What A Color Space Actually Constrains

Three numbers describing a pixel mean nothing without a definition. A color space supplies that definition, and mismatched definitions are behind a large share of color problems.

The definition has three parts

A color space specifies the exact colors its primaries represent, which fixes the corners of the range it can express. Different standards place those corners differently.

It specifies a white point, the color that appears neutral when all three channels are equal. Shifting the white point tints the entire image.

It specifies a transfer function relating stored values to light output. That relationship is deliberately non-linear because vision is more sensitive to changes in dark tones.

The same numbers describe different colors

Maximum red in a narrow space and maximum red in a wide one are physically different colors. The stored values are identical.

Interpreting a wide-space file as narrow makes colors look muted and slightly off. The reverse makes them look oversaturated and cartoonish.

Neither is a processing error. The values were never altered; only the definition applied to them changed.

The delivery standard is chosen, not discovered

Web video and conventional broadcast largely share one long-established space, which is why untagged content usually looks acceptable. That default has held for a long time.

High-definition and high dynamic range delivery use wider definitions capable of expressing more saturated colors. They require correct signaling to be interpreted properly.

Cameras frequently record in a manufacturer-specific space that is wider still, intended as a working format rather than a delivery one.

Conversion is a mapping problem

Moving between spaces means recalculating every value so the described color stays the same. Where the source contains colors the destination cannot express, something must give.

Clipping snaps out-of-range colors to the nearest available one, preserving accuracy elsewhere while flattening saturated areas. Compression scales the whole range inward, preserving relationships while desaturating everything.

Neither is universally correct. The choice depends on whether the saturated content is the subject or the background.

Tags must survive the whole chain

Every stage from camera through editing to delivery must agree on which definition applies. One mislabeled step shifts the picture for everything downstream.

Because the shift is often subtle, it can pass unnoticed until the material is seen beside a correct reference. Skin tones are usually the first place it shows.

Conversion tools that discard tags cause this routinely. Verifying declared color properties after any format change catches it while it is still cheap to fix.