The invention relates to a method for forecasting the appearance of a textile surface to be produced from a given yarn, wherein an image database of image data sets is provided (201). Each image data set comprises an image of a textile surface produced from an image yarn and a metadata group assigned to the image. Parameter values from one metadata group each form an image vector. The given yarn is measured (202), and the parameter values (203) obtained from the measurement (202) form a yarn vector. Distances between the yarn vector on the one hand and the image vectors on the other hand are determined (206) and are compared with one another (207). Based on the comparison (207), an image data set is selected (208) from the image database, and the corresponding picture is output (209). The method provides realistic results. The method is simple, can be executed quickly, and is dynamically expandable.