A7s sample footage

I'm going to post this here since I don't have my FS700 to test against for my own thread.

Cine4:

1/640th, F/4.0, ISO 800
Cine4.png


Slog 2:

1/640th, F/4.0, ISO 3200
slog2.png



As you can see, Premiere shows Cine 4 saturating the entire signal (Blue line in the Y component) while Slog 2 only does from 10%-90% of the Y component. On the FS700 the Y component will go all the way to 100%, but it will never go down to 0%.

You can download the camera clips here and see if your software tells you something different, but these were handled natively by Premiere Pro CC 2014.

Obligatory Gif (256 color and will have banding)

cine4_vs_slog2.gif


Test setup: Sony A7s 1.0 firmware US model, 24-70 Zeiss f/4 OSS, Manual focus mode.
 
Thanks for this comparison and making original files available. Mostly Premiere user here, nothing to add. I'll load the S-Log2 file later and see what happens with some LUTs in Premiere and Resolve. Cheers

Edit: The S-Log2 tops out at 90 IRE, and it is clipping by the look of it, correct? Is this a likely firmware issue to fix? Thanks.
 
Last edited:
Christopher Barry said:
Edit: The S-Log2 tops out at 90 IRE, and it is clipping by the look of it, correct? Is this a likely firmware issue to fix? Thanks.
Click to expand...

If you look at the Green area 100 IRE is 235 in 8-bit. Most video cameras shoot within the 16-235 range. On the FS700 since it's a full range 8-bit camera you can have the green range go as low as -10 and as high as 110 IRE.
 
evanburke said:
If I buy this camera I'm not shooting S-Log2
Click to expand...


Waveforms are great but from what I've shot, I vastly prefer the look of the Slog2 files than any other profile so far. The latitude is much more akin to film and it shows in the final image. If I was controlling the light, I could see not needing every last ounce out of this thing but for natural light, it's been awesome. Also can confirm that SLOG2 is not clipping when coming into FCPX but Cine 4 requires some highlight saving (as Demistate had first posted). Corrected both.
1.jpg2.jpg
 
Last edited:
cyork said:
Waveforms are great but from what I've shot, I vastly prefer the look of the Slog2 files than any other profile so far. The latitude is much more akin to film and it shows in the final image.
Click to expand...

Yeah that's what I've been seeing too. Which PP are you using with S-Log? So many peeps testing, hard to keep track. I've been having issues with FCPX.

I just started an A7s post workflow thread http://www.dvxuser.com/V6/showthread.php?326181-A7s-post-workflow&p=1986457571#post1986457571 can everyone post their findings there, there's already been some interesting colour space stuff discussed in this and other threads.
 
Kholi, very useful results - you should get yourself one of these too, i think you deserve it :)

4K is very noisy, on the O7Q RAW as well, it's just the sensor. But 4K + noise reduction and decent downsample in post would give very nice 1080p images. What James Miller was doing strapping an AV 4K down converter onto the camera makes some sense although i think a better down converter is needed.

The question we have is very simple:

We need to understand whether the Slog2 original YUV signal is compromised in the XVAC container compared to the other profiles. Is the range cut down inside. Not what the various apps tell us because i don't think we can rely on any of them not to automagically do stuff.

So, Demistate, if you scaled that Slog signal to 0 to 109 would it be identical. Sadly it won't because one is Slog and the other is a Cine profile. So can't test that way.

There are two issues here:

1) How is the XAVC being natively recorded in Slog2 - the values in the MP4 file. (a quantifiable question)
2) When you do unwrap the log file (Slog2 curve or LUT back to linear) how much tonality is lost in the shadows and mids compared to a cine profile (a subjective question)

and then compared to the HDMI signal.

I need to find a way to get into the MP4 YUV data to see the real luma values...

cheers
Paul
 
Demistate said:
If you look at the Green area 100 IRE is 235 in 8-bit. Most video cameras shoot within the 16-235 range. On the FS700 since it's a full range 8-bit camera you can have the green range go as low as -10 and as high as 110 IRE.
Click to expand...

I am not sure what values the a7S S-Log2 should be producing on the scopes, interpretation is a work in progress, as discussed yesterday in this thread. Thanks, Demistate.
 
squig said:
HDMI out: YCbCr 4:2:2 8bit/ RGB 8bit. https://developer.sony.com/devices/cameras/a7s/
Click to expand...

Not very helpful from Sony there. The camera settings just have 1080i or 1080p, nothing about format out. But it does introduce a new question that whether the external recorder can choose what format it requires.

AFAIK YUV can actually hold an image better than RGB.

Maybe the fact that i've seen the external recorder not produce quite so much range is down to this format question, i will find out what the recorder is asking for, maybe that will help us

cheers
Paul
 
I'll be recording into a pix220 tomorrow, it records prores4444 RGB and YCbCr, it's not switchable so we'll see what it picks up.
 
squig said:
I'll be recording into a pix220 tomorrow, it records prores4444 RGB and YCbCr, it's not switchable so we'll see what it picks up.
Click to expand...

The recorder just says 1080psf 25 4:2:2, now AFAIK the 422 must mean YUV? Because isn't RGB 444 by definition? So that's worth looking into, if the recorder can be made to request RGB i wonder if it would get full 444 sampling...

cheers
Paul
 
paulcurtis said:
The recorder just says 1080psf 25 4:2:2, now AFAIK the 422 must mean YUV? Because isn't RGB 444 by definition? So that's worth looking into, if the recorder can be made to request RGB i wonder if it would get full 444 sampling...

cheers
Paul
Click to expand...

Yep RGB is 444
 
You must log in or register to reply here.
Back
Top