Research map / Video / Video generation
AutoCut: End-to-end advertisement video editing based on multimodal discretization and controllable generation
Milton Zhou, Sizhong Qin, Yongzhi Li, Quan Chen, Peng Jiang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add multimodal alignment stage before SFT (ablation) | AutoCut (Qwen3-8B) · Ad video editing (364 videos) | CRA: 0.08242 -> 0.10714 | Improved |
| add multimodal alignment stage before SFT (ablation) | AutoCut (Qwen3-8B) · Ad video editing (364 videos) | CSA: 0.4780 -> 0.6593 | Improved |
| add multimodal alignment stage before SFT (ablation) | AutoCut (Qwen3-8B) · Ad video editing (364 videos) | WCD: 4.4346 -> 3.0182 | Improved |
| add extra pre-training stage after alignment (ablation) | AutoCut (Qwen3-8B) · Ad video editing (364 videos) | CRA, SQ, WCD: CRA 0.10714 -> 0.05770; SQ 84.6255 -> 78.9644; WCD 3.0182 -> 4.4984 | Hurt |
| add extra pre-training stage after alignment (ablation) | AutoCut (Qwen3-8B) · Ad video editing (364 videos) | CSA: 0.6593 -> 0.7170 | Improved |
| fine-tune Qwen3-8B on caption inputs with same SFT data | Qwen3-8B (Caption) · Ad video editing (364 videos) | CSA: 0.1374 -> 0.5687 | Improved |
| use discretized video tokens instead of captions with SFT | Qwen3-8B · Ad video editing (364 videos) | SQ: 59.1592 -> 84.6255 | Improved |
| use AutoCut instead of GPT-4o pipeline | ad video editing · Inference cost per 100 videos | cost: $2.5 -> $0.015 | Improved |
Closest papers by meaning
- EditBoard: Towards a Comprehensive Evaluation Benchmark for Text-Based Video Editing Models (AAAI 2025 · similarity 0.83)
- FADE: Frequency-Aware Diffusion Model Factorization for Video Editing (CVPR 2025 · similarity 0.83)
- Transforming Image Editors into Video Editors (NeurIPS 2026 · similarity 0.83)
- MaskINT: Video Editing via Interpolative Non-autoregressive Masked Transformers (CVPR 2024 · similarity 0.82)
- AVENUE: Audio-Video Editing Understanding and Evaluation (NeurIPS 2026 · similarity 0.82)
- Align-A-Video: Deterministic Reward Tuning of Image Diffusion Models for Consistent Video Editing (CVPR 2025 · similarity 0.82)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
