# Video Generation on EntityBench Cross-shot 1.0

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

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

Data updated: 2026-05-15T10:07:25.351Z

Evaluation of video generation models on the EntityBench Cross-shot 1.0 dataset, focusing on cross-shot consistency metrics and LLM-based scoring for face, object, and scene fidelity.

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.

| Scope | Recorded value |
| --- | --- |
| Task | Video Generation |
| Dataset (including any stated split) | EntityBench Cross-shot 1.0 |
| Metric | Cross-shot Face Consistency |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 4 of 4 |
| Coverage | All results currently returned for this benchmark |

Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol.

## Metrics

- [Cross-shot Face Consistency](<https://www.sota2.com/research/sota/video-generation-on-entitybench-cross-shot-1-0.md?metric=Cross-shot%20Face%20Consistency>): Higher is better; unit: PERCENT.

- [Cross-shot Object Consistency](<https://www.sota2.com/research/sota/video-generation-on-entitybench-cross-shot-1-0.md?metric=Cross-shot%20Object%20Consistency>): Higher is better; unit: PERCENT.

- [Cross-shot Transition Boundary Score](<https://www.sota2.com/research/sota/video-generation-on-entitybench-cross-shot-1-0.md?metric=Cross-shot%20Transition%20Boundary%20Score>): Higher is better; unit: PERCENT.

- [LLM Face Accuracy](<https://www.sota2.com/research/sota/video-generation-on-entitybench-cross-shot-1-0.md?metric=LLM%20Face%20Accuracy>): Higher is better; unit: PERCENT.

- [LLM Face Mean Score](<https://www.sota2.com/research/sota/video-generation-on-entitybench-cross-shot-1-0.md?metric=LLM%20Face%20Mean%20Score>): Higher is better; unit: PERCENT.

- [LLM Face Fidelity Score](<https://www.sota2.com/research/sota/video-generation-on-entitybench-cross-shot-1-0.md?metric=LLM%20Face%20Fidelity%20Score>): Higher is better; unit: PERCENT.

- [LLM Object Accuracy](<https://www.sota2.com/research/sota/video-generation-on-entitybench-cross-shot-1-0.md?metric=LLM%20Object%20Accuracy>): Higher is better; unit: PERCENT.

- [LLM Object Mean Score](<https://www.sota2.com/research/sota/video-generation-on-entitybench-cross-shot-1-0.md?metric=LLM%20Object%20Mean%20Score>): Higher is better; unit: PERCENT.

- [LLM Object Shape Score](<https://www.sota2.com/research/sota/video-generation-on-entitybench-cross-shot-1-0.md?metric=LLM%20Object%20Shape%20Score>): Higher is better; unit: PERCENT.

- [LLM Scene Accuracy](<https://www.sota2.com/research/sota/video-generation-on-entitybench-cross-shot-1-0.md?metric=LLM%20Scene%20Accuracy>): Higher is better; unit: PERCENT.

- [LLM Scene Mean Score](<https://www.sota2.com/research/sota/video-generation-on-entitybench-cross-shot-1-0.md?metric=LLM%20Scene%20Mean%20Score>): Higher is better; unit: PERCENT.

- [LLM Scene Layout Score](<https://www.sota2.com/research/sota/video-generation-on-entitybench-cross-shot-1-0.md?metric=LLM%20Scene%20Layout%20Score>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Cross-shot Face Consistency | Result date | Report paper |
| --- | --- | --- | --- |
| StoryMem | 79.2 | Not recorded | 6a06a8ccdbc836b87d82a3a2 |
| CineTrans | 77.2 | Not recorded | 6a06a8ccdbc836b87d82a3a2 |
| HoloCine | 75.1 | Not recorded | 6a06a8ccdbc836b87d82a3a2 |
| EntityMem | 73.7 | Not recorded | 6a06a8ccdbc836b87d82a3a2 |

### StoryMem

Cross-shot Face Consistency: 79.2. Result ID: 6a06f01b47f34ce924ef2841.

[Reporting paper](<https://www.sota2.com/research/paper/6a06a8ccdbc836b87d82a3a2.md>)

[Method origin paper](<https://www.sota2.com/research/paper/694a224c9eebe296e72aaac2.md>)

[Implementation](<https://github.com/Kevin-thu/StoryMem>)

Recorded evaluation configuration: {"Paradigm":"two-stage keyframe-then-animate"}. This record does not assert independent reproduction.

### CineTrans

Cross-shot Face Consistency: 77.2. Result ID: 6a06f01b47f34ce924ef2843.

[Reporting paper](<https://www.sota2.com/research/paper/6a06a8ccdbc836b87d82a3a2.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7ab41e209131cea84b40.md>)

Recorded evaluation configuration: {"Paradigm":"holistic"}. This record does not assert independent reproduction.

### HoloCine

Cross-shot Face Consistency: 75.1. Result ID: 6a06f01b47f34ce924ef2842.

[Reporting paper](<https://www.sota2.com/research/paper/6a06a8ccdbc836b87d82a3a2.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fc597a3ef7ccd74e45278b.md>)

Recorded evaluation configuration: {"Paradigm":"holistic"}. This record does not assert independent reproduction.

### EntityMem

Cross-shot Face Consistency: 73.7. Result ID: 6a06f01b47f34ce924ef2840.

[Reporting paper](<https://www.sota2.com/research/paper/6a06a8ccdbc836b87d82a3a2.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a06a8ccdbc836b87d82a3a2.md>)

Recorded evaluation configuration: {"Paradigm":"two-stage keyframe-then-animate"}. This record does not assert independent reproduction.

Page 1 of 1. This is an index page, not the full collection.