Cineon Conversion: convert_log2lin in other applications to match VFB

[Jorq1H]

Hi all. I'm in need of some assistance regarding the way VFB handles lin2log conversion.
I'm trying to replicate this effect in After Effects, to make sure I'm getting the same result using log luts there as in the VFB.

Looking at the convert code in vraylut.cpp shows that the standard values of 95 black and 685 white going from linear to log:

...
static float convert_lin2log(float linValue, float cinBlack, float cinWBdiff) {
float res=log10f(cinBlack+linValue*cinWBdiff)/3.41f;
return res;
...

...
float blackCode=95.0f;
float whiteCode=685.0f;
float cinBlack=powf(10.0f, blackCode*0.002f/0.6f);
float cinWhite=powf(10.0f, whiteCode*0.002f/0.6f);
float cinWBdiff=cinWhite-cinBlack;
...

Now, I'm not pretending to understand C++ and whatever is going on in this code, I just wanted to look for any clue as to which cineon convert settings VFB uses.

So, to my problem when applying the same settings on a Cineon Converter plugin in After Effects, I get really different results then what Vray does, with the same convert settings:



However, eyeballing a little I can get closer:



But I never seem to get it exact.
I can only test this in After Effects, but I guess Cineon Conversion is the same across all grading applications. Or maybe each one handles this differently?

Anyway, any feedback on this would be much appreciated.

Thanks!

[Jorq1H]

Bumpin' this.

[vlado]

What is the respective conversion code in the Cineon Converter plugin in After Effects that you are using? If you didn't write it yourself, maybe ask the developers?

Best regards,
Vlado

[Jorq1H]

I used the default Cineon Converter filter in AE, but I can't tell what goes on behind the curtain there. As you can see from the screenshot, they use "standard linear to logarithmic conversion curves".
However, after setting their settings to the same as in your code, the result is way off, which leads me to believe gamma plays a huge part in the conversion.

So I'm wondering how your conversion handles gamma and why the results were so different when handling 32bit renders.

My goal here is obviously to create a project in AE which accurately mimics how the VFB applies LUTs, so that I can do my post-processing in the VFB and simply port the settings over when doing the final comp.