The usage of wide color volumes within the camera during image acquisition for HDR-imagery is common nowadays. However, these color volumes are often manufacturer-specific and get minimized and transformed into a higher-order, standardized color system, such as ITU-R BT.709. Not every process-chain for video-imagery is currently capable of handling such wide color volumes. The outstanding color-capture-characteristics of prevailing image sensors therefore do not get used enough yet. Most image reproduction devices won’t be able to display High Dynamic Range (HDR) – and Wide Color Gamut (WCG) – content accurately in the near future. This justifies the current development of products that convert HDR-imagery to SDR-imagery with the aim of keeping the original look and the picture intent of the grading – or video-engineer. When we bring up HDR- to SDR-transformation, a focus should be set on an accurate color volume transformation. This aspect of a conversion process is of particular importance because certain colors, eg. patented colors, should cause the same visual impression on all reproduction devices. Visual artefacts, such as hueshifts, banding or unwanted constant color regions, caused by clipped or incorrect converted colors could violate this rule. In order to transform HDR- to SDR-content, a brightness compensation (tonemapping) will usually be performed as the difference in brightness between HDR and SDR is most visible. Rarely an accurate color volume transformation is performed. A novel HDR-SDR-transformation-algorithm will be presented in this paper as well as the underlying basics of constant-luminance color volumes and general transformation procedures. The newly designed algorithm is adaptable for different use cases. It can be used on an image after a tonemapping operation was conducted, or it can be used as a static conversion method between HDR and SDR. The aim of the algorithm is to convert colors from a source color volume into a target color volume in a way that the converted colors stay distinguishable and no clipping is necessary. The algorithm combines advantages of legacy conversion methods, such as hue-adaptation, softclipping and lightness- and core-region-preservation. It is based on a constant-luminance color-representation, enabling the display of colors with a luminance of up to 10.000 cd/m² (PQ-EOTF, ITU-R BT.2100) and enabling the display of all colors inside the spectral locus.
Technical Depth of Presentation
The paper/presentation will cover fundamentals of color-representations. These are fairly easy to understand and have a basic to fundamental technical depth. The paper/presentation will also cover the mechanics of the color-volume transformation algorithm. These could be categorized with an intermediate technical depth. The overall technical depth of the paper/presentation should therefore be categorized with an intermediate technical depth.
What Attendees will Benefit Most from this Presentation
The ideal audience would be a mix of engineers, technologists and researchers. The presented algorithm and the scientific field are more targeted towards researchers, developers and production engineers. Engineers could benefit from this presentation due to a better understanding of the underlying concepts of color-representations.
Take-Aways from this Presentation
• The importance of the usage of perception-accurate and constant-luminance color-representations for HDR-imagery will be presented. The consequences of a HDR-SDR conversion will be shown, if the conversion takes place in a non-constant luminance color representation with a less than optimal decorrelation between the color-difference and luminance channels.
• A comparison of simple HDR-SDR conversion methods with sophisticated approaches (proposed algorithm) will be shown. If the color- and luminance differences between HDR and SDR are small, simple conversion methods, such as RGB-clipping or CIE1931-xy-clipping provide sufficient accuracy. For HDR-imagery that highly exceeds the color- and luminance range of SDR, these conversion methods introduce unwanted artefacts.
• The importance of the usage of perception accurate color-volume mapping systems will be presented and how such a mapping system can improve simple conversion techniques. The advantages of perception accurate conversion techniques over clipping will be shown, as well as the technical background of it. A visual quality comparison will be drawn, that also shows how color-volume mapping can further improve HDR-SDR-tonemapping operations on an image.