FS7: Shooting Custom Slog3 can we see a Look on a monitor out of camera?

Mitch's exaltation is excellent because again, one size does not fit all. Which curve "appears" better depends on the scene, and then depends where it will go in post.

Overall, it's true slog3 has less bits than slog2 does... but that is in an extreme case with a full 14 stop dynamic range scene. Maybe you shoot those, maybe you don't.

Slog3 however because it has more data at the lower end means theoretically if in post you were to do a big Exposure adjustment it would probably fair out better than slog2 would.

But again, do you go into a shoot saying "Hm I'm really gonna screw up this Exposure so is better shoot in slog3". Probably not.

I have personally moved to slog3 because of sgamut3.cine and that's way easier to grade for both me and any clients I have footage over to who are use to cineon based workflows....
 
starcentral said:
But again, do you go into a shoot saying "Hm I'm really gonna screw up this Exposure so is better shoot in slog3". Probably not.
Click to expand...

Yep you can do.. if you are used to shooting raw because you are completely evaluating the scene.. maybe you know you dont mind blowing your speculars then you could expose right up.. or to hold them you might iris right down - thats why Scurves (Hgammy) has always done me in.
 
So this suggests that S-Log3 is better for good shadow detail, given more levels to work with due to the way it's distributed and even though over a 14-stop range, it doesn't go all the way to 1023 bits white level.

I've gotten mixed results where, hand grading without a LUT, S-Log3 seemed noisier, but with a REC709 LUT applied in post, it was actually quieter than some S-Log2 footage, while having more shadow detail. Thus, I've switched to shooting S-Log3 much of the time. I do have some S-Log2 footage that's incredibly clean however.

The odd thing about the sensor in the FS7 is that the noise floor seems to come up as you close the iris. This, despite gain being set to a fixed level. I got around this by changing the ISO to 500, which pushed the noise below the point of visibility. But still curious why night scenes are noisy unless you go way down on ISO, while scenes with a range of tones from black to white, the black and near black areas seem totally noise-free. Almost seems like some sort of AGC is operating in the sensor, based on how much light it's receiving.
 
Hi Mitch

Just wonder are you doing your re rating in custom or Cine EI mode.. Im confused when you say you are losing the bottom two stops.. by over exposing and dragging down in post.. arnt you pulling dark area,s out of noise the first place.. ? .. thanks for the graph.. very interesting..
 
Donny 123 said:
Hi Mitch

Just wonder are you doing your re rating in custom or Cine EI mode.. Im confused when you say you are losing the bottom two stops.. by over exposing and dragging down in post.. arnt you pulling dark area,s out of noise the first place.. ? .. thanks for the graph.. very interesting..
Click to expand...
How would I be pulling the dark areas out? I guess it's all semantics, but here's what I'm talking about:

I generally would shoot Cine EI, S-Log3, Sgamma.cine, etc.
I would set the camera to 2000ISO because that's all I could do anyway.
Feed the signal to the Odyssey7Q+ and use an Exposure Compensation LUT to rerate -2 stops, the equivalent of 500ISO.
In post I would apply the same LUT, or at least use that as my guide and a starting point to whatever I might choose to actually do.

Since I use the LUTted signal to set exposure (I could just as easily use a lightmeter) but record either the S-Log3 or a RAW signal, I am effectively overexposing everything by two stops. I make sure to watch out that I don't let any highlights clip unless I don't care, but there's a lot of room up there to play. What I am essentially doing is a "controlled ETTR," where I am working in the upper part of the sensitivity range of the sensor, and I'm doing it in a consistent manner rather than the shot-by-shot method some employ on ETTR. (I think that method is completely inappropriate for motion picture imaging. It should only be used in stills where shots don't need to match one another and intercut seamlessly.) Working in the upper part of the range (and by upper part I only mean the top 12 out of 14 stops, so not that big a difference) means that my middle gray along with everything else gets two stops more light then it would otherwise. It means that I'm allotting more bits of information to the parts of the image that I'm going to use.

Now, of course the sensor is in fact going to expose those bottom two stops as well. But the first thing I'm going to do in post is apply the LUT or something close to it to darken everything and shove those two stops into blackness. Along with burying those two stops, all the rest of the image gets darker, and something that was captured at say four stops over the bottom and has a fair amount of noise now gets pushed down two stops darker and that noise becomes much less visible.

I guess semantics could say that because I am calling the sensor less sensitive I am therefore opening up the iris or adding light or whatever else I need to do in order to make the information hitting the sensor that much brighter, so I guess that could be construed as "pulling dark areas out of noise" but I don't see it that way. When I hear that phrase it sounds like something one does in post, and that is very much not the case here. You say potato I say potahto?

Basic formula: Expose brighter + post darker = less noise, better bit allotment, less dynamic range.

BTW, this can work in Custom just as well as Cine EI. In fact Custom is more made for this sort of thing in certain ways, but I record to an Odyssey7Q+ so I don't bother with that.
 
I would think the right exposure is the one in which the whites of importance are just under the sensor's clip limit. If you're able to 'overexpose' the image by two stops, it suggests you've been underexposing if you have headroom on the sensor and not the S-Log curve. Of course one can overexpose and obtain more shadow information above noise floor, but the whites get clipped out by the sensor and no manipulation with LUTs is going to bring that back because it was lost in the acquisition phase.
 
Basspig.. but you have so much high light latitude there is room to spare.. sacrifice a bit for a cleaner SNR .. a lot of people are working this way..

Mitch.. thanks for the run through of your set up.. yes semantics .. just getting better SNR..
 
basspig - just a thought re the mosquito noise in the EX3 footage recorded by external recorder.
If you had display info turned on in the menus but then turned off the display button on the outside of the camera to get a clean exported video you could get mosquito noise. Display must be turned off in the menus. That's how it was on the F3 anyway. Not sure about the Ex3.
 
Donny, I think clipped whites look ugly, therefore I try to avoid them when reasonable.

Lenilenapi, I had menus turned off, since SDI is the recorded output. I suspect the noise levels put the CODEC at a disadvantage.
 
You must log in or register to reply here.
Back
Top