We introduce the notion of image-driven simplification, a framework that us
es images to decide which portions of a model to simplify. This is a depart
ure from approaches that make polygonal simplification decisions based on g
eometry. As with many methods, we use the edge collapse operator to make in
cremental changes to a model. Unique to our approach, however, is the use o
f comparisons between images of the original model against those of a simpl
ified model to determine the cost of an edge collapse. We use common graphi
cs rendering hardware to accelerate the creation of the required images. As
expected, this method produces models that are close to the original model
according to image differences. Perhaps more surprising, however, is that
the method yields models that have high geometric fidelity as well. Our app
roach also solves the quandary of how to weight the geometric distance vers
us appearance properties such as normals, color, and texture. All of these
trade-offs are balanced by the image metric. Benefits of this approach incl
ude high fidelity silhouettes, extreme simplification of hidden portions of
a model, attention to shading interpolation effects, and simplification th
at is sensitive to the content of a texture. In order to better preserve th
e appearance of textured models, we introduce a novel technique for assigni
ng texture coordinates to the new vertices of the mesh. This method is base
d on a geometric heuristic that can be integrated with any edge collapse al
gorithm to produce high quality textured surfaces.