Prof. Dr. Karol Myszkowski (Max Planck Institut
Saarbrücken): Perceptually-based global illumination, rendering, and animation
techniques.
Many improvements in image synthesis and computer animation have
resulted from exploitation upon human perceptual capacities and
insensitivities. This is because it is the appearance of the resulting
images that is of primary relevance to the successful development and
deployment of new algorithms for image synthesis and computer animation.
In this study, we investigate the applications of the perceptually-based
Visual Difference Predictor (VDP) developed by Daly to guide the realistic
rendering computation.
First, we discuss the results of our psychophysical experiments
involving human observers, which were designed specifically to validate
the performance of this general purpose predictor in the tasks typical for
the global illumination computation. Then we show an example of the VDP
application to evaluate progressive changes in image quality at various
stages of the global illumination computation. The VDP responses are used
to support off-line decisions regarding the selection of the best
technique from a pool of complementary algorithms, which at a given stage
of computations minimize the perceivable differences between the
intermediate and final images. Using this approach we are able to provide
the high quality images of complex environments within single minutes or
seconds using physically-based partial solutions.
Next, we discuss the extensions of VDP required to measure the quality
of animated sequences. We use the resulting Animation Quality Metric (AQM)
to develop an efficient antialiasing technique handling both still and
animated images. Our antialiasing solution is based on a
motion-compensated filtering, and the filter parameters have been tuned
using the AQM predictions of animation quality as perceived by the human
observer. These parameters adapt locally to the visual pattern velocity.
|