Interactive storytelling via video content recombination

Abstract

Rights'ACM Allow an authors' version of their own ACMcopyrighted work on their personal server or on servers belonging to their employers ' ABSTRACTIn the paper we present a prototype of video-based storytelling that is able to generate multiple story variants from a baseline video. The video content for the system is generated by an adaptation of forefront video summarisation techniques that decompose the video into a number of Logical Story Units (LSU) representing sequences of contiguous and interconnected shots sharing a common semantic thread. Alternative storylines are generated using AI Planning techniques and these are used to direct the combination of elementary LSU for output. We report early results from experiments with the prototype in which the reordering of video shots on the basis of their high-level semantics produces trailers giving the illusion of different storylines.

Automatic generation of video narratives from shared UGC

“…Nevertheless, examples of configurable or interactive video narratives exist. A recent result is reported in [Porteous, 2010], which describes a video-based storytelling prototype that is able to generate multiple story variants from a baseline video. This is more related to video summarization than to narrative aggregation of content from various sources.…”

Generating story variants with constrained video recombination

“…Our earlier approach [23] attempted to directly combine state-of-the-art plan-based narrative generation with video content. It used somewhat naive mappings between planning actions and video segments with the consequence that the output was restricted to variants that were simple rearrangements of the original input story, without fundamentally changing the original semantics of the actions.…”

“…This can't be done without video analysis level reasoning about local causality and consistency and hence our approach integrates it with a joint though separate high level reasoning about global narrative properties. Each action in an output narrative can map to multiple points in the shared semantic subspace (see Section 4), so the video presentation of individual narrative actions isn't limited to any single segment of the baseline video (as it is with the naive mapping used in [23]); instead, appropriate video subparts for a narrative action can be selected from any video segments which map to relevant points for that action in the semantic subspace, while preserving fundamental consistency in the content. This is a powerful result since it provides a complete decoupling of the narrative model and the baseline video content which is only described by its semantics.…”

“…In this section we consider the key enablers of narrative integration in the system: the narrative domain actions and mappings from them to semantic patterns; and the use of the shared semantic space in constraining narrative generation (further details of the planner can be found in [23]). …”

Script-to-Movie: A Computational Framework for Story Movie Composition

“…These steps usually contain massive manual operations, which are both labor-intensive and skill-required. To ease the burden, researchers proposed mashup [2] and remix [3] ideas to collect materials from existing movies, and used interactive approaches [4], [5] to simplify the video content organization. However, in any case, the manual intervention remains an indispensable step to illustrate various ingenious story plots and complex character interactions.…”