We want to analyse the faces depicted in the Wenceslas Bible, an illustrated bible from the late 14th century (circa 1390), however, this requires automatic finding of faces in the illustrations of the bible. This is a difficult task due to the fact that we are working with illustrations of human faces that are often interwoven into the background and in odd poses. This paper presents an analysis of prominent face detection methods and how their performance translates from real-world facial images to painted imagery. We will make use of a scale and rotation cascade on top of these methods to see if the detection of faces in odd poses can be improved. Finally, most methods are designed to handle a particular face size, but for the purpose of finding faces in the illustration of historical books the relative size of the image and face differs from most real-world cases. An attempt to fix this is to use tiling of the input image to adjust for the relative scale difference. We will see that tiling, scaling, and rotation help, and while no general “best” setting can be given, we also see that these can boost some methods from non-working to being quite good.