A7s: S-log2 too noisy? Then underexpose cine4

For me PP1 is my personal variant of slog2 (which I call Flaat) and PP2 is a monochrome profile that I use only to set WB.
 
Just going back to the original post. Initially, I too found Slog2 too noisy, and only used cine2 or cine4. But after a lot of testing and field work, I now LOVE SLOG2! Consider me a SLOG2 fanboy. Especially after getting the 7Q+, using the waveform, the highlights just keep going up and up and up. After pushing the exposure up +2 stops, there is no noise, even in 4K. I found good results with Alisters AC A7s Film-Like 2-OVER LUT. Quinte Conservation Permits & Regulations.jpg
 
re_curly81 said:
Just going back to the original post. Initially, I too found Slog2 too noisy, and only used cine2 or cine4. But after a lot of testing and field work, I now LOVE SLOG2! Consider me a SLOG2 fanboy. Especially after getting the 7Q+, using the waveform, the highlights just keep going up and up and up. After pushing the exposure up +2 stops, there is no noise, even in 4K. I found good results with Alisters AC A7s Film-Like 2-OVER LUT. View attachment 101402
Click to expand...

Overexposing slog by two stops is basically mapping the mid grey in a higher zone, therefore losing highliggts. Of course the noise will be buried deep down the shadows but then you end up being with a similor dr ratio as using cine4. So not big of an advantage there.
 
macgregor said:
Overexposing slog by two stops is basically mapping the mid grey in a higher zone, therefore losing highliggts. Of course the noise will be buried deep down the shadows but then you end up being with a similor dr ratio as using cine4. So not big of an advantage there.
Click to expand...

Well... you have to be aware that S-Log2 has 6+ stops of latitude above it's "mid grey". Stills cameras only have 4 stops above mid grey. Therefore, "over" exposing the A7s in S-log2 by 2 stops, brings you right back to "normal" stills territory. IMHO Sony had to rate the A7s ISO3200 to fulfill the S-Log2 DR specs. Practically I'd just rate her at ISO800.

Just my 2c,
Michael
 
macgregor said:
Are you telling me or just quoting my graph in the 1st post?
Click to expand...

Actually I'm referring to this chart:
S-Log2vsS-Log3.jpg

... and am speaking of my experience contributing the S-Log2 code for MlRawViewer.

macgregor said:
Overexposing slog by two stops is basically mapping the mid grey in a higher zone, therefore losing highliggts. Of course the noise will be buried deep down the shadows but then you end up being with a similor dr ratio as using cine4. So not big of an advantage there.
Click to expand...

The advantage of S-Log2 vs Cine(1-4) is, that you can linearize the signal again, do an exact exposure correction in linear light, and then apply whatever curve you want afterwards (in Resolve). Also S-Log2 has the most code values in the first stops after clipping:
SLog2 Stops.jpg
This is especially important for the A7s, since we are recording an 8 bit signal and need all the code values per stop that we can get!
 
macgregor said:
OK so where is the info about ISO 640?
Click to expand...

Could you further explain what you mean?
Do you mean
1) the real native ISO (S-Log2 has about 6 1/3 stops above grey) or
2) underexposing cine4 by 1 2/3 like you mentioned, and treating it as 640?

If you need the "extra" latitude and a lower ISO, sure, you can do it that way. I like to be able to linearize my footage and make exposure adjustments easily though.
Also could you explain why you need so much DR in the hightlights? As a photographer I only need that if I shoot against the sun/light. I'm fine usually with the 4 stops normal camera raw gives me!
 
MichaelFischer.at said:
The advantage of S-Log2 vs Cine(1-4) is, that you can linearize the signal again, do an exact exposure correction in linear light, and then apply whatever curve you want afterwards (in Resolve). Also S-Log2 has the most code values in the first stops after clipping:
View attachment 101440
This is especially important for the A7s, since we are recording an 8 bit signal and need all the code values per stop that we can get!
Click to expand...
This is contradicted by Alister Chapman's a7s article here:
http://www.xdcam-user.com/2014/08/e...-on-the-sony-a7s-part-one-gamma-and-exposure/
In the section on over exposing s-log2 he appears to demonstrate that pushing more than 2 stops over will start to reduce fine detail and texture (in an 8bit recording).
I haven't tested this myself. I've been shooting cine4 to save the headache :)
 
bags said:
This is contradicted by Alister Chapman's a7s article here:
http://www.xdcam-user.com/2014/08/e...-on-the-sony-a7s-part-one-gamma-and-exposure/
In the section on over exposing s-log2 he appears to demonstrate that pushing more than 2 stops over will start to reduce fine detail and texture (in an 8bit recording).
I haven't tested this myself. I've been shooting cine4 to save the headache :)
Click to expand...

I really can't see that in his test, do you see it? I don't see any loss of detail...
 
MichaelFischer.at said:
I couldn't see it either in his tests or in mine. The only thing that changed were the noise levels.
Click to expand...

I think sometimes we're just looking at the trees and not the forest

I've side by sided Slog vs Cine and tried to break both to find the differences.

First, if you expose like for like the noise is the same. Second this is 8 bit in a YUV container, not RGB. It's surprisingly robust. Third, if you can take it out of the HDMI because there seems to be a bit more detail there. (If you can get UHD out and downsample to HD yourself, that's pretty damn good)

Basically whether you choose cine or slog then the sensor and noise are the same, it's just the method of getting it out of the camera. If you swap between either of those, assuming the same exposure, they're near identical which is what the OP was basically getting at. I think your choice should be dictated by your post workflow.

You can get Slog2 very wrong and also a lot of applications don't immediately show full range. So if you're into using LUTs then you may be using them wrong. For example PPro hides the superbright info, your highlights are clipped by default. If you pop a LUT on that is clamped 0 to 1, so actually (assuming the LUT is correct) you're loosing highlight detail. If you bring down the exposure in PPro first, to ensure your superbrights are in range and then apply the LUT that is much better. But some applications don't even give you the superbrights, SpeedGrade for example appears to never allow you to get that. I've side by sided various apps with my control which is using Nuke to get the raw YUV data and convert to RGB myself, that way i can see every last pixel in the files. So i know when there are highlights missing. One day i'll write this up. Our blog does have an article about noise and SLog2 though.

The other caveat is that the Slog2 out of the HDMI is not standard, the black levels are very different. So again if you just dump a LUT over it then the starting point is wrong, therefore your result will be too.

Hence my comment that if you can work with Slog2 properly that's (my anyway) preferred route....

cheers
Paul
 
paulcurtis said:
The other caveat is that the Slog2 out of the HDMI is not standard, the black levels are very different. So again if you just dump a LUT over it then the starting point is wrong, therefore your result will be too.
Click to expand...

Hi Paul,
first of all, cool article about the A7s and the Odyssey, which I already read earlier! The reduced banding when recording UHD and downsampling to HD is exectly what I was hoping for :grin:

Could it be that the A7s transmits the Signal over HDMI in full range (0-255) and the Odyssey is flagging it as "TV Levels" (16-235) which makes is more contrasty at playback?
http://en.wikipedia.org/wiki/YUV#Numerical_approximations
A simple "Levels" correction (input 0, 1.0, 255, output 16-235) should fix it.
Could you try and let us know?

Even tough we live in the digital age, we still have this issue and even Windows 10, which I tried recently, displays black (0) as grey (16), if you use incorrect settings...

BR
Michael
 
MichaelFischer.at said:
I couldn't see it either in his tests or in mine. The only thing that changed were the noise levels.
Click to expand...
I see loss of details on the highlights. Wooden peg on his right.

its minor. but i expose using false color. Always kept skin in 40-50 ire and i never had a noisy skin. Shadows can be noisy and really who cares?
 
Last edited:
I am quite new into color grading, but I want to get deep into it. I just bought the color correction handbook. I've read your comments because I have the same trouble with my a7s, but there are so many things I dont't understand yet. Could someone tell me which would be a good starting point? Thanks! sorry for my bad english
 
I've read this thread with interest, as an A7Rii owner for just couple of weeks. From my own experience so far, nothing I do to Cine curves (especially Cine4) can beat what S-log2 is the best at: ability to preserve highlights (like the blue sky with white clouds), even if I overexpose the S-log2 by 2 stops to bring it don in Resolve and get rid of some mid-and-low area noise. Compared to S-log2, all my Cine-gamma based experiments failed to preserve those white sky and particularly the blue color of the sky! HOWEVER - and I wonder why nobody even mentions it in this thread - the 8bit nature of A7 series cameras make S-log so prone to banding (like in the blue sky), that I rarely use S-log2 anymore! Can you explain to me why blue sky (or white wall, for that matter) is not banding at all (provide of course that it's not totally blown out) when using Cine gammas, and banding like hell with S-log2, practically making it unusable ? TIA

Piotr
 
<<Compared to S-log2, all my Cine-gamma based experiments failed to preserve those white sky and particularly the blue color of the sky!>>

I hate to say it but if you're blowing highlight detail in cine 4 you aren't exposing correctly:
https://vimeo.com/139061000

 
You must log in or register to reply here.
Back
Top