3 minute read
A color space is a way of organizing colors. In the context of images and monitors, we describe color spaces in terms of their RGB primaries and white point. It is important to know what color space an image is in so that colors can be correctly interpreted and displayed. As part of a workflow, colors may need to be converted from one color space to another, and a knowledge of color spaces can help us avoid loss of information or inaccurate display of colors.
A color space definition consists of the CIE 1931 chromaticity coordinates (x,y) of the three primaries (red, green and blue) and the white point. Plotting a triangle on a CIE 1931 xy diagram provides an outline that helps visualize the visible colors a color space encompasses.
The gamut of a color space is the set of all colors that can accurately be represented in that space. When an image in one color space is converted to another color space, the relative sizes of the gamuts become important as some colors that can be represented in a larger gamut cannot be accurately represented in a smaller gamut. Typically any color space that has a wider gamut than REC709 is referred to as “wide gamut”.
Throughout a project, color spaces will need to be chosen. The camera will be recording into a color space (even RAW is a color space in the sense that it encompasses a set of colors the camera can see), images will be in a color space for viewing while being edited, grading will require images to be in a color space and probably transform them all into a deliverable color space.
The process of converting from one color space to another is part of color management and therefore it’s important that, along the path the image takes through the workflow, it is not unduly restricted in color space. Once a color space is limited to a smaller gamut, it cannot be expanded back up to a larger gamut – that information is lost. This is analogous to how dynamic range is handled through capture, editing, grading, output whereby the best workflows capture as high a possible dynamic range in the camera (recording RAW to ensure nothing is lost, or using a log recording where, if careful, little dynamic range is lost), only compressing that dynamic range to that which is pleasing on a given display as part of the grading process.
Contributed by the team at Adobe.
Camera sensor data doesn’t have a well-defined color space. Camera sensors are not colorimetric devices, and although the mathematics of color spaces are often used to describe camera data, it is transformed into a color space before use. That color space could be a display space such as Rec. 709, Rec. 2020, DCI-P3, or a more camera-specific color space designed to encompass all the colors a camera captures into a wide gamut space suitable for grading. Examples include REDWideGamutRGB, Arri Wide Gamut, and Sony S-Gamut. These wide gamut spaces will need proper conversion to the display or deliverable color space to ensure accurate portrayal of captured colors. The conversion should be done with care to ensure any colors which are in the camera space gamut, but outside the display gamut are handled in such a way as to avoid clipping which can produce ugly results.
Sometimes a common intermediate color space will be used for all elements of a project. The Academy’s ACES project provides a color space for this purpose AP0, but in other projects the most commonly used source’s wide gamut colors space will be used as the common space and the other elements all converted to it.
The display or deliverable color space will be defined by the nature of the project. For cinema release it’s typically DCI-P3, for TV and web it’s Rec. 709, and for HDR UHD it’s Rec. 2020. As a project will usually encompass a range of source material, elements that are not in the target deliverable space will need to be converted. Special care needs to be taken with graphic elements and logos to ensure they display correctly on the final export.
Video collaboration solved.