GigaPatchExplorer enables users to filter billions of possible exploration regions or patches of gigapixel images (a) into a manageable subset (f) using patch set queries (d). A patch set is a subset of all possible patches from one or more images. In our implementation it consists of (1) a set of patches, and (2) a permutohedral lattice for fast indexing of patches in a given high-dimensional feature space. The permutohedral lattice implicitly groups similar patches into meaningful cells. For instance, cells with low patch count correspond to unique or salient patches in the feature space. Using a thresholding filter operator, the salient subset of a patch set can be found for a given feature space: (a) shows an input image, (b) shows region covariance after tresholding, and (c) shows MPEG-7 color structure after thresholding. The red dots indicate patches in a patch set. Using the AND operator results in the intersection of the patch sets (b) and (c). This reduces the patch sets into a more manageable, more salient patch set (f). Several of the patches in (f), shown with context in (g), match community user tagged interest regions from gigapan.org. The patch explorer (e) visualizes the patch sets according to the grouping of the permutohedral lattice. Together with the patch set query, the patch explorer provides an efficient interface for the guided exploration of gigapixel images.