April 15, 2016 at 5:33AM, Edited April 15, 6:20AM

6

S-LOG 2 or S-LOg 3 for grading?

I have purchused the Sony a6300, after using BMPCC using S-LOG is all new to me. So the question is which of these two picture profiles preserves more dinamic range, data and detail for better grading in post?

11 Comments

Long answer: http://www.liftgammagain.com/forum/index.php?threads/sony-f5-s-log-2-vs-...

Short answer: S-Log3 is closer to Cineon curve, so if that matters to you, choose that.

April 15, 2016 at 8:04AM, Edited April 15, 8:04AM

0
Reply

That’s a lot of useful information. Thanks.

Ivan (Fidel) Fedorych

April 15, 2016 at 3:39PM

Anyone can correct me if I am wrong but the dynamic range of a camera is primarily determined by the sensor. A higher ISO boosts brightness in the lows and, heaven forbid, could even clip the highlights but raising the ISO comes with the price of having more noise. And finally a CODEC may not handle a super flat profile very well because a lot of compression is perceptual compression and this is especially true if the CODEC is 8 bits.

So I think the bottom line is that encoding your video flat can help with color grading but will hardly do anything for extending the dynamic range.

But anyone feel free to correct me if you think I am wrong about it.

April 15, 2016 at 12:25PM, Edited April 15, 1:06PM

0
Reply
avatar
Cary Knoop
Member
2111

You are right in what you're saying about a high compression codec not doing you any favors when shooting flat, but shooting flat does extend your dynamic range. When not shooting flat you're not using all of the dynamic range that the sensor produces, since doing this would result in a dull (flat) and uninteresting picture (and what consumer would want that?), so the standard profiles compress the image into a 6-8 stop high contrast and "visually pleasing" image. A flat profile will take the highlight and shadow information and shift them towards the mids to allow more of that sensor data to be captured. In post we then grade the image back into a 'low dynamic range' image with higher contrast and saturation, but the difference is that now we have access to that extra highlight and shadow information and can tweak it more to our liking (eg. maintain some extra highlight or shadow detail).

Tobias N

April 17, 2016 at 5:56AM

I am not fully agreeing with this.

Would you agree with any of the following statements?

1. Profiles do not influence what a sensor can and cannot capture.
2. Unless a profile actually clips data from the sensor the dynamic range remains the same, all the profile does is to present a different tonal range to the encoder.
3. In 8-bit representation any tonal range is a compromise. If a given profile maintains more information in some range it must lose some in another range. It's a zero sum game.
4. A profile will actually destroy information at any place in the curve where it tries to record tonal information that is beyond the capability of the sensor.

April 17, 2016 at 12:17PM, Edited April 17, 1:08PM

0
Reply
avatar
Cary Knoop
Member
2111

1. Not quite. Shooting with a Rec709-profile (or most Standard profiles) captures around 6-7 stops of dynamic range. You normally lose some of that usable dynamic range for a higher contrast image.
2. Punchy profiles clip - Flat profiles squeeze.
3/4. Flat profiles definitely compromise your tonal precision, but it does preserve some of that usable dynamic range that would otherwise not be represented in the captured footage.

This said, I am a strong proponent of capturing the look directly in-camera whenever possible. The absolutely best and cleanest image you will capture in an 8-bit camera will be the one you have to touch the least in post.

Tobias N

April 17, 2016 at 3:02PM, Edited April 17, 3:02PM

I do not think that modern video cameras (HD and 4k) actually record according to Rec 709, they just produce an image that fits Rec 709. Even most news cameras do not record pure Rec 709.

Nevertheless you cannot influence what the sensor captures you can only interpret the linear output by assigning custom brightness levels.

Cary Knoop

April 17, 2016 at 3:39PM

For instance take the GH4, here is an article comparing the dynamic range for all profiles:

http://www.dvinfo.net/article/acquisition/micro-43/panasonic-gh4-dr-and-gamma-timelapse-and-more.html

Conclusion the dynamic range does not change at all with changing profiles with the exception of Cine-V and Vivid that clip one stop.

Cary Knoop

April 17, 2016 at 4:06PM, Edited April 17, 4:06PM

Does S-log 3 really add noise to shadows and blacks, and how bad it is?

April 17, 2016 at 3:40PM

0
Reply
avatar
Ivan (Fidel) Fedorych
Director, DP
127

It doesn't actually add noise but it gives you that extra shadow detail; which is quite noisy. You can also just grade the shadows back down if you don't need the information.

Tobias N

April 17, 2016 at 3:54PM

Slog3 is not more noisy it just raises more shadow and therefore keeps a bit more details in it than slog2 but the idea is to put down those raised shadows so the black is black. Here is the detail : http://www.xdcam-user.com/2014/03/understanding-sonys-slog3-it-isnt-real...
Also for the first question people agree that slog3 is a bit better for grading than slog2

April 18, 2016 at 8:31AM

0
Reply
AvdS
1288

Your Comment