Cognitive psychology has a long history of using physiological measures, such as pupillometry. However, their susceptibility to confounds introduced by stimulus properties, such as color and luminance, has limited their application. Pupil size measurements, in particular, require sophisticated experimental designs to dissociate relatively small changes in pupil diameter due to cognitive responses from larger ones elicited by changes in stimulus properties or the experimental environment. Here, building on previous research, we present a pupillometry paradigm that adapts the pupil to stimulus properties during the baseline period without revealing stimulus meaning or context by using a pixel-scrambled image mask around an intact image. We demonstrate its robustness in the context of pupillary responses to branded product familiarity. Results show larger average and peak pupil dilation for passively viewed familiar product images and an extended later temporal component representing differences in familiarity across participants (starting around 1400 ms post-stimulus onset). These amplitude differences are present for almost all participants at the single-participant level, and vary somewhat by product category. However, amplitude differences were absent during the baseline period. These findings demonstrate that involuntary pupil size measurements combined with the presented paradigm are successful in dissociating cognitive effects of familiarity from physical stimulus confounds.