Gemini 4:4:4 frame Grabs

[Billy Steinberg]

Just as one shouldn't "look down" on getting the pictures right in the first place, one also shouldn't look down on being able to take advantage of what S-Log provides. I'm not trying to play the devil's advocate to my previous message about Gamma Curves and Dynamic Range, but I do appreciate that having the ability to utilize the sensor's full range after the fact is very powerful stuff...

Being given the choice is always the best way, in my book.

Billy

[Brian Drysdale]

The camera does give you the option of going either way, depending on your workflow of choice and your schedule.

[Alister Chapman]

Originally Posted by Billy Steinberg

While one might run into banding/quantization when grossly stretching out a section of video, the same doesn't happen when shrinking it (pulling the black levels down). In fact, the opposite happens. And the cameras I work with have 12 (and occasionally 14) bit data, not 8 bit. We are talking about Hyper (Cine) gamma here, which happens in the camera.

When you pull the blacks down with an 8 bit recording as in the F3 internal recordings, you will often get banding in the area of the image just above where you have done the pull down as you will create a bit starved area between your original exposure and the parts of the image (blacks/mids) where you have reduced your levels and this can lead to banding. When stretching or squeezing different parts of an 8 bit image you have to consider what is happening around the periphery of the manipulated areas.

If you have 100 even spaced grey shades and then take the bottom 50 and crush them to fit in the space occupied by 25 original steps, then there must either be a big jump between level 25 and 50 or levels 50 to 100 are now stretched to fill 25 to 100. In either case banding can become an issue.

[Billy Steinberg]

Absolutely true; when you compress in one area, you expand in another. And this is why it's semi-worthless to shoot in S-Log and record in 8-bit, with the expectation to make things wonderful in post.

But regardless of whether you record internally in the F3 (which I am not familiar with, but I trust that you are, and that the recordings are indeed 8-bit), when you are using Hyper (Cine) Gamma, you are working with at least 10 bits, probably 12, maybe even 14, but certainly not 8; the 10 (or 12/14) bit world of the camera part of the F3 is where you're making the changes, long before the video is recorded, and there's plenty of information to stretch and compress. And since you're making the alterations in the camera, you're not likely to have to make huge gamma changes in post, meaning even an 8-bit recording should look fine, with no banding or quantization artifacts.

Not only are Hyper (Cine) Gammas much less severe than S-Log, and hence require less stretching and compressing, but on rare occasion I have done some extreme compressions with zero banding or quantization in the areas (35%-65% input levels) above it. (-40 to -50 setting of the black gamma, bringing the 35% area back down 15% to make up for the lift of one of the more severe Hyper (Cine) gammas).

Billy