thank you for the LUT, Kala!
I ran it through the analyzer and indeed I was right in not releasing the LUT I made a long time ago for D-log.
- Screenshot from 2024-07-01 12-50-35.png (24.18 KiB) Viewed 4178 times
- Screenshot from 2024-07-01 12-52-34.png (27.08 KiB) Viewed 4178 times
The first one is the one made by the manufacturer that you shared, while the second one was the one we produced initially when we didn't have any reference.
As you can see, the one from the manufacturer starts much higher with the blacks, thus preserving them and it's also shifted to the right with the knee start level raising a whole stop later.
Both flattened at 6 stops which is 100 nits for BT709 SDR.
The final LUT result was therefore different:
- Screenshot from 2024-07-01 13-01-13.png (22.03 KiB) Viewed 4178 times
- Screenshot from 2024-07-01 13-01-23.png (23.64 KiB) Viewed 4178 times
For reasons everyone would probably understand, we can't ship any manufacturer-created LUT, so we had to create our own which are shared publicly for free just like FFAStrans.
I'd like to ask you one additional big favor: can you test the following LUT I'm about to share in attachment and let me know how it goes when you apply it on a real D-Log footage you have compared to the manufacturer LUT?
Please note that it's my second attempt and it might not work correctly.
I still think I haven't got the black gamma right and in fact you're gonna see a lot of zero at the very beginning of the LUT as those resulted in negative values which I zeroed out to avoid math problems, but please let me know.
I will almost definitely release a v3 version.
If you could also share a sample privately that would be awesome.
Hopefully, at the end of this, we're gonna get a proper LUT and - perhaps - D-Log support in a future version of FFAStrans.
