To provide a high-speed parallel processing device requiring small memory capacity by converting a graphical object into a chunk image and by synthesizing the chunk image to generate a display image.
In order to reduce a rendering overhead, this device executes affine transformation on a g-sprite 552 for simulating motion of a three-dimensional object rather than re-rendering the object for frame animation. Rendering geometry in a shank enables complicated fragment antialiasing. The device stores fragmentary or semitransparent pigments for expressing a partially covered pigment position in a fragment buffer. After rasterizing a primitive for the shank, a fragment decomposition auxiliary device identifies the fragment for calculating output pigments. A rasterizing component of the device tries to unite fragments for saving a fragment memory.
KAJIYA JAMES T (US)
TORBORG JOHN G JR (US)
KENWORTHY MARK L (US)
TOELLE MICHAEL A (US)
GRIFFIN KENT E (US)
LENGYEL JEROME E (US)
GABRIEL STEVEN A (US)
VERES JAMES E (US)
CHAUVIN JOSEPH W (US)
GOOD HOWARD (US)
POWELL WILLIAM CHAMBERS III (US)
