Blog / Gear

Canon C70 CLog3 to Rec.709: Why the Built-In LUT Crushes Your Shadows

8 Min Read

The C70's built-in CLog3-to-Rec.709 conversion looks fine on a monitoring scope until you push it into post, then the shadows go muddy, the grade fights you, and you spend an hour wondering what you did wrong in the field. You didn't do anything wrong. The LUT did.

Canon's Cinema EOS line has earned its reputation in the Indian production circuit, and the C70 sits in a genuinely useful position: compact enough to fit on a gimbal beside a Sony FX3, capable enough to cover a full Punjabi baraat through a 400mm on a PL adapter. But there's a persistent frustration among shooters new to CLog3 who apply Canon's own supplied LUT in Resolve and find that their shadows look crushed, the dark fabric of a sherwani collapses into a single grey-brown mass, and recovering detail in a dimly lit mandap becomes a fight they can't win.

This post explains exactly why the Canon-supplied CLog3 → Rec.709 LUT behaves the way it does, what it does to your histogram, and how to build a proper transform chain in Resolve that keeps the shadow detail you actually captured on sensor.

What CLog3 Is Actually Storing

Canon Log 3 is a logarithmic gamma curve designed to record the full dynamic range of the Cinema EOS sensor, roughly 14 stops on a C70 body, into a compressed tonal range that fits cleanly onto 10-bit media. The critical thing to understand is where it places middle grey and where it parks the low end.

In CLog3, middle grey (18%) sits at approximately 40 IRE on a 0–100 waveform scale. The curve is relatively steep in the shadows, meaning there is real, recoverable detail packed between 5 IRE and 25 IRE on a correctly exposed shot. When you use a scope in Resolve and look at a CLog3 clip of a dimly lit phera ceremony with candles and 2700K tungsten table lamps, you'll see shadow detail riding between 8 and 20 IRE. That information is there. It was captured.

The problem starts when you apply the transform to bring it into Rec.709 for delivery.

What the Built-In LUT Does to Shadows

Canon provides a .cube LUT titled something along the lines of CanonLog3_Cinema_EOS_to_Rec709_v1.1. It ships with Canon's official download, and it also appears as an in-camera LUT applied for monitoring. When you drop this onto a node in Resolve and read the waveform, you will typically see the following:

The result is that detail which sat at 12 IRE in your CLog3 file, the shadow side of a bride's silk saree, the darker panels of a mandap drape, skin in a backlit vidaai moment, gets pushed to 7–9 IRE in the output and then clipped at the display floor. You can lift it in a downstream node, but you're lifting a compressed, nearly clipped signal, not genuine shadow detail. The noise amplifies before the texture does.

This is not a bug exactly. The Canon LUT was designed to give a "broadcast safe" output directly from a log timeline, the kind of output a news or corporate client can drop into an existing SDR workflow without any further grading. It is conservative about shadows on purpose. For a filmmaker doing any real grading in post, it is the wrong starting point.

The Right Transform: CST Over a Clip-Based LUT

The correct approach for a graded workflow is to bypass the Canon LUT entirely and use a Color Space Transform node (Resolve's built-in OFX) to handle the input transform, then grade on top of a linearised, properly placed signal.

Here is the node structure to use as your base grade for C70 CLog3 material:

Node 1, CST In
  Input Color Space: Canon Log 3
  Input Gamma: Canon Log 3
  Output Color Space: DaVinci Wide Gamut
  Output Gamma: DaVinci Intermediate
  (Tone mapping: OFF)

Node 2, Creative grade (exposure, contrast, primaries)

Node 3, CST Out
  Input Color Space: DaVinci Wide Gamut
  Input Gamma: DaVinci Intermediate
  Output Color Space: Rec.709
  Output Gamma: Gamma 2.4 (or 2.2 for web delivery)

With this chain, your CST In does the colour science work of converting Canon Log 3 / Canon Cinema Gamut into Resolve's own wide gamut working space, which is linear in character and handles the shadow end without the artificial clip the Canon LUT introduces. When you check your waveform after Node 1, you'll see middle grey sitting around 40–42 IRE and your shadow detail intact, your 12 IRE shadow information from the ceremony hall is still at around 11–13 IRE in the working space, still real pixels, still gradeable.

The tone mapping toggle in the CST deserves a word. When you have a large gamut like Cinema EOS to DWG, leaving tone mapping on can introduce its own midtone compression. Turn it off and handle your own contrast in Node 2. You are the colorist. Let the CST do only the colour science.

Dialling In Shadow Latitude on Node 2

With your CST In in place, Node 2 is where you do real work. For a typical indoor wedding situation, the reception at a hotel in Bengaluru with LED uplighting at 3200K, mixed with fluorescent fills, here is a starting point:

After these adjustments, the dark sherwani fabric reads as a distinct deep blue-grey rather than a collapsed black mass. The difference between 8 IRE crushed and 12 IRE with shadow detail is visible on any calibrated monitor and absolutely visible in a WebGL player on a colour-accurate phone screen.

CLog3 and Skin in Tungsten-Heavy Mandap Lighting

Indian wedding venues lean heavily into warm sources, from 2700K LED battens and marigold-bounced tungsten to the open flame of the havan. This creates a problem specific to CLog3 that is separate from the shadow crush issue: skin in the warm part of the frame and skin in the cooler shadows of the same shot can read very differently after transform, because the Canon curve handles the high-chroma orange-yellow channel (the dominant channel in tungsten-lit Indian skin) with more compression than it applies to neutral tones.

When the Canon LUT processes this, it simultaneously crushes your shadows and slightly over-saturates the warm skin, giving you a double problem. The CST approach handles this more accurately because DaVinci Wide Gamut has a larger container for high-chroma yellows and oranges. After your CST In, the skin in the warm zone sits at a more accurate hue angle and is easier to handle with a HSL Qualifier if you need to separate it further.

As a check: after your CST In, pull a vectorscope. Warm tungsten-lit Indian skin should land just above the skin tone line (the line pointing roughly from centre toward the yellow-red boundary). If it's pulling hard toward yellow, your white balance from the CLog3 encode is drifting. A small negative adjustment on the Master Gain in Blue in Node 2's primaries, something like -0.02 to -0.04, usually centres it without affecting the shadow grade you've already built.

From the Positiva LUT Library

Indian Wedding LUTs

Built for the exact tungsten-LED-fire mix of Indian ceremony halls: pre-constructed grade nodes that protect shadow detail and keep warm skin tones on the skin line.

View Pack →

When You're Working With In-Camera Log Assist

Some C70 shooters enable the Log Assist (view LUT) on the monitor output but shoot in CLog3 to card. This is correct practice for on-set monitoring. The issue arises when a client's EVF operator or the second shooter's AC is used to seeing the Log Assist preview and reads a correctly exposed CLog3 frame as "dark." They'll ask you to lift exposure on set. Resist this. A correctly exposed CLog3 frame will look flat and low-contrast on a monitor without a LUT, that is the point.

If you are delivering directly for broadcast with no grading pass, then yes, the Canon-supplied LUT applied in Resolve at export is fast and predictable. But if any colorist is touching this footage, the CST workflow above is the right handoff. Export a DPX or ProRes 4444 from your graded timeline, not a baked-LUT H.264.

Matching C70 CLog3 to FX3 or A7S III in a Multi-Camera Cut

Multi-camera is common at Indian weddings: a C70 on the main aisle and a Sony FX3 or A7S III on a second angle. The FX3/A7S III will typically be shooting S-Log3 / S-Gamut3.Cine. After both are through their respective CST In nodes into DaVinci Wide Gamut / Intermediate, they share the same working space and your Node 2 grade applies equally to both. This is the strongest argument for the CST approach: it creates a common container so that multi-camera matching is a contrast and colour balance problem, not a transform problem.

For a deeper look at how shadow grading interacts with location colour science, the post on grading Ladakh's cold blue shadows without cyan-shifting skin works through a very different lighting scenario but uses the same CST logic. Seeing the same node structure applied to two opposite colour problems clarifies what the working space is actually doing.

One check before you export anything: set your viewer to Rec.709-A or Rec.709 in Resolve's colour management settings. If you're working in an unmanaged timeline and your CST Out is set to Gamma 2.4, make sure your monitor is calibrated to 2.4 or you'll be grading for a display you're not seeing. This is where crushed shadows get missed in QC and go out to clients looking wrong on a calibrated TV.

The C70 is a capable camera and CLog3 stores real, gradeable latitude in those shadow stops. The Canon LUT is a monitoring convenience, not a grade. Treat it that way and the camera repays you.

✦ GRADE EVERY CAMERA

GRADE EVERY CAMERAOne pack, every log format

The Positiva Bundle covers CLog3, S-Log3, D-Log M and BRaw in one download, pre-built CST nodes, skin tone protection and shadow shaping for mixed-camera Indian wedding and travel shoots.

Get the Bundle →