# EntityBench: Towards Entity-Consistent Long-Range Multi-Shot Video Generation

Source: [EntityBench: Towards Entity-Consistent Long-Range Multi-Shot Video Generation](<https://www.sota2.com/research/paper/6a06a8ccdbc836b87d82a3a2>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-05-16T05:00:12.406Z

Authors: Ruozhen He, Meng Wei, Ziyan Yang, Vicente Ordonez

Published: 2026-05-14T17:59:55.000Z

## 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. We introduce EntityBench, 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 that disentangles intra-shot quality, prompt-following alignment, and cross-shot consistency, with a fidelity gate that admits only accurate entity appearances into cross-shot scoring. As a baseline, we propose EntityMem, a memory-augmented generation system that stores verified per-entity visual references in a persistent memory bank before generation begins. Experiments show that cross-shot entity consistency degrades sharply with recurrence distance in existing methods, and that explicit per-entity memory yields the highest character fidelity (Cohen's d = +2.33) and presence among methods evaluated. Code and data are available at https://github.com/Catherine-R-He/EntityBench/.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Pillar 2 Intra-shot prompt-following alignment](<https://www.sota2.com/research/sota/pillar-2-intra-shot-prompt-following-alignment.md>): Intra-shot prompt-following alignment; Pillar 2 Intra-shot prompt-following alignment; Intra-shot Character Presence: 96.7.

- [Intra-shot quality evaluation on EntityBench](<https://www.sota2.com/research/sota/intra-shot-quality-evaluation-on-entitybench.md>): Intra-shot quality evaluation; EntityBench; Subject Consistency: 88.1.

- [Video Generation on EntityBench Intra-shot 1.0](<https://www.sota2.com/research/sota/video-generation-on-entitybench-intra-shot-1-0.md>): Video Generation; EntityBench Intra-shot 1.0; Imaging Quality: 66.

- [Pillar 3 Cross-shot consistency](<https://www.sota2.com/research/sota/pillar-3-cross-shot-consistency.md>): Cross-shot consistency; Pillar 3 Cross-shot consistency; CS Consistency (Face): 73.7.

- [Video Generation on EntityBench Cross-shot 1.0](<https://www.sota2.com/research/sota/video-generation-on-entitybench-cross-shot-1-0.md>): Video Generation; EntityBench Cross-shot 1.0; Cross-shot Face Consistency: 73.7.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2605.15199>)

[PDF](<https://arxiv.org/pdf/2605.15199.pdf>)