Using Zebras when measuring S-log3.

[alisterchapman]

If you use Zebras to measure the levels of an S-Log3 image then you should consider using a narrower Zebra range or window than the normal default 10% (+5% / -5%).

Zebras were originally conceived for use with Rec-709 and other normal gamma curves. With Rec-709 above middle grey each stop of exposure occupies around 16% of the recording range. So when using Zebras with a 10% or -5% to +5% window to measure a specific value, perhaps a white card or grey card on average you will be accurate to around +/- 1/3rd of a stop. On a grey card if you have set your zebras to 50% your zebras would start to appear at 45% and disappear at 55%, you will have an error of under half a stop at most. This is reasonable and when measuring something like skin tones which may cover a large range anyway works perfectly fine.

But when using Zebras to measure the levels of S-Log3 you need to use a much narrower range to achieve the same accuracy. Between middle grey and white each stop of the S-Log3 curve only occupies approx 8.5% of the recording range. So, if you use the default 10% zebra range your accuracy will at best be +/- 3/4 of stop. With the default -5% to +5% range if you set your zebras to 41% to measure the level of a grey card Zebras will appear at 36% , almost a stop under exposed and go all the way up to 46%, almost 3/4 of a stop over. With a zebra window 1.2 stops wide thats a lot of potential for incorrect exposure.

I would suggest that when using Zebras to measure S-Log3 you should reduce the range of the zebra window to -3% +3% to achieve a similar level of accuracy as you would have when measuring rec-709. If you are only using Zebras to determine the exposure of a white card or grey card you may want to go all the way down to -1% +1% if you want even better accuracy.