Sony F3 and S-Log


With its large sensor and PL mount it is easy to be seduced by the Sony F3. That is until you run into highlight clipping and a dynamic range that reminds you that this is after all an HD video camera. Even with Sony’s hyper-gamma curves enabled and attaching a recorder such as the Nano flash the camera is basically an EX 3 with a big chip.
It was no secret that Sony intended to release a firmware upgrade that added S-log and 4:4:4 output to the camera, and when it was introduced a couple of months ago, Sony offered a camera to the NZCS for testing.
Thom Burstyn fresh from having shot ‘Underbelly’ with the F3 was asked to supervise the testing and together with myself plus Scott and Clive from Sony we put the camera through its paces, comparing all its configurations including basic 709 video mode, hyper gamma mode and S-log mode.

So what exactly is S-Log?  
We know that compared to digital sensors, our eyes (and film) are much better at coping with high dynamic range. Highlights do not blow out and shadows don’t suddenly block up. This is because our eyes and film respond in a compressive non-linear (logarithmic) way. Look at a film emulsions characteristic curve and where you see the toe and shoulder you are looking at a form of non-linear compression.
A CCD or CMOS sensor responds to the light in a linear fashion, output voltages produced by a camera are proportional to the light intensity hitting the sensor.  That means that as the sensor fills up with light, changes in the amount of light being collected by the sensor are always perceived in the same way.
Or slightly more scientific… On a linear scale, a change between two values is perceived on the basis of the difference between the values. Thus, for example, a change from 1 to 2 would be perceived as the same amount of increase as from 4 to 5.
On a logarithmic scale, a change between two values is perceived on the basis of the ratio of the two values. That is a change from 1 to 2 (ratio1:2) is perceived as the same amount of increase as a change from 4 to 8 (also a ratio of 1:2)

Remapping something in log space is a way of manipulating data. It gives us the ability to squeeze (compress) more stuff into the same amount of space. If we could take the linear sensor data and convert it to non-linear (log) data, we should then be able to increase the dynamic range recorded by the camera.
So how do we convert our linear image to a log image?  We use what is known as a Gamma function, and you guessed it,  S-Log is a Gamma function.

If you read the official Sony literature it explains that ‘the S-Log curve provides sufficient quantization bits to offer exceptional tonal reproduction in both the low light and high light region’.
If like me even just trying to say quantization hurts your brain, don’t panic. What we as cinematographers need to know is that shooting in log means capturing greater dynamic range… and that is really all we need to know. *
Using a camera in log mode, does necessitate the use of a viewing LUT for accurate monitoring on set as the outputted image is low in contrast, and it does mean that some colour correction is vital as part of the rushes / editorial process (as if it wasn’t before). However the bottom line is that once an image is captured in log it can be post processed with similar techniques as those employed for film originated material, making the conceptual leap to shooting log far easier to make for the, (cough), more mature cinematographer.
Since S-Log can supposedly reproduce the entire tonal range captured by the camera sensor it is sometimes referred to as a ‘digital negative’. Personally, with all the processing involved I prefer to think of it more as a ‘low contrast digital print’, leaving the digital negative title for RAW.

So back to the test, how much extra dynamic range does the S-Log curve allow you to capture?  The short answer is a lot!  
Rather than try to describe the differences between the various shooting modes of the Sony F3, below are a selection of frame grabs photographed as part of the testing.
 As always, test for yourself.  However these images are a great starting point for your venture into S-log shooting.

*For the pointy heads in the audience, quantization is the process of converting a continuous range of values into a finite range of discreet values and is a function of analog to digital converters. The converter creates a series of digital values to represent the original analog signal. Bit depth determines the accuracy and quality of the quantized value.

S-Log 5.jpg

Unfortunately, the cine-gamma shot of the model was unusable due to fluctuations in the sun during testing; needless to say the S-Log allowed for much more information in the highlights. My feeling over all was that the S-Log affected the mid, 1/4 tones and highlights much more aggressively than the ¾ tones and blacks.
An interesting aside is the difference between turning Auto Knee on and off. Clive the engineer from Sony whose brain is like Marvin’s from ‘The Hitchhikers Guide to the Galaxy’ explained it in a way that seemed entirely obvious but I still really have no idea why it was so… however the visual difference was quite surprising… I do believe that using Auto Knee in more high contrast situations will give different results… so as always… test, test, test.
Bottom line... is  S-Log a worthwhile upgrade for the F3?… Absolutely, it is a completely new camera with the addition of S-Log. Not to mention the 4:4:4 recording that also comes with the firmware upgrade. This is no longer an EX-3 with a big chip.