Details:
Summary:
Digital photography has become a widely accepted method for capturing images for printed matter. There remains however considerable subjectivity regarding the demosaicing and color processing of raw images. This subjectivity affects ICC profiling, in that profiling an image can have variable results based on the software used to demosaic the raw data, and the color rendering settings that are manipulated prior to conversion. Camera profiling allows the shooting of a color target in a controlled environment that is limited in environmental variances and offers controlled exposure. The resulting target image is then profiled to achieve color information that is representative of the target under this controlled environment. In an uncontrolled environment, both the images shot and the observer visually assessing the scene are subject to environmental variables, such as variations in lighting, shadows, movement, and observer response. If these images are profiled using an accurate camera profile, the resulting image should be a color managed image from the perspective of the observer. The color of the image will be affected by the subjectivity of the scene, but the profile allows us to manage the variability in a controlled and predictable way. The research for this project explored colorimetric analysis of how camera raw color rendering settings of different camera raw editors affect the ability camera profiles to achieve accurate color prediction. To explore this, a GretagMacbeth Digital ColorChecker SG chart was shot in raw format using both a Canon and a Nikon digital SLR camera in a controlled lighting environment. The raw files were then converted into RGB using various approaches. The findings showed that color managed workflows with tailored profiles have the ability to overcome some inconsistency between images, producing images that more accurately predict color representation, even when the degree of accuracy of the initial image varies. In general, camera raw images color rendered in a linear state with minimal highlight and neutral adjustment produced profiles that were consistently empirically close to the target values (low delta E and variance). Generic color settings in Vendor and Mainstream software solutions did not adequately remedy exposure imbalances, as they are bounded by algorithms which do not base corrections on reference data relating to actual color information. Conversely, custom made profiles apply data-fitting algorithms or transformational matrices based on the variations between the captured data and actual color information, allowing them to blindly compensate for variations in individual images. When the linear raw data was compensated for lightness and gray balance, the results of the proofing were more consistent and repeatable, suggesting that these simple adjustments allowed the profiling software to better compensate for variations between images by providing a base constant to reference. If the goal of photography is to create a representation of the scene that appeals to our visual assessment (pleasing color), then the goal of profiling a camera should not be to make the camera match the scene. Instead, camera profiling should attempt to provide more stability and predictability in the way the scene is altered in order to achieve that desirable pleasing color image.