camera output RAW without a gamma adjustment? I know it looks horrible because of how dark and flat it looks, but it seems like it would be a helpful tool to people trying to really figure out how exposure works on the camera.

Thanks,

-Russ
 
Russ L said:
MIf the DPs could see the RAW output on the camera as linear (without a gamma adjustment) then it seems they could make better exposure decisions.
Click to expand...

The gamma adjustment would not cause anyone to underexpose significantly: something else must have caused the mistake. RED already provides a veritable buffet of tools to help guide exposure.
 
Noone really answered this from RED, but it's not raw/lin. It's the raw colors, with a lift in the bottom. It's a bit off REC 709, though...

Edit: It isn't apparently...

If you set the cam to rec 709 and native iso/color, the waveform and other exposure tools are pretty close to what you'll see in post.
 
The gamma adjustment would not cause anyone to underexpose significantly: something else must have caused the mistake. RED already provides a veritable buffet of tools to help guide exposure.
Click to expand...

I've seen this mostly with DPs that are new to shooting RED. They're struggling to really understand how the image is really translating to the RAW file. When they turn the camera monitor to view RAW and the exposure looks okay (a little dark, but still okay) then they say, "I can live with that."

But when I bring it into Scratch and see the RAW with a linear gamma, it's underexposed and the whole image fits between 0 to 15 or 20 IRE. Then the DP says to me, "Why is it so dark? I don't think I exposed it that way." Because of the flexibility of the R3D file, we've been able to make the image look acceptable, but there is definitely more noise in the image than there would be if it was exposed properly.

Again, on set or even with the default settings in REDAlert or REDCine, it doesn't look as bad. But if you reset the gamma curve to one, then you see that the exposure isn't optimized.

These DPs aren't inexperienced guys. They really know their stuff when it comes to film and HD, but they're trying to understand how the RED camera works (underneath all of the LUTs and gamma adjustments). I should add that most of the footage looks great.

So really, my main questions are:

Does the true R3D file have a linear gamma? If so, what is the gamma correction of the RAW output on the camera? Is there an option for turning off this gamma adjustment on the camera outputs so that the RAW output has a linear gamma?

Thanks,

-Russ
 
RAW view is no matrix or colour or exposure adjustment and a pure REC709 gamma curve. Showing linear light would lead to massive over-exposure. Linear light view is very, very hard to expose correctly - it's so counter intuitive and non of us are used to looking at linear light images.

The R3D file itself is linear 12 bit, just as the sensor is linear 12 bit.

Graeme
 
Graeme Nattress said:
The R3D file itself is linear 12 bit, just as the sensor is linear 12 bit.
Click to expand...

Graeme, don't know how to express this in the most polite way, but may I point out that in the rigorous sense saying "linear light" or a "file is linear" has little, if any meaning. One may talk of linear maps, linear operators or of linear spaces but linear light or linear files do not make much sense.

But, I guess the point is that the input of the sensor pixels (the amount of light) and the output (the pixels voltages) are in a linear relation. This implies, the amount of light and the pixel voltages are the same up to a scaling factor. Then your "linear file" is a compoud word meaning the pixels voltages are stored into the R3D file "as is". Consequently, the linear relationship with the input is still retained with the voltages stored in R3D files.

Perhaps by "linear light" you mean, if these same voltages were shown on the monitor, it looked akward. Thus, you introduce another map (the REC709 gamma curve) between the voltages and the EVF/monitor.
 
Graeme - When you say 'no exposure adjustment', what is the base uncorrected ISO? I know 320ISO is a commonly quoted figure, but in my tests the un-adjusted exposure of RAW view is between 200 and 250ISO. The point of this is being able to bring a r3d into RedCine / RedAlert and see the same image you saw in the LCD in RAW view.
 
ISO320 has no gain what-so-ever applied and hence is base ISO.

Graeme
 
graeme, is the raw histogram "accurate"? meaning it is the real raw data? so if that clips, fuhget about it?
 
Yes, raw histogram is accurate.

Graeme
 
Well, this is a surprise.
I thought RAW view was RAW.
Apparently, it's "RAW color with a Rec 709 gamma curve"
Which is fine, if you know it.
Graeme, would there be any benefit to a fourth view? Linear or whatever? My view of exposing for digital is that more information is better (though I'm not a fan of RedSpace) and being able to evaluate using many tools is ultimately helpful. Unless linear is so difficult to evaluate on a monitor without a .45 gamma applied...
I remember in the dawn of my career as a 2nd AC doing "slop tests" in the hotel bathroom at the end of the day so the DP could look at a bit of negative (we were on a location movie and did not see dailies until they were, well, weeklies)
It's good someone brought this up...
Cheers,
Harry
 
I think in the very earliest Frankie, we had a linear light view, and it was utterly dreadful to use.

Graeme
 
I think the use for a true lin preview would be much broader today, after several lof us have been correcting images from lin for a while...
 
A linear light output wouldn't show you what's happening in the darkest stops though. Using it to expose would lead to over-exposure as it looks so dark. The curve used is the simplest curve to perceptually linearize the data. It does not induce clipping. It allows you to see white clipping and more going on in the shadows. It looks "reasonable".

"Lin" has been so abused by the lin=video gamma crowd that I don't use that terminology. I try to be extra clear and say linear light. I think that's good practice for all of us.

Graeme
 
Actually I am thinking of "lin" like when you start in Scratch or hit "F" in RC or choose "lin" in color.

I agree about the lowlights, but it would be nice to check what's actually recorded in the highlights :)

I agree this is purely academic, as you don't plan to release such an option, but that's what I hoped RAW view would be.

I was off, though :)

No complaints, btw. Just what I hoped for and how I'd use it.
 
You must log in or register to reply here.
Back
Top