Before imposing liability for copyright infringement, a court analyzes whether the defendant’s allegedly infringing work is substantially similar to the copyright-holder plaintiff’s allegedly infringed work. This substantial similarity analysis broadly contains two steps. First, facts and ideas do not receive copyright protection and are filtered out. Second, the two works are compared to see if there is material overlap between the two works’ remaining creative expression—i.e., whether or not the two works are substantially similar. This two-step approach furthers the delicate dual goal of copyright law to keep ideas and facts freely available as raw material for creation while awarding an author copyright over particular creative expression. While this substantial similarity test is noncontroversial, drawing the elusive line between idea and expression has been challenging for the courts. This task is especially difficult in the context of pictorial, graphic, and sculptural (PGS) works. The substantial similarity test evolved largely in the context of literary works and is not perfectly transposable to PGS works. Courts have resorted to two pernicious tendencies in applying the substantial similarity test to PGS works. They have used either what this Note calls a “literal descriptive approach,” comparing identifiable objects across visual works, or a “myopic visual approach,” comparing abstract visual components such as colors and shapes across visual works. Both of these tendencies fail to approximate how human vision works. Instead of identifying objects first or seeing shapes and colors in isolation, visual experience is initially nonverbal and inherently contextual. A substantial similarity test divorced from visual perception may erroneously draw the line between copyright protection and no protection, deviating from the idea–expression divide. Drawing on the science of visual perception and the nature of artistic training, this Note proposes an alternate framework for substantial similarity analysis of PGS works. In this truly visual substantial similarity test, the first filtration step would sieve out what this Note calls “perceptual facts” often reflected in artistic techniques. Visual works would then be compared in terms of how perceptual facts were combined to form an aesthetic, visual composition. This approach correctly conceptualizes the visual idea–expression divide and therefore allows meaningful protection for copyright holders without removing raw material for creativity from subsequent creators.