FS700 RAW files / R5 explained

C

chrditlev

New member
I recently did some exposure tests on the FS700 (shooting RAW to the R5) to see how the RAW files came out.

My question is:
I use the Slog2 picture profile and use my light-meter to set exposure (ISO2000) and shoot a clip. Then afterward I shoot another clip with another pricture profile (cine-gamma (at 500ISO)) and I again use my light meter to expose correctly. Then I should assume that the RAW file would look identical - since the picture profiles are just for viewing and only exists as Meta data on the R5 files. Correct? But they don't! They look different and the highlights are not as unrecoverable from the cinegamma RAW clip. SO: What is different and what affects what gets recorded? I viewed the files RAW in Sony's RAW viewer without the meta data applied (Also tested this in TechniColor's DI suite without any metadata and the still look different).

Thanks,

Chrdit.
 
Last edited:
chrditlev said:
I recently did some exposure tests on the FS700 (shooting RAW to the R5) to see how the RAW files came out.

My question is:
I use the Slog2 picture profile and use my light-meter to set exposure (ISO2000) and shoot a clip. Then afterward I shoot another clip with another pricture profile (cine-gamma (at 500ISO)) and I again use my light meter to expose correctly. Then I should assume that the RAW file would look identical - since the picture profiles are just for viewing and only exists as Meta data on the R5 files. Correct? But they don't! They look different and the highlights are not as unrecoverable from the cinegamma RAW clip. SO: What is different and what affects what gets recorded? I viewed the files RAW in Sony's RAW viewer without the meta data applied (Also tested this in TechniColor's DI suite without any metadata and the still look different).

Thanks,

Chrdit.
Click to expand...

The gamma curve gets baked into the raw output on the FS700. Raw just means 'un-bayered' in this instance - some signal processing occurs before being sent to the recorder. White balance (RGB gain adjustments) and gamma are known about). If you want max range from the raw the camera needs to be in Slog2.
 
Thanks Colin. I spoke with SONY support yesterday and the tech assured me that nothing (besides exposure related settings, F-stop and aperture) is recorded. Everything else is metadata. I was told that the R5 records an unaltered signal of exactly what the Sensor is "seeing" or registering. I am quoting Sony when I say: "Nothing from the picture profile settings including slog2 and gammas are baked into the Raw signal - it's all metadata recorded alongside the RAW file. When brought into the DI at Technicolor all meatadata was stripped (and as I mentioned previously - there was a difference in ho w the files looked).


What you say makes sense and explains it; in fact I am seeing now in the 4K upgrade manual it says that chaining the gamma curve WILL change the dynamic range of the recorded image. (The FS700/R5 tech did promise me on the phone that that wasn't the case.)


Thanks,

Chrditlev.
 
That was something that got me confused too. A Sony Engineer has told me the very same thing - that the output of the RAW signal is Linear RAW and the meta data for knee and other settings is how the PP was set up to monitor on the FS700.

I'd have naturally assumed that the output was affected by the Profile settings, it makes sense.

The issue is 12 bits is not enough for Linear RAW. It maybe be enough for Slog based RAW or other gamma curve, but certainly not *true* linear.

Colin, have you heard different or just based on evidence?

cheers
Paul
 
paulcurtis said:
That was something that got me confused too. A Sony Engineer has told me the very same thing - that the output of the RAW signal is Linear RAW and the meta data for knee and other settings is how the PP was set up to monitor on the FS700.

I'd have naturally assumed that the output was affected by the Profile settings, it makes sense.

The issue is 12 bits is not enough for Linear RAW. It maybe be enough for Slog based RAW or other gamma curve, but certainly not *true* linear.

Colin, have you heard different or just based on evidence?

cheers
Paul
Click to expand...

Mitch has said it quite a few times - I believe that's the reason the Odyssey currently only works with an Slog2 or ITU709 gamma (and hey he advises white balancing the footage) other gammas are not read correctly by the Odyssey. I could have got the wrong end of the stick from what Mitch has said, but that was what I took home from all the discussions had here about the nature of the 12bit 'Raw' from the FS700. I think it is baked in as some form of compression/data reduction is needed to get an accurate reproduction of the 13+ stops of DR the sensor sees via a 12bit stream down a single 3G SDI cable. Adding a gamma correction to the 'linear' output is one part of this I guess.


I suspect the issue is that the F5 and F55 don't do this, but they have 16bit Raw output, rather than 12bit Raw and interface with the R5 directly, with a special connector, rather than via a 3G SDI cable and interface. So probably don't need to bake in the gamma - because they don't have the same limitations. Odds are the tech jus assumed they're the same.

Given that Convergent Design, by necessity, have been analysing the hell out of the data from the FS700, I'd be minded to believe them over some random Sony tech.
 
Colinelves said:
Mitch has said it quite a few times - I believe that's the reason the Odyssey currently only works with an Slog2 or ITU709 gamma (and hey he advises white balancing the footage) other gammas are not read correctly by the Odyssey. I could have got the wrong end of the stick from what Mitch has said, but that was what I took home from all the discussions had here about the nature of the 12bit 'Raw' from the FS700. I think it is baked in as some form of compression/data reduction is needed to get an accurate reproduction of the 13+ stops of DR the sensor sees via a 12bit stream down a single 3G SDI cable. Adding a gamma correction to the 'linear' output is one part of this I guess.


I suspect the issue is that the F5 and F55 don't do this, but they have 16bit Raw output, rather than 12bit Raw and interface with the R5 directly, with a special connector, rather than via a 3G SDI cable and interface. So probably don't need to bake in the gamma - because they don't have the same limitations. Odds are the tech jus assumed they're the same.

Given that Convergent Design, by necessity, have been analysing the hell out of the data from the FS700, I'd be minded to believe them over some random Sony tech.
Click to expand...

I hope so too. But Sony have been *overly* insistent on calling the output of the FS700 12 bit Linear RAW. If it wasn't for that word linear then i wouldn't have had a second thought. I've had discussions in the past with others about methods of compressing (without profiles) too, but inconclusive.

There's also a chance that CD are adding a REC709 themselves. In that if you set the camera to 709 (which according to Sony only affects the metadata and the way the LCD monitors) CD can read that metadata and then turn the incoming Linear into REC709 via their own LUT. One test of this would be : do the profile settings in camera change, is the black gamma and the knee and the other settings fed through to the O7Q?

And they don't appear to affect the O7Q image (just tried superficially).

This leads onto a wider question that if you cannot use the Knee and Gamma settings then you're quite limited in how you squeeze the full range into that curve.

So it leaves it a bit of a mystery really. More than happy for Mitch to chime in and stomp all over my theories though - Calling Mitch?

cheers
Paul
 
I'm very curious about all this as well, is the FS raw from the FS700 true linear raw or has a baked in gamma? (which would be a good thing! If we're to squeeze the full dynamic range into those 12 bits)
 
You must log in or register to reply here.
Back
Top