Home · Lab · Color

C4 Log: why footage looks flat

Many stops packed into 0–1

Color · ~6 min · Free

1 The idea

Cameras record more stops than sRGB can hold. Log squeezes that wide range into 0–1, giving each stop roughly equal room. Viewed raw, it looks flat and grey.

Before comp, log must be converted back to linear (in Nuke, set the colorspace on the Read). Light maths like plus or defocus belongs in linear; log suits some jobs, such as grading HDR fire or keying.

LogC3: c·log10(a·x + b) + d (x > cut)

2 Hands-on

Slide right: sRGB clips from about +2.5 stops

SkyVFX Lab · C4interactive · 1001

3 Deep dive

Equal room for every stop

In log, ×2 becomes “add a constant”. In ARRI LogC3 (EI 800) one stop near mid grey costs about 0.069–0.074. Notice that black (linear 0) does not sit at 0 but around 0.093.

LinearStops from greyLogC3 (EI 800)sRGB
0.0225−30.2000.162
0.1800.3910.461
0.72+20.5360.865
1.0≈ +2.50.571clips at 1
≈ 55≈ +8.31.0clipped

Log is a container, not a look

Each camera brand has its own log curve, and some change with camera settings (LogC3 depends on EI). Pick the wrong curve and every light ratio is wrong. The symptom: a tiny grade makes a practical lamp jump far brighter than it should.

Order for finding the right curve: file metadata → camera report → camera brand → the closest curve, then test it.

  1. 1File metadata
  2. 2Camera report
  3. 3Camera brand/model
  4. 4Closest curve + test

When to work in log on purpose

Some jobs behave better in log: grading hot fire or sparks, keying dark hair against a bright sky, transforms with sharp filters that ring into negatives. Do the job, then go straight back to linear.

Merging in log stops soft foreground edges being swallowed by a very bright background, but it is a perceptual trick, not physics. Emissive light such as glows or sparks still gets added (plus) in linear.

  1. 1linear
  2. 2to log
  3. 3grade / key / filter
  4. 4back to linear (same curve)

Offset in log ≈ exposure in linear

Because log turns multiplies into adds, adding about 0.074 in LogC3 is close to +1 stop once back in linear (away from the toe). That is why colourists use offset in log to change exposure.

Gain in log is the dangerous one: a small amount becomes a big jump once back in linear. Use it lightly or not at all.

Know each curve’s ceiling

If log values are kept in 0–1, each curve has a ceiling: Cineon (Log2Lin defaults, 95/685) tops out near 13.5 linear, LogC3 EI 800 near 55, ACEScct near 222. Sparks at 62 clip in Cineon and LogC3 if anything clamps.

In float, log values above 1 survive as long as no node in between clamps them.

For hot fire pick a log with headroom such as ACEScct, and never clamp between the conversion pair

4 In real work

In Nuke

  1. Read: choose the camera’s curve, e.g. AlexaV3LogC in Nuke mode or the ARRI LogC3 input in an ACES config, so the plate arrives linear.
  2. Log2Lin: operation lin2log and log2lin, Cineon defaults (black 95, white 685).
  3. OCIOColorSpace: in = working space (e.g. ACEScg), out = a log space (e.g. ACEScct); a second one with in/out swapped brings it back.
  4. Read has a raw data checkbox to see the file’s log values unconverted, for checking only.
  5. Keep each conversion pair inside a backdrop labelled “LOG” so it is obvious which part is not linear.

Common mistakes → the fix

  • ✗ Reading a log plate as sRGB or linear✓ Set the camera’s correct curve, then check mid grey lands near 0.18 and practical lamps read sensibly.
  • ✗ Comping the whole script in log, including plus, glows and light defocus✓ Go to log only for the jobs that need it, then straight back to linear.
  • ✗ Heavy gain in log✓ Use offset in log, or multiply in linear.
  • ✗ Converting in with one curve and out with another (Cineon Log2Lin in, LogC out)✓ Both halves of the pair must use the same curve. Test with a difference: it should be black.
  • ✗ Using Cineon on fire that reaches 50✓ Cineon tops out near 13.5 if clamped; use ACEScct or LogC.

Before you send the shot

  • The plate’s log curve confirmed from metadata or the camera report
  • Every log section has a matching conversion back with the same curve
  • No clamps between the conversion pair
  • Light-adding merges (plus, glow) happen in linear
  • The element’s peak values stay under the chosen log curve’s ceiling

5 Check yourself

One question to check yourself

Log footage looks flat when viewed raw. What does that mean?

What you learned

🔒 Answer the question above and this card opens

More questions

1. In LogC3, what does adding about 0.074 to every pixel do once back in linear?

2. Why grade a fire element around 7 in log rather than linear?

Where the numbers come from: Curves and matrices are computed from published definitions (sRGB IEC 61966-2-1, ARRI LogC3 EI 800, Rec.709/P3/Rec.2020/ACES AP1 primaries) and checked against published reference numbers. The tone curve is an example formula, not ACES/AgX/Filmic.