Purchase this article with an account.
Ran Carmi, Laurent Itti; Why do we fail to perceive jump-cuts in motion pictures?. Journal of Vision 2005;5(8):548. doi: https://doi.org/10.1167/5.8.548.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Motivation: For more than a century, motion pictures have been extremely successful in attracting people's attention, yet their psychology is poorly understood or even addressed by the scientific community. One of the more puzzling practices in motion pictures is the frequent use of jump-cuts (abrupt transitions between adjacent scene shots), which are the staple of music television (MTV). Attention research suggests that our seemingly continuous and detailed perception of the real world is the product of highly incomplete internal representations that depend on selective on-demand sampling of continuous environmental inputs. If input continuity is so important, why do jump-cuts often go unnoticed?
Methods: In order to examine the effects of jump-cuts on attentional allocation, we first constructed MTV-style clips, which featured persistent context for 1–3 seconds, from a diverse collection of continuous clips that depict photography-based and computer-generated dynamic scenes. We then performed two eye-tracking experiments with separate groups of subjects, each inspecting either continuous or MTV-style clips. In order to measure the persistence of attention-guiding representations (AGRs), we quantified changes in their inputs and outputs using either local intensity contrast or saliency as probes for the ongoing impact of bottom-up influences on saccade target selection.
Results: Jump-cuts update AGRs within less than 250 ms. AGRs persist for less than 2 seconds even during inspection of continuous clips.
Conclusions: We propose that perceptual continuity is often unperturbed across jump-cuts, despite physical discontinuities, thanks to the briefness and sparseness of mental representations, combined with the ingenuity of moviemakers in manipulating these representations. Our results indicate that integrating computational attention research with the art and technology of moviemaking is technically feasible, and can advance the understanding and practice of both fields.
This PDF is available to Subscribers Only