What's the difference? LOG vs RAW footage for color grading

May 30, 2022
What's the difference? LOG vs RAW footage for color grading
Posted in: All Film Science
Better films every fortnight. Subscribe to Syrp Lab YouTube.

You’ve just got your footage back, you start to grade and everything just breaks… How do you fix that? We’ve all been there and oh boy - it doesn’t feel great.

In this episode of Film Science we deep dive into the science of LOG and RAW footage and why sometimes it just looks trash.

So to understand bad footage, we first need to understand how your camera actually works and what it’s doing to create that image. Because when you shoot LOG or RAW, the process is different.

The Bits & Basics

As you point your camera at a subject, light enters through the lens and hits the camera sensor, a collection of millions of tiny photosites. These collect the light that hits them and convert it into voltage – the more bright, the more voltage. But before we have video, we have to turn this into a digital format. The problem is, this voltage is analogue, so it’s a continuous number and to record it in digital, we need simpler whole numbers. To do this your camera breaks down the analogue signal into digital steps. Think about it as turning a slide into a staircase – less fun. More useful.

 

If we have a value of 1.2, and only 1 and 2 are available, the camera will round down and give it a value of 1. The number of these available values is called bit depth. 8 bits, which is the standard for displays like computer monitors, has 256 possible values. However most modern cameras, like the R5c, do it in 12 bits which has 4096 values available – much more than that display.

For RAW footage, the full 10, 12 or 16 bits of RAW sensor information is saved to the card and all the processing of turning that into watchable video, happens when you come to edit your footage. However, if you shot in LOG, this is where it differs.

 

So we tested it out. In the Syrp Lab studio we took one shot on some background paper with some nice smooth gradients and tried to push it to the point of breaking. When we shoot it in RAW, we can move it up and down, grade the colours and cant really push it to the point of breaking. When we do the same shot in 10 bit LOG we still have a bunch of latitude but eventually, if we do some pretty extreme changes, we get some banding. And shooting the same in 8 bit log, when we start to grade, it’s definitely noticeable in gradients.

If you’re a bit of a gear head, we shot this on the new Canon EOSR5C (thanks Canon!) on a Manfrotto 635 FAST tripod with some Colorama ‘Forget Me Not’ background paper.

 

Banding Basics

So why does this banding happen?

See, to save space, time and make your footage playable, the camera does a bunch of extra processing before it saves the LOG footage. Most importantly for us, the 12 bit Linear image is mapped onto a 8 or 10 bit logarithmic curve. Don’t worry, this isn’t about to become a math class.

This logarithmic processing is a further step of simplification where the footage is mapped to a non linear curve of values. Doing this lets the camera store less data, but the that’s data most important to human vision. But why?

Human vision isn’t simple and how we perceive the world is actually non-linear – like that curve. This thing called Weber’s law states that the human eye can detect changes in brightness, only when they are about 2-3% different than a previous or surrounding value. If we take 10 as our starting value, we could detect a difference from 10 to 10.3, but at 100 it would require a change to 102 or 103 for this to be perceivable.

Receive our latest videos and participate in camera gear giveaways

We send a newsletter once or twice a month and we make it easy to unsubscribe at any time.

This means that in the darkest sections of a video, we can detect small changes in brightness, but in the brightest sections, we can only detect larger changes. By mapping out our values on a logarithmic curve we can keep all the information in the areas it is most perceivable, but ignore and get rid of the information we cant see. Because of this, LOG footage is far more efficient with it’s bit depth than RAW so it’s file sizes are much smaller. RAW usually comes in the 12,14 or 16 bit flavours, whereas LOG only uses 8 or 10 bits.

 

So why shoot RAW?

So why bother shooting these larger, less compressed files?
Grading.

Once you get into the grade and you realise, “hey this shot could do with being a little different”, you start to stretch your values along that curve and things… break. Think of it as that staircase again. When you need to change the shape of the staircase, some of the stairs become too big to climb up. Where, in RAW footage you have a whole lot of values to bring up and fill in the gaps to give smooth gradients, in our LOG footage, we discarded a lot of that data to get smaller file sizes.

There’s also something else at play when we shoot in a non-RAW format. To get the best image straight out of the camera and keep file sizes as low as possible, most cameras do a bunch of compression, sharpening, noise reduction, as well as storing the colour information in a way that discards information you can’t perceive, as long as you don’t grade or manipulate the footage.

 

So how do we fix it?

Thankfully there are a few things we can do to make sure we never run into this problem again.

  • Shoot in a higher bit depth

    If your camera can shoot 10 bit LOG, you should be shooting in 10 bit LOG, not 8. While there may only be 2 extra bits of data, each bit doubles the amount of points you have. You actually have 768 additional steps to, give you room to grade the footage and avoid banding.

  • Check your LOG footage is being converted to SRGB using the correct profile.

    Different manufactures have different curves, sometimes several like CLOG, CLOG2 and CLOG3 all for Canon alone. By using the wrong conversion, either as a LUT or inside the colour space transform, you’ll be mapping the values incorrectly and your footage will look whack.

  • Shoot truer to final

    If none of this is possible – you’ll need to film truer to the final image, exposing correctly and getting your contrast ratio right in the scene so you need to grade less or ideally, not at all.

  • But ideally, if you can, consider shooting RAW.
    But maybe order some more LaCie’s

 

Thanks for watching – we hope you enjoyed this video and learnt something useful! Leave us a comment on YouTube with any questions or ideas and don’t forget to subscribe to Syrp Lab. Happy shooting.

Receive our latest videos and participate in camera gear giveaways

We send a newsletter once or twice a month and we make it easy to unsubscribe at any time.

May 30, 2022

Vitec Group
  Our Brands