Summary
EntityBench is a new benchmark for evaluating entity consistency in multi-shot video generation, introduced by Ruozhen He, Meng Wei, Ziyan Yang, and Vicente Ordonez (arXiv:2605.15199). While multi-shot video generation extends single-shot generation into coherent visual narratives, keeping characters, objects, and locations consistent across long sequences remains difficult, and existing evaluations rely on independently generated prompts with limited entity coverage and simple consistency metrics. EntityBench contains 140 episodes totaling 2,491 shots derived from real narrative media, with explicit per-shot entity schedules tracking characters, objects, and locations simultaneously. It spans easy, medium, and hard difficulty tiers of up to 50 shots, with 13 cross-shot characters, 8 cross-shot locations, 22 cross-shot objects, and recurrence gaps of up to 48 shots. The benchmark is paired with a three-pillar evaluation suite enabling standardized comparison of long-range entity consistency in video generation models.
Paper Overview
Field: Computer Vision (CV)
Authors: Ruozhen He, Meng Wei, Ziyan Yang, Vicente Ordonez
Published: 2026-05-14
arXiv: 2605.15199
Abstract
Multi-shot video generation extends single-shot generation to coherent visual narratives, yet maintaining consistent characters, objects, and locations across shots remains a challenge over long sequences. Existing evaluations typically use independently generated prompt sets with limited entity coverage and simple consistency metrics, making standardized comparison difficult.
EntityBench is a benchmark of 140 episodes (2,491 shots) derived from real narrative media, with explicit per-shot entity schedules tracking characters, objects, and locations simultaneously across easy / medium / hard tiers of up to 50 shots, 13 cross-shot characters, 8 cross-shot locations, 22 cross-shot objects, and recurrence gaps spanning up to 48 shots. It is paired with a three-pillar evaluation suite for standardized assessment of entity consistency in multi-shot video generation.
*Auto-collected on 2026-05-15.*
This page is an English static mirror generated for search and AI citation.
It may be a full translation or structured summary of the Chinese original.
Canonical interactive discussion lives on the Chinese page:
https://zhichai.net/topic/177620056