Raw color values without gamma correction — what the sensor actually captures. Essential for accurate compositing math and grading in DaVinci or Nuke.
You're working in Nuke or After Effects and suddenly realize your color values aren't behaving as the math predicts — multiplications turn to mush, additive compositing looks wrong. The problem: You're in non-linear color space, while the compositing engine expects you to calculate linearly. Linear color space is the opposite of what your monitor shows you. It's the actual light values as captured by the sensor — without the gamma curve your eye needs to look correct on a display.
In practice, this means: If a sensor captures double the light value, it stores it as exactly double the numerical value — linear. Your monitor, however, never displays these raw values directly; it applies gamma (usually 2.4) to adjust the image for the human eye. This isn't aesthetic — it's physics. All modern compositing packages calculate linearly internally because only there do mathematical operations function correctly. A blur in linear space distributes light values according to physical principles. A blur in gamma-corrected space becomes dark and muddy because the math is based on false assumptions.
On set, this specifically means: Your log footage (Alexa LogC, Sony S-Log, RED Dragon, etc.) is not linear — log is a non-linear encoding that compresses dynamic range. You need to convert it to linear format before you composite or grind in Nuke. This is done through LUT application or in the color management system of your editing system. DaVinci Resolve works internally with linearized values if you're configured correctly — for this, you need the correct Input Colour Space definition (LogC → Linear Rec.709, etc.). Many VFX shots go wrong because someone did color grading in gamma space and then the compositor continued working linearly — the grade becomes unexpectedly harsh or soft.
The practical rule: Always work in linear space in compositing and color-critical grades. Save your intermediate results (mattes, gradeouts) in 16-bit linear, not in 8-bit gamma. Only your final delivery goes back into gamma for the monitor or cinema. And yes — this makes files larger and work slower. But it's the only way for math to work.