Visible Difference Predictors 10 Image-quality evaluation is important in many applications, from image acquisi- tion, synthesis, and compression to restora- tion, enhancement, and reproduction. Quality metrics are commonly used in algorithm validation and comparison tasks. For example, to speed up rendering com- putation, simplifying assumptions are often used in designing novel algorithms, and quality metrics can be used to determine whether this leads to visible image distor- tions or artifacts. A typical application of quality metrics is the evaluation of various algorithms' efficiency in terms of texture, image, and video compression. Another group of quality metric tasks is the optimization of algorithms in terms of visual quality. Here, quality metrics are often embedded into control loops, and the goal is to minimize perceivable error. In this case, quality metrics can be used to deter- mine optimal values of certain parameters or to decide upon stopping condition