cameramanben
Well-known member
Hi All,
I'm currently working on an update for LUTCalc, and have uploaded an online version to LUTCalc Testing.
I've been playing with the picture profiles in an a7s mkII and had a go at analysing the Cinegamma 3 and Cinegamma 4 curves. I checked my process against the Cinegamma 1 and Cinegamma 2 curves, which are said to be HG4 and HG2 respectively. Basically I shot charts at various exposures in S-Log2, S-Log3 and the Cinegammas, all supposedly at the same ISO (1600 for the a7s mkII). I reset all the picture profiles before I began, and for Cinegamma 3 and Cinegamma 4 I adjusted picture profiles 5 (default CG 1) and 6 (default CG2). I don't spend enough time with the a7s to be 100% sure that nothing had been tweaked, but I had reset things as far as I knew how.
The data came out very nicely, though Cinegamma 1 and Cinegamma 2 appeared around a stop and a bit brighter than they should have been. Could be finger trouble over the ISO, could well just be the way that Sony has chosen to set things up in the stills cameras against the dedicated videos.
As such, when the Cinegamma 3 and Cinegamma 4 looked to be quite bright (mid grey around 50%, as seems quite common in stills cameras where sRGB is the base gamma, rather than Rec709), I corrected the stop on both by half a stop, which drops mid grey into alignment with Rec709 and everything below that quite neatly akin to Sony's Rec709(800%). I mention all this to see what anyone thinks of the approach.
The other thing I have been spending a deal of time on is working through the HDR colourspaces. I have included PQ and HLG curves for ages, but frankly having now worked through a bunch of technical documentation I needed to completely overhaul things (and fix some misunderstandings). A big element for me was in not grasping the significance that HLG is scene referred whilst PQ is display referred; fundamentally different approaches to how they work.
The code should now be pretty robust (and correct), and I won't go further here as it gets nuts techie easily, and if you know what an OOTF is for, you don't need me stumbling my way over it!
If you do know your OOTF from your EOTF and something doesn't seem right, please do let me know.
Lastly, I've added a tool called 'Display Colourspace Converter'. This sits after any adjustments you have made, and acts as a final chance to change the basic colourspace of your output. In other words, if you work on a Rec709 LUT, then want to make a LUT for DCI-P3 displays which should look the same as the Rec709 LUT on a Rec709 display, use the Display Colourspace Converter.
As ever, please let me know your thoughts,
Ben
I'm currently working on an update for LUTCalc, and have uploaded an online version to LUTCalc Testing.
I've been playing with the picture profiles in an a7s mkII and had a go at analysing the Cinegamma 3 and Cinegamma 4 curves. I checked my process against the Cinegamma 1 and Cinegamma 2 curves, which are said to be HG4 and HG2 respectively. Basically I shot charts at various exposures in S-Log2, S-Log3 and the Cinegammas, all supposedly at the same ISO (1600 for the a7s mkII). I reset all the picture profiles before I began, and for Cinegamma 3 and Cinegamma 4 I adjusted picture profiles 5 (default CG 1) and 6 (default CG2). I don't spend enough time with the a7s to be 100% sure that nothing had been tweaked, but I had reset things as far as I knew how.
The data came out very nicely, though Cinegamma 1 and Cinegamma 2 appeared around a stop and a bit brighter than they should have been. Could be finger trouble over the ISO, could well just be the way that Sony has chosen to set things up in the stills cameras against the dedicated videos.
As such, when the Cinegamma 3 and Cinegamma 4 looked to be quite bright (mid grey around 50%, as seems quite common in stills cameras where sRGB is the base gamma, rather than Rec709), I corrected the stop on both by half a stop, which drops mid grey into alignment with Rec709 and everything below that quite neatly akin to Sony's Rec709(800%). I mention all this to see what anyone thinks of the approach.
The other thing I have been spending a deal of time on is working through the HDR colourspaces. I have included PQ and HLG curves for ages, but frankly having now worked through a bunch of technical documentation I needed to completely overhaul things (and fix some misunderstandings). A big element for me was in not grasping the significance that HLG is scene referred whilst PQ is display referred; fundamentally different approaches to how they work.
The code should now be pretty robust (and correct), and I won't go further here as it gets nuts techie easily, and if you know what an OOTF is for, you don't need me stumbling my way over it!
If you do know your OOTF from your EOTF and something doesn't seem right, please do let me know.Lastly, I've added a tool called 'Display Colourspace Converter'. This sits after any adjustments you have made, and acts as a final chance to change the basic colourspace of your output. In other words, if you work on a Rec709 LUT, then want to make a LUT for DCI-P3 displays which should look the same as the Rec709 LUT on a Rec709 display, use the Display Colourspace Converter.
As ever, please let me know your thoughts,
Ben
Last edited: