Open Archduk3 opened 3 years ago
Hi Archduk,
As far as I can see (Admittedly a quick skim read) that paper seems to show S-Log1 mid grey (18%) at 10-bit CV 394, which is what it always has been.
Once defined, curves don't generally get changed, they just come up with new ones :-) . S-Log2 is literally S-Log1 underexposed half a stop to allow extra headroom with quieter sensors, since white clip is set at CV 1016 for both.
Ben
I would say the paper that @Archduk3 has posted is the older paper. It is dated 2009, and the S-Log equation given there does not include the linear extension below linear light zero. This is added in the S-Log2 paper (dated 2012). Other than that the equations are the same (taking into account the fun that is reflectance vs 'linear IRE' and full/legal range scaling).
This is the latest paper that Sony Pro officially supplied with, as well as the actual CDL. I had to get them to search through the hard drives to get it.
The base ISO for F35 has been argued as 320 for a long time though, even in testing. And 450 should be noted more as a number on camera than anything actually used in real word measurements.
mån 12 juli 2021 kl. 07:08 skrev Nick Shaw @.***>:
I would say the paper that @Archduk3 https://github.com/Archduk3 has posted is the older paper. It is dated 2009, and the S-Log equation given there does not include the linear extension below linear light zero. This is added in the S-Log2 paper (dated 2012). Other than that the equations are the same (taking into account the fun that is reflectance vs 'linear IRE' and full/legal range scaling).
— You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub https://github.com/cameramanben/LUTCalc/issues/18#issuecomment-878188634, or unsubscribe https://github.com/notifications/unsubscribe-auth/AUXIF3B3FOZTMVP4VMNH7PLTXLELRANCNFSM5AFJBI3A .
-- Have a nice day Carl Nenzen Loven (415) 802-3254 Cinematographer IATSE Local 600 | First assistant camera S.O.C. Associate Member Based in Los Angeles
I would suggest that the ACES IDT provided by Sony supersedes a paper written in 2009. And the IDT includes the linear extension.
In terms of "native ISO", the paper you posted says the ISO is 450 at 0dB gain, and simply recommends settings it to 320 (-3dB) "when a quieter noise floor is the priority".
Best
N
I note the S-Log IDT was updated to add the linear extension in a commit in January 2014.
Ah yes,
I get it now. The dB / ISO relationship is more than a little fuzzy in the world of arbitrary transfer functions. S-Log1 is, was and forever will be defined by the log function as used in LUTCalc. The one change that has occurred is to define superblacks as Nick mentions, as log doesn't appreciate zeros.
That S-Log1 function defines the output relative to black (10-bit 90) and 18% grey reflectance (10-bit 394) and the intention to allow five and I think a third stops headroom. It makes no judgement upon the fundamental 'base' ISO of the camera.
Working down from sensor white clip on the S35 that headroom seemingly sets the 'base' ISO as 500 ish. However, from what you suggest, Sony considers that a bit noisy, recommending using a CineEI or 320, losing a bit of headroom to keep the shadows a bit further from the noise floor. Losing headroom on a cinema camera doesn't sound so good on the sales pitch though!
Sony cameras record log the same regardless of CineEI setting, just placing a note in the metadata and adjusting the feed to MLUTs, so, CineEI 320 is the same as overexposing about half a stop. In S-Log1 terms, that puts mid grey at about 10-bit 450.
This is probably totally unhelpful, but for me I'm happy to see where the numbers come from!
Thanks,
Ben
Sorry for the confusion then, thanks for educating me.
In terms of 10-bit. You’re referring to the LUT samples or actual recording. Since F35 does 12-bit.
C
mån 12 juli 2021 kl. 11:50 skrev Ben Turley @.***>:
Ah yes,
I get it now. The dB / ISO relationship is more than a little fuzzy in the world of arbitrary transfer functions. S-Log1 is, was and forever will be defined by the log function as used in LUTCalc. The one change that has occurred is to define superblacks as Nick mentions, as log doesn't appreciate zeros.
That S-Log1 function defines the output relative to black (10-bit 90) and 18% grey reflectance (10-bit 394) and the intention to allow five and I think a third stops headroom. It makes no judgement upon the fundamental 'base' ISO of the camera.
Working down from sensor white clip on the S35 that headroom seemingly sets the 'base' ISO as 500 ish. However, from what you suggest, Sony considers that a bit noisy, recommending using a CineEI or 320, losing a bit of headroom to keep the shadows a bit further from the noise floor. Losing headroom on a cinema camera doesn't sound so good on the sales pitch though!
Sony cameras record log the same regardless of CineEI setting, just placing a note in the metadata and adjusting the feed to MLUTs, so, CineEI 320 is the same as overexposing about half a stop. In S-Log1 terms, that puts mid grey at about 10-bit 450.
This is probably totally unhelpful, but for me I'm happy to see where the numbers come from!
Thanks,
Ben
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/cameramanben/LUTCalc/issues/18#issuecomment-878391917, or unsubscribe https://github.com/notifications/unsubscribe-auth/AUXIF3HLPH4O4DEZ3IFU7ETTXMFNFANCNFSM5AFJBI3A .
-- Have a nice day Carl Nenzen Loven (415) 802-3254 Cinematographer IATSE Local 600 | First assistant camera S.O.C. Associate Member Based in Los Angeles
Would is be possible to have LutCalc output a format that could load this in as an MLUT for the F35?
I have been trying ot figure out the right format. Having a conversion LUT to C-log would probably be more ideal since most monitors do NOT have a standard Slog lut.
C
sön 11 juli 2021 kl. 16:30 skrev Ben Turley @.***>:
Hi Archduk,
As far as I can see (Admittedly a quick skim read) that paper seems to show S-Log1 mid grey (18%) at 10-bit CV 394, which is what it always has been.
Once defined, curves don't generally get changed, they just come up with new ones :-) . S-Log2 is literally S-Log1 underexposed half a stop to allow extra headroom with quieter sensors, since white clip is set at CV 1016 for both.
Ben
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/cameramanben/LUTCalc/issues/18#issuecomment-877856571, or unsubscribe https://github.com/notifications/unsubscribe-auth/AUXIF3CKLKDD4MJW7BDWRI3TXH5O5ANCNFSM5AFJBI3A .
Hi,
It seems the data for SLOG1 is based on the first edition white paper, not the later one from the last update to the F35.
While the camera reads 450 for Cine EI still, the later white paper rates it at 320. I am posting the most recent white paper here as well.
C SonyLogWhitepaperEng1110.pdf