Writing study: OmniScript: Towards Audio-Visual Script Generation for Long-Form Cinematic Video
← ECCV 2026 oral writing corpus
JUNFU PU, Yuxin Chen, Teng Wang, Ying Shan · Official ECCV 2026 oral record
Figure, table and section ledger
Editorial contribution type: Datasets, benchmarks & measurement (primary).
Layout extraction, not a required paper template. Counts are distinct caption labels; a multi-panel figure counts once. Section spans are estimated from heading positions, including floats and whitespace. Method and limitations.
- Canonical PDF
- 17 physical pages
- Before references / appendix
- ≈ 14.51 page-equivalents
- Main figures / tables
- 4 / 5
- Attached appendix figures / tables
- 0 / 0
- Attached appendix span
- No boundary detected; not assumed zero
Extraction notes: No attached appendix boundary detected. Zero means no recognized caption, not independently verified absence.
Source-bound correction: All top-level main headings and the reference boundary checked against source PDF text; source-bound positions use physical single-column y/page-height, not a two-column reading-order surrogate. Running page-number/author titles, affiliations, numbered contribution bullets, footnotes and bibliography continuations are excluded. This linked manuscript contains no appended supplementary section; separate supplement links were not scanned.
Section-by-section lengths and source pages
The title/abstract/front matter precede the first numbered heading. A span ends at the next detected top-level heading, not at the last sentence; end-page whitespace is included. References and statements are listed separately. These are not measurements of text-only or artwork area.
| Top-level heading | Starts on PDF page | Estimated span | Share of entire PDF |
|---|---|---|---|
| 1 Introduction | 1 | ≈ 1.86 pp. | 10.94% |
| 2 Related Work | 3 | ≈ 0.83 pp. | 4.88% |
| 3 Video-to-Script Generation | 4 | ≈ 2.15 pp. | 12.65% |
| 4 The V2S Benchmark: Metrics and Dataset | 6 | ≈ 2.07 pp. | 12.18% |
| 5 OmniScript | 8 | ≈ 2.09 pp. | 12.29% |
| 6 Experiments | 10 | ≈ 4.58 pp. | 26.94% |
| 7 Conclusion | 15 | ≈ 0.3 pp. | 1.76% |
| References | 15 | ≈ 2.49 pp. | 14.65% |
All detected figure and table locations
Main text
Attached appendix
No recognized caption labels.
Argument and organization
- 1 Introduction; 2 Related Work
Defines the missing output as a temporally grounded hierarchical script with multiple fields, contrasting it with short captions and plot summaries.
- 3 Video-to-Script Generation; 4 The V2S Benchmark; 5 OmniScript
Task schema, annotation pipeline and evaluation metrics precede the model, allowing the reader to understand what counts as a correct output before seeing architecture.
- 6 Experiments; 7 Conclusion
Tables separate event and scene fields, identify audio availability and report temporal overlap; subtitle masking and removing audio are separate controls.
Problem, claim and contribution
Defines the missing output as a temporally grounded hierarchical script with multiple fields, contrasting it with short captions and plot summaries.
Abstract and Section 1, pp.1–2 · Main paper
Argument and section order
Task schema, annotation pipeline and evaluation metrics precede the model, allowing the reader to understand what counts as a correct output before seeing architecture.
Sections 3–5, pp.4–9 · Main paper
Figures, captions and visual explanation
The architecture aligns video frames, audio and timestamped token streams; duration plots use the same horizontal groups for F1 and recall.
Figure 3, p.8; Figure 4, p.13 · Main paper
Tables, comparisons and evidence
Tables separate event and scene fields, identify audio availability and report temporal overlap; subtitle masking and removing audio are separate controls.
Tables 1–2, pp.11–12; Tables 4–5, p.14 · Main paper
Appendix and reproducibility
The hierarchical metric definition points to supplementary details, but the metric stages remain described in the main manuscript. The separate supplement was not retrieved.
Section 4.1, pp.6–7 · Main paper
Clarity and writing conventions
Schema-level field names recur across task description, model outputs and tables, reducing the distance between a qualitative example and measured errors.
Sections 3.1 and 6.2, pp.4,11 · Main paper
What to reuse—and what not to infer
Transferable lesson: A new structured-output benchmark needs an output example, annotation-quality explanation and metric decomposition before its model leaderboard.
Boundary: The benchmark contains ten cinematic works; avoid claiming universal long-video coverage. The source describes four progressive stages in Section 5.2 but calls the recipe three-stage in Section 6.1, a wording inconsistency to avoid.
Apply this lesson using the argument and evidence worksheet.
Inspection record and sources
- Checked
- 2026-10-02
- Reading status
- Writing-focused manuscript scan + selected visuals inspected
- Main text
- Writing-oriented scan across the manuscript main argument, physical PDF pp.1–15: framing and section roadmaps, method/argument explanations, figure and table captions, comparison protocols, conclusions and limitations. Selected full-page rasters were inspected. This is not exhaustive line-by-line technical reading, proof verification or experimental reproduction.
- Appendix / supplement
- Main-manuscript pointers to appendix/supplement and the reference boundary were checked. Separate supplementary PDFs were not retrieved or inspected; their contents are not included in inspected-page counts.
- Visually inspected PDF pages
- 8, 13
This is a writing-focused scan of the manuscript and the stated supplementary scope: framing, organization, visual explanation, evidence presentation and wording. It is not a line-by-line proof audit, reproduction of results or exhaustive scientific review.