# Cinematography Language Understanding on ShotBench (test)

Source: [Cinematography Language Understanding on ShotBench (test)](<https://www.sota2.com/research/sota/cinematography-language-understanding-on-shotbench-test>)

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

Data updated: 2026-03-31T06:59:54.486Z

Evaluation of models on the ShotBench dataset focusing on understanding cinematographic language and structure, measured across specific components like shot sequence, framing, and linguistic cues.

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 | Cinematography Language Understanding |
| Dataset (including any stated split) | ShotBench (test) |
| Metric | Shot Sequence Score |
| 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

- [Shot Sequence Score](<https://www.sota2.com/research/sota/cinematography-language-understanding-on-shotbench-test.md?metric=Shot%20Sequence%20Score>): Higher is better; unit: PERCENT.

- [Shot Framing Score](<https://www.sota2.com/research/sota/cinematography-language-understanding-on-shotbench-test.md?metric=Shot%20Framing%20Score>): Higher is better; unit: PERCENT.

- [Camera Angle Understanding](<https://www.sota2.com/research/sota/cinematography-language-understanding-on-shotbench-test.md?metric=Camera%20Angle%20Understanding>): Higher is better; unit: PERCENT.

- [Linguistic Structure Score](<https://www.sota2.com/research/sota/cinematography-language-understanding-on-shotbench-test.md?metric=Linguistic%20Structure%20Score>): Higher is better; unit: PERCENT.

- [Long-term Temporal Understanding](<https://www.sota2.com/research/sota/cinematography-language-understanding-on-shotbench-test.md?metric=Long-term%20Temporal%20Understanding>): Higher is better; unit: PERCENT.

- [Linguistic Content Score](<https://www.sota2.com/research/sota/cinematography-language-understanding-on-shotbench-test.md?metric=Linguistic%20Content%20Score>): Higher is better; unit: PERCENT.

- [Shot Context Score](<https://www.sota2.com/research/sota/cinematography-language-understanding-on-shotbench-test.md?metric=Shot%20Context%20Score>): Higher is better; unit: PERCENT.

- [Cinematic Mapping Score](<https://www.sota2.com/research/sota/cinematography-language-understanding-on-shotbench-test.md?metric=Cinematic%20Mapping%20Score>): Higher is better; unit: PERCENT.

- [Average Score](<https://www.sota2.com/research/sota/cinematography-language-understanding-on-shotbench-test.md?metric=Average%20Score>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Shot Sequence Score | Result date | Report paper |
| --- | --- | --- | --- |
| ShotVL | 77.9 | Not recorded | 68fa79b31e209131cea7ef72 |
| Qwen2.5-VL-72B-Instruct | 75.1 | Not recorded | 68fa79b31e209131cea7ef72 |
| GPT-4o | 69.3 | Not recorded | 68fa79b31e209131cea7ef72 |
| Qwen2.5-VL-3B-Instruct | 54.6 | Not recorded | 68fa79b31e209131cea7ef72 |

### ShotVL

Shot Sequence Score: 77.9. Result ID: 69cb6fef47f34ce924b334eb.

[Reporting paper](<https://www.sota2.com/research/paper/68fa79b31e209131cea7ef72.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa79b31e209131cea7ef72.md>)

Recorded evaluation configuration: {"Parameters":"3B"}. This record does not assert independent reproduction.

### Qwen2.5-VL-72B-Instruct

Shot Sequence Score: 75.1. Result ID: 69cb6fef47f34ce924b334e8.

[Reporting paper](<https://www.sota2.com/research/paper/68fa79b31e209131cea7ef72.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa773a1e209131cea71d3c.md>)

Recorded evaluation configuration: {"Parameters":"72B"}. This record does not assert independent reproduction.

### GPT-4o

Shot Sequence Score: 69.3. Result ID: 69cb6fef47f34ce924b334e9.

[Reporting paper](<https://www.sota2.com/research/paper/68fa79b31e209131cea7ef72.md>)

Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction.

### Qwen2.5-VL-3B-Instruct

Shot Sequence Score: 54.6. Result ID: 69cb6fef47f34ce924b334ea.

[Reporting paper](<https://www.sota2.com/research/paper/68fa79b31e209131cea7ef72.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa773a1e209131cea71d3c.md>)

Recorded evaluation configuration: {"Parameters":"3B"}. This record does not assert independent reproduction.

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