Skip to content

Color spaces & camera log, properly

IntermediateDuration ~12 min video + 30 min hands-onTools DaVinci Resolve (free)

In Level 1 you learned to recognize log footage and normalize it with a Color Space Transform or RCM — as a recipe. You typed “S-Log3 / S-Gamut3.Cine” into a box because a tutorial told you to. This lesson makes that recipe understood. Every camera’s footage lives in a color space, and a color space is really two separate ideas welded together: which colors it can hold, and how it encodes brightness. Once you can pull those two apart, “what do I put in the input field?” stops being a guess. You’ll be able to look at any clip, name its log-and-gamut pair, and know exactly why the normalization step does what it does — the theory Level 1 deliberately skipped.

Watch for: Qazi opens by separating the two ideas on a chart (around 3–4 min): the big triangles are color spaces / gamuts of different sizes, and Rec.709 is one of the smaller ones. The key practical demo is the 'choked display' section (around 8–11 min): he sets Resolve to output a wide HDR gamut, and his ordinary monitor can’t show it, so the image falls apart — proving your display, not your project, is often the bottleneck. Watch him then match a Rec.709 export against a P3 screen. Ignore the monitor-shopping section at the end unless you’re buying.

Start with the word everyone blurs. A color space is a full specification of how color is stored, and it contains two independent parts.

The first is the gamut — the range of colors the space can represent. Draw the visible spectrum as a horseshoe and a gamut is a triangle inside it: a bigger triangle holds more saturated, more extreme colors. As Qazi shows on his chart, Rec.709 (what TVs, phones and YouTube expect) is a fairly small triangle; DCI-P3 is bigger; camera “wide gamut” spaces and the working space DaVinci Wide Gamut are bigger still, some reaching past what the eye can see. Gamut answers “how saturated a red can this hold?”

The second is the transfer function — how the space encodes brightness, the mapping between the numbers in the file and the light on screen. This is the “gamma” half. A standard display uses a fixed power curve: Rec.709 is mastered to roughly gamma 2.4 in a dim room; sRGB, the web’s standard, is close to gamma 2.2. A log transfer function is a different animal entirely — a deliberately flat, shadow-lifted curve that crams an enormous brightness range into the file so nothing in the highlights or shadows is thrown away before you grade. That’s why log looks milky and broken on a normal screen: it isn’t wrong, it’s encoded for latitude, not for viewing.

Here’s the payoff Qazi drives home with his “choked display” demo. Inside Resolve your project can work in a gamut as wide as you like, but you are always limited by what your monitor can physically reproduce. When he pushes the output to a wide HDR gamut his ordinary display can’t speak that language and the picture collapses. Set it back to Rec.709 and it’s correct again. So a wide working space is good — it gives your grade room — but a color-accurate display is what lets you trust what you see, which is exactly why Level 3 spends real money on monitors and this level does not.

Now the practical skill: every camera records in a log-plus-gamut pair, and your whole job at normalize time is to name it. The CST’s “input” is two fields — a gamma and a gamut — and they must match what the camera shot:

  • Sony: S-Log3 paired with S-Gamut3.Cine.
  • ARRI: LogC — LogC3 on older Alexas, LogC4 on the Alexa 35 — paired with ARRI Wide Gamut.
  • Panasonic: V-Log paired with V-Gamut.
  • RED: the IPP2 pipeline, REDWideGamutRGB with the Log3G10 curve.
  • Blackmagic: BRAW, decoded via Blackmagic Design’s Film Gen 5 gamut and gamma.

Get the pair right and normalization is invisible and correct — the flat log resolves into a healthy, gradeable image, just as Cullen Kelly’s color-managed clips snap into place the instant he tags each shot’s real input (you’ll see that in 2.2). Get it wrong — tell Resolve an ARRI clip is Sony — and the picture shifts, because you’ve asked the math to undo a curve the camera never applied. RAW formats (BRAW, ARRIRAW, R3D) are the friendly exception: they carry their space in metadata, so a color-managed project maps them automatically and you often can’t even change the input, as Kelly demonstrates on his Pocket Camera shots.

  1. Gather three clips shot on different cameras — or grab Blackmagic’s free training media, which mixes ARRI and RED. If you only have one camera, shoot one log clip and one Rec.709 (standard) clip.
  2. For each clip, find its log-and-gamut pair. Check the camera’s manual or the clip’s metadata (right-click → Clip Attributes, or read the file name). Write the pair down before touching Resolve.
  3. Drop a Color Space Transform as your first node. Set the input to the pair you identified and the output to Rec.709 / Gamma 2.4. Confirm the flat log resolves to a healthy image on the waveform.
  4. Now deliberately break it: set the input to the wrong camera’s space and watch the picture shift on the parade. Set it back. That shift is what a wrong guess costs.
  5. Change the CST output from Gamma 2.4 to Gamma 2.2 and note the small lift in the shadows. That’s the transfer-function half acting alone, with the gamut untouched.
Log formats table — camera log/gamut pairs for the common cameraslog-formats-table.pdf154 KBOriginal course material — free to useLevel 2 workbook — every Do it exercise, 2.1–2.12, plus the capstone (printable)level-2-workbook.pdf799 KBOriginal course material — free to use

Check yourself

  1. A color space bundles two independent things together. What are they?

  2. What does a camera’s log curve actually do?

  3. You are handed a clip shot on a Sony camera in log. To normalize it with a Color Space Transform, what pair do you need to identify and enter as the input?

  4. Your Resolve project is working in a wide gamut, but your laptop panel can only show Rec.709. What happens to the colors outside Rec.709?

You can move on when you can… take any clip, name its gamut and its transfer function separately, identify the camera’s log-and-gamut pair, and explain why entering it correctly in a CST makes normalization invisible — and why your display can still be the limiting factor.

Gerald Undone — How to Correct Log & HLG Footage & Make LUTs: a hands-on pass at correcting log by hand and baking the result into a LUT. It’s an older, livestream-style video with a dated by-eye approach, so treat it as background — useful for seeing why the color-managed method in 2.2 is cleaner than hand-correcting every clip.

Watch for: [Optional/background.] How he manually lifts and shapes a log curve, and the point where making a LUT out of a correction is worth it — versus letting color management handle it.

Cullen Kelly — 36 Project Settings (Camera RAW segment): a short, precise look at the Camera RAW decode controls, where a RAW clip’s gamut and gamma are set per format (BRAW, ARRIRAW, RED, etc.) rather than in a CST — the metadata exception in action.

Segment: 35:03–39:52 — Camera RAW decode controls per camera profilewatch full video

Watch for: How the RAW tab exposes gamut/gamma per camera profile, and why RAW is auto-mapped where a baked codec is not.

Next up: 2.2 · Color management for real — turning this understanding into three real pipelines you can set up and choose between.