Architecture and method to transfer data in generation, display or printing
high edge placement accuracy images from multiple exposure of plurality of
predefined patterns with lower edge placement accuracy. A pattern is laid
out on a grid finer or different from grid size of image transducer pixel
size, overlaid by transducer grid and converted to n patterns compatible
with transducer grid. When combined by partial exposures weighting
patterns unevenly, the n patterns generate an image with line edge
positions a fraction (1/(2.sup.n -1)) of transducer grid size. For most
picture display and step-and-repeat lithography applications, pattern
stored in first memory is displayed or partially exposed once, and
remaining patterns are displayed or partially exposed 2.sup.m-1 times m
being copy number of the pattern. Superimposing 2.sup.n -1 exposures in
human eye to scene integration time, picture with improved line placement
accuracy is perceived.