Abstract
Digital reproductions often aim to recreate the visual experience of viewing a physical scene. Current reproduction pipelines are limited by the luminance dynamic range of typical cameras and displays, which is much narrower than the dynamic range of many physical scenes. Thus, reproducing visual contrast and detail in a way that replicates the experience of a scene is a challenge, which requires mapping high dynamic range content to lower dynamic range via tone mapping algorithms. However, because tone mapping is a lossy process, it can compromise visibility and perceptual realism. It has recently been suggested that binocular vision can be leveraged to improve perceived luminance contrast by showing differently tone mapped images to the two eyes (dichoptic tone mapping). Several such tone mapping methods have been proposed but the results for improving perceptual quality are mixed. We hypothesized that these mixed results were related to the different baseline tone mapping algorithms used in perceptual comparisons. To address this issue, we conducted perceptual studies using image quality ratings and a two-alternative forced choice task to ask whether dichoptically tone mapped images were systematically preferred over a range of alternative tone mapping methods that presented identical images to both eyes. We also examined the perceived three-dimensionality of dichoptically tone mapped images for potential applications in stereoscopic 3D content. Our data suggest that preferences for dichoptically tone mapped images may be driven by a preference for one image out of the dichoptic pair, rather than a preference for the dichoptic pair together. Consistent with prior work, we found that dichoptically tone mapped images were perceived as appearing more three-dimensional. In conclusion, when the goal of digital reproduction is to convey realistic contrast, dichoptic tone mapping may not consistently improve over conventional tone mapping, but it may be used to enhance the 3D impression.