AbelCine S-Log on the F3, part 2 with EI

This was very interesting to watch (part 2). I had thought that the EI (exposure index) was supposed to increase the dynamic range from the prior version by recording at essentially ISO 800 instead of 1600 with the EI giving a more accurate view of the final product. However, in the Andy video, he seems to say that there is in fact no change at all in the s-Log output from the prior version.
 
lasikdenver said:
This was very interesting to watch (part 2). I had thought that the EI (exposure index) was supposed to increase the dynamic range from the prior version by recording at essentially ISO 800 instead of 1600 with the EI giving a more accurate view of the final product. However, in the Andy video, he seems to say that there is in fact no change at all in the s-Log output from the prior version.
Click to expand...

The initial hype is kinda right, but the end fact that EI mode doesn't touch the S-Log output on the dual link ports is correct.

At 800ISO, the sensor is at it's 'optimum'. It has the most latitude above and below midpoint. When you add traditional gain, you are of course making a dark image brighter, but you are also trading highlight latitude for shadow detail...the latitude goes lopsided so to speak. This is true with all cameras, it's just the nature of adding gain.

So the supposition is that if you keep the sensor at 800 at all times, record that, and do your gain operations in post (color-correction), you can have much greater control of where that 12 stops clips. EI mode is simply just a way to facilitate that: Change the monitoring LUTs to suit a given ISO rating, always record the sensor at 800ISO to a high quality medium, and boost that image brightness in post. Which, by the way, is not a bad way to be if you are recording S-Log 10-bit to uncompressed or close to it.

For the average joe, EI mode is non-player. It's just easier to keep it off, add gain when you need it, and call it a day. It's really for shoot where you need to match F35 workflows and/or you got the outboard recorder and post path to make it happen. I have all those things to make it viable, and I still couldn't care less it's available and won't be using it.

[edit: in practice, right now, it's only usable for folks that are recording S-Log via the dual link ports, i.e. Gemini folks or SR-R1 folks. The rest of us recording S-Log to Pix240s and Samurais have to use the HD-SDI port intended for monitoring, which kills the whole point of EI mode because we have to shoot/record with LUTs off]
 
Last edited:
But just to add to Nate's last statement, sometime in January we should see firmware rev 1.3 which will allow an S-Log 4:2:2 output via the A port, allowing LUT monitoring from the SDI port. But still, EI mode is only for LUT viewing. Nate's explanation is excellent.

Ned Soltz
 
I thought it was interesting that Andy admitted that he had mismarked middle grey in part 1. That throws the earleir interpretation off.

Since EI won't make much difference for most of us until we have v1.3 it will be interesting to see what he says in part 3 when testing.
 
Hey guys

Thanks for checking out the videos. Sounds like you all picked what I was saying about S-Log changing, from first descriptions by Sony it sounded like EI would increase the camera range. However this was not really the case. So EI won't do much for most people until vs 1.3 of the firmware hits later this month.

I'm waiting for a nice bright day to shoot part 3, but it should happen this week. I'll be recording both internally and to the Gemini444, which in the future you could substitute for any 422 10bit recorder. This way we can see what Slog uncompressed gives us compared to the LUTed internal recording.

Happy New Year to all,

Andy
 
I find it even slightly more amusing that the firmware description touts the advantages of EI mode, but no official LUTs exist for S-Log from Sony from what I've seen, much less an entire set of LUTs that you would need for every ISO to do EI mode right.

I paid £30-ish to have somebody make an S-Log LUT for me in Europe that I don't even use because I can make things look better faster without it in Resolve.
 
Nate

I mentioned this to Sony as well. Arri's LUT generator has it's faults, but at least you can get different LUTs sat different EI modes. Any LUT would be just be a starting point for Resolve, but they are still important. I plan on making some available as part of the series.

Andy
 
Sony's early description led a lot of people to believe that EI was something entirely different than what it was. Thank god we have people that actually test things to see what's actually happening.

My real guess is that by shooting in 800 ISO there will be so little noise that footage can be graded to get that little bit of extra detail out of it. However, the chip's DR is the chip's DR.

EI will just let us have an idea of what else is going on in high contrast scenes, where when we set exposure to protect the highlights the darks and mids look crushed together on the monitor until we fiddle with it in the edit bay.
 
Last edited:
Duke M. said:
My real guess is that by shooting in 800 ISO there will be so little noise that footage can be graded to get that little bit of extra detail out of it. However, the chip's DR is the chip's DR.
Click to expand...

There was a long discussion on Reduser some years ago with Graeme discussing the pros and cons of applying gain AFTER analog to digital conversion. The short story is that if you record the digitized signal with enough bit depth (10bit vs 12 bit vs 16bit), you record the noise floor of the chip with enough fidelity that it doesn't matter where you make the manipulation in the chain.

My quick playing around with lifting 800ISO S-Log that was 3 stops underexposed in post, it seemed it was dirtier than in camera gain. But I need to revisit it, don't take that statement as real fact. It could be that S-Log to 10bit ProRes isn't enough. Might have to be 10bit uncompressed.
 
I think at 3 stops you were into the noise floor. The description was that it would give you another stop. Even half a stop would be good.
 
Duke M. said:
I think at 3 stops you were into the noise floor. The description was that it would give you another stop. Even half a stop would be good.
Click to expand...

Very much into the noise floor. But you understand what I was trying to say/do, right?

EI mode would have you set it to, say, 3200 ASA (a two stop 'push'). So your monitoring LUT shows you what the camera output would be like if you gained up to 3200, but your dual-link S-Log output is two stops underexposed, leaving it for you to lift 2 stops in post.

So 2 stops or 3, your mid tones are yes, likely into the noise floor of the camera.
 
You must log in or register to reply here.
Back
Top