Odyssey Firmware 2015.5

stefanhuberfilms said:
does that mean that the Odyssey chooses "extended range" because the LUT i'm using was designed for that specific range? (i was using a custom LUT from a friend.) doesn't quite make sense to me, since the LUT shows the superwhites only if i chose "legal", but clips the superwhites in extenden mode.
Click to expand...
In an "extended range" coding of an image 1023 represents 100%, and since in 10 bit you can't have a higher value than 1023, it can't represent super-whites. So they get clipped off. This mirrors the behaviour in Resolve, where although super-whites are available to be brought in by grading, if you just apply a standard 3D LUT they will be clipped.
 
Last edited:
Clermond said:
Does that mean the supplied slog2 lc709a LUT is not suitable for the FS700?
Click to expand...
The S-Log2 to LC709A LUT is absolutely suitable for use with the FS700. That is what it was intended for. But if you don't like the look it gives your image, don't use it.

I would say your original image is slightly cool. It is an exterior in shadow after all. LC709A is a "Low Contrast Rec.709" look. From your grade, you clearly want a higher contrast, warmer look, so maybe LC709A is not for you. What constitutes a "well balanced neutral image" is actually quite subjective.
 
Nick_Shaw said:
In an "extended range" coding of an image 1023 represents 100%, and since in 10 bit you can't have a higher value than 1023, it can't represent super-whites. So they get clipped off. This mirrors the behaviour in Resolve, where although super-whites are available to be brought in by grading, if you just apply a standard 3D LUT they will be clipped.
Click to expand...

if we take this vice-versa... does that mean that in "legal range" the value 1023 represents 109% - hence being able to display superwhites?

in Resolve we're either working as "limited" video or "full" data range (data beeing the setting where i always was able to see superwhites), no?
 
Clermond said:
hmm.. for my understanding that's why we shoot color charts to define well balanced neutral images.
Click to expand...
Of course colour charts are a reference for the colours in front of the camera. But an image is only "neutral" insofar as white balancing makes the red, green and blue levels identical for the grey chips on the chart. The values are not identical in your source image, nor in your grade of it. It is not the purpose of a standard LUT to correct white balance. If you want an image out of the LUT with an absolutely neutral white balance, you need to put a neutral image into it.

If you want to see how subjective a "well balanced neutral image can be" take e.g. an Alexa LogC image into Resolve in "DaVinci YRGB" mode, and apply the built in ARRI LUT. Then turn off the LUT and switch to ACES mode and apply the ARRI IDT and the Rec.709 ODT. The two results are ARRI's concept of a well balanced image, and The Academy's. They are dramatically different, and neither is right or wrong.
 
stefanhuberfilms said:
if we take this vice-versa... does that mean that in "legal range" the value 1023 represents 109% - hence being able to display superwhites?
Click to expand...
Correct

in Resolve we're either working as "limited" video or "full" data range (data beeing the setting where i always was able to see superwhites), no?
Click to expand...
Resolve works in unclamped floating point, so although you may not be able to see super-whites greater than 1.0, they are available for you to pull back in with grades. Unless that is you apply an operator like a standard 3D LUT or some curves which clamp to the 0.0 to 1.0 range.

What you are doing in the Clip Attributes of Resolve is choosing how the data of your source clip is mapped to Resolve's floating point representation. "Video" maps 64 to 0.0 and 940 to 1.0 (reserving values greater than 940 as float values greater than 1.0) and "Data" maps 0 to 0.0 and 1023 to 1.0. The reverse happens on the Deliver page. The 7Q's Extended mode does the same mapping from 64-940 to 0-1023, then applies the LUT, then maps back to 64-940. This means the same EE LUT will give the same result in the 7Q and Resolve. Both the 7Q and Resolve will clamp super-whites in this process if you use the default Resolve processing of ProRes (interpret as "video", i.e stretch 64-940 to 0.0-1.0).

I think some of the confusion stems from the fact that when you set "video levels" in the Clip Attributes for your media, that does not mean you are "working in video range". The opposite is true. You are stretching the video range to fill the whole 0.0-1.0 range. If you want to actually work in video range you need to "lie to Resolve" by telling it your media is "data range" so it does not get stretched out. Personally I do not recommend this approach.
 
Nick_Shaw said:
I think some of the confusion stems from the fact that when you set "video levels" in the Clip Attributes for your media, that does not mean you are "working in video range". The opposite is true. You are stretching the video range to fill the whole 0.0-1.0 range. If you want to actually work in video range you need to "lie to Resolve" by telling it your media is "data range" so it does not get stretched out. Personally I do not recommend this approach.
Click to expand...

FWIW Avid Media Composer works the same way, altho' using slightly different terms. I/O, in Avid, refers to RGB and 709 levels, which is another mixed up nomenclature. Selecting RGB levels forces Avid to remap to 64-940, while selecting 709 levels forces Avid to not remap any values. This has always seemed backwards, to me, yet this seems to be somewhat of a convention, as you describe.

I've noticed instances where roundtripping to Resolve causes some clipping when selecting "Video". It's become de riguer for me to select "Data Range" to avoid the clipping.

Altho', since log color gammas clip below 100 IRE, I presume this isn't that big of an issue when using a log gamma?
 
Last edited:
thanks for the detailed explanation Nick, much appreciated! Makes much more sense now. Premiere handles this the same way and needs, for example, a fast color corrector to map the values above 940 and below 64 to the visible range.

Important things to consider if you want to gain the maximum DR out of a cameras signal.
 
cuervo said:
Altho', since log color gammas clip below 100 IRE, I presume this isn't that big of an issue when using a log gamma?
Click to expand...
Unfortunately that's not universally true.

REDlogFilm and ARRI LogC only go up to 100 IRE. All the variants of S-Log, for example, go up to 109 IRE. I prefer the ARRI/RED approach is it's very easy for people to accidentally clamp at 100 IRE, probably without even realising there was ever anything there to clamp off.
 
Nick_Shaw said:
Unfortunately that's not universally true.

All the variants of S-Log, for example, go up to 109 IRE. .
Click to expand...

hmmm....I'm showing the limits of my knowledge. Perhaps, rather than using the term clipping, to be more correct, the slog gamma, after a LUT like LC709A, is applied, rolls off below 100 IRE, as they have a knee.

Using the LUTCalc tool from cameramanben, I can get a pretty good picture of almost any gamma curve with a LUT applied, where 18% gray, 90% white and roll off occurs.

(thanx for explaining this. It's something I'd like to have a much better understanding of)
 
Last edited:
cuervo said:
hmmm....I'm showing the limits of my knowledge. Perhaps, rather than using the term clipping, to be more correct, the slog gamma, after a LUT like LC709A, is applied, rolls off below 100 IRE, as they have a knee.
Click to expand...
Sadly again it's slightly more complicated. The LC709A LUT does indeed roll off to 100 IRE, because it is based on the ARRI approach. Sony's original Rec.709(800%) LUT rolls off to 109 IRE. The Odyssey's preset R800 LUT, being "EE" cannot reproduce the last part of the roll off. The "LL" version I posted earlier in this thread can.
 
stefanhuberfilms said:
So back to my initial question, why doesn't the 7Q remember if i had set legal or extended range? On purpose? no?
Click to expand...
I believe that it is set to default to Extended range, as that it correct of all our built in LUTs. Most people simply do not understand what a LUT RANGE is. Just look at the last few pages of this thread and then think about all the people out there who don't know the difference between RAW and Log, etc. So we want the system to default to the correct setting.

But if you choose to switch to Legal then we have to assume that you meant to do that. What we do not do is change this setting when you switch between different LUTs. We have room in both the naming convention as well as in the notes field for you to label the LUT as to Legal or Extended. If you make your own there's no way for us to know what you based it off of and what you are intending so you need to decide for yourself.
 
Mitch,
I think what Stefan meant was that if I select Legal, then, when I power down and then power up, again, the O7Q automatically resets to Extended.
 
Has anyone had any issues with 4ktoHD with the 4 second burst on the FS700? I was recording to ProRes HQ.
Monitoring was really choppy-like 3fps choppy. When writing the buffer, playing it back on the Odyssey, and back home on my computer it appeared fine.
Just glad it records fine, but makes framing shots tricky when really choppy.
 
Nick_Shaw said:
Unfortunately that's not universally true.

REDlogFilm and ARRI LogC only go up to 100 IRE. All the variants of S-Log, for example, go up to 109 IRE. I prefer the ARRI/RED approach is it's very easy for people to accidentally clamp at 100 IRE, probably without even realising there was ever anything there to clamp off.
Click to expand...

I have tested s-log2 on my Sony A7s and it definitely clips below 100 IRE ( about 98 IRE ) irrespective of the level of overexposure. It also clips blacks at about 10 IRE irrespective of the level of underexposure. Why would you create an extended range LUT for this camera ?
 
You must log in or register to reply here.
Back
Top