Color space where brightness values don't map linearly to physical light intensity — sRGB, Rec.709, Log curves. Critical for grading and delivery.
You're in a color grade and quickly realize: working in a linear space is impractical. That's why nonlinear color spaces exist — they encode brightness information not by physical intensity, but by a curve that mimics the human eye. This isn't a mistake, it's intentional. Human perception is exponential; we see differences in the shadows more clearly than in the highlights. A nonlinear color space like sRGB or Rec.709 uses this property to store and transfer image information more efficiently — fewer bits for the same visual quality.
In practice, this means: when you work in DaVinci or Nuke, you're usually already working in a nonlinear space. The footage coming from the camera is log-encoded or gamma-corrected to Rec.709. Saving storage space, optimizing transfer, and handling grading curves more intuitively — these are the real reasons this standard exists. However, when you're doing compositing, you often need to convert to linear space because mattes, lights, and physical effects function correctly mathematically there. The conversion between linear and nonlinear happens via the gamma curve or specialized LUTs.
Log color spaces — such as Alexa LogC, Red LogFilm, or DaVinci Intermediate — are extreme forms of nonlinear: they maximize usable dynamic range by compressing highlights more strongly. This gives you more grading latitude, but requires an understanding of how they work. Many beginners confuse themselves because they don't know if their current working space is linear or nonlinear. This leads to incorrect color and exposure corrections.
The practical tip: Always explicitly set your color space — define in the project setup whether you are working in linear or nonlinear space. Your color management depends on it. And remember: sRGB and Rec.709 are your standard output color spaces for web and SDR broadcast. They are nonlinear, and that's exactly right for them.