FS7: Slog3 clipping level

L

Ljung

Member
I've noticed something that strikes me as odd: When shooting Slog3 with the FS7, the camera clips whites at 93 IRE. I've confirmed this both with a waveform on an external monitor and with the camera's own zebras; I can't get a white higher than 93. (I'm in CineEI mode, at 2000 ISO, with no LUT applied.) I even brought some footage into Resolve and looked at the scopes there; white clips before 896 instead of going up to 1024 like you might expect. Anyone know why this is the case? I would think Slog3 should go up to 109 IRE (or 1024 digital units) to take advantage of all available bits... Why keep the whitest white slightly gray? Have others found this to be the case as well, or is there some odd setting buried somewhere that I'm missing?
 
61 is the recommended value to expose a white card leaving some space for speculars it is not the actual clip point
 
Last edited:
Point being if 61% is a white point (plus space from some speculars) then it's fairly likely something exposed at 109 would be clipping like crazy. No?
 
I think it was explained to me that in order the follow the cineon curve it had to be that way as Slog3 like Cineon is designed to go beyond 14 stops to be used in future higher DR cameras and that is why the room is preserved. I don't remember where I read this so I can't post the article. It might have been when I was talking to Alister Chapman at the Prfusion expo in Toronto.
 
A white card with 90% reflectance is 61% IRE in slog3. After that, the camera in slog2 or slog3 mode captures levels 14.4 times brighter than that white card which is incredible… except when its not completely needed.

As for how data gets distributed Sony's own white paper on slog3 says "Data range is used from 0 to 1023 (from -7IRE to 109IRE)" so I'm not sure why it clips around 94%.

What could make a bit of sense is if this is how "other" manufacturers map to the cineon gamma curve with their cameras.
 
Ben Scott said:
Point being if 61% is a white point (plus space from some speculars) then it's fairly likely something exposed at 109 would be clipping like crazy. No?
Click to expand...

Lets say you are shooting a wide of an INT and there are LED spots on the ceiling - the 61-100 range should enable that you wont get the horrid halos round the LEDs (even if the globes themselves clip) im sure those horrid halos and thier opposite - the alexa look - are about dedicating enough space to that roll to white, which is what the 61-100 element of the file looks after.
 
starcentral said:
I'm not sure why it clips around 94%.
Click to expand...

I would still think that it is not making best use of the 1024 levels and wonder of another (Slog2 as Doug has suggested) would make better use of the sensor (if you can nicely un-fungle Slog2 which 90% of shooters cant!)

We have the first 9bit camera?

S
 
morgan_moore said:
Lets say you are shooting a wide of an INT and there are LED spots on the ceiling - the 61-100 range should enable that you wont get the horrid halos round the LEDs (even if the globes themselves clip) im sure those horrid halos and thier opposite - the alexa look - are about dedicating enough space to that roll to white, which is what the 61-100 element of the file looks after.
Click to expand...

My point was that with 90% reflective white at 61% and the camera's dynamic range mapped into S-Log3 it really doesn't come as any major surprise to find 94% clipping
 
So is it best to set Zebras at 90 for S-Log 3?

Does it go by increments of 5 on the FS7? So if zebra was at 95, it might clip a bit?
 
Ljung said:
I've noticed something that strikes me as odd: When shooting Slog3 with the FS7, the camera clips whites at 93 IRE. I've confirmed this both with a waveform on an external monitor and with the camera's own zebras; I can't get a white higher than 93. (I'm in CineEI mode, at 2000 ISO, with no LUT applied.) I even brought some footage into Resolve and looked at the scopes there; white clips before 896 instead of going up to 1024 like you might expect. Anyone know why this is the case? I would think Slog3 should go up to 109 IRE (or 1024 digital units) to take advantage of all available bits... Why keep the whitest white slightly gray? Have others found this to be the case as well, or is there some odd setting buried somewhere that I'm missing?
Click to expand...

In Resolve do you have the project set to "Data" or "Video" levels? and do you have the Clip Attributes for every shot in your timeline set to Data or Video?
 
Ben Scott said:
My point was that with 90% reflective white at 61% and the camera's dynamic range mapped into S-Log3 it really doesn't come as any major surprise to find 94% clipping
Click to expand...

Observe the WFM on the camera - you cannot go above about 95 or so if you point the camera at a very light thing and open right up.

--
When you apply a LUT in your NLE you would indeed expect stuff exposed much above 65 to go white - but then you can pull it back with curves or whatver and get nice highlight rolloff (which is what the LC709 lut does anyway)

S
 
Since EI mode shifts the middle grey point, you will notice the clipping point go up and down as you increase and decrease from EI2000.

for example at EI500, white clips at about 85% IRE.
 
marvinhello said:
Since EI mode shifts the middle grey point, you will notice the clipping point go up and down as you increase and decrease from EI2000.

for example at EI500, white clips at about 85% IRE.
Click to expand...

No it won't. That is completely wrong. The underlying slog3 will remain exactly the same.
 
Lots of people here seem confused between the IRE levels for slog and the IRE levels displayed by a LUT. Two very different things.
 
Liam Hall said:
Lots of people here seem confused between the IRE levels for slog and the IRE levels displayed by a LUT. Two very different things.
Click to expand...

The op mentioned there's no LUT applied.

Alexa LogC for example, if you want white to clip at 109%IRE you have to use EI1600 or above, at EI800 or lower white clips well below 109%.

Same applies for Slog
 
You must log in or register to reply here.
Back
Top