Behind Video Generation: Observations and Insights on Multimodal Understanding from Kling
cs.CV
Submitted: 2026-11-03
Updated: 2026-11-03
Terminology
Sources
- VideoPhy-2: A Challenging Action-Centric Physical Commonsense Evaluation in Video Generation
- Benchmarking and Confidence Evaluation of LALMs For Temporal Reasoning
- CapRiCorn-1K: A Comprehensive Benchmark for Video Captioning and Subject Referential Consistency Across Temporal Scales
- Towards Event-oriented Long Video Understanding
- Geometry-Guided Camera Motion Understanding in VideoLLMs
- ARC-Hunyuan-Video-7B: Structured Video Comprehension of Real-World Shorts
- Language Models Represent Space and Time
- VideoScore: Building Automatic Metrics to Simulate Fine-grained Human Feedback for Video Generation
- VideoScore2: Think before You Score in Generative Video Evaluation
- WorldSense: Evaluating Real-world Omnimodal Understanding for Multimodal LLMs
- MotionBench: Benchmarking and Improving Fine-grained Video Motion Understanding for Vision Language Models
- HunyuanVideo: A Systematic Framework For Large Video Generative Models
- GRAN-TED: Generating Robust, Aligned, and Nuanced Text Embedding for Diffusion Models
- Semantic Routing: Exploring Multi-Layer LLM Feature Weighting for Diffusion Transformers
- OmniBench: Towards The Future of Universal Omni-Language Models
- Seeing Before Reasoning: A Unified Framework for Generalizable and Explainable Fake Image Detection
- PerceptionBench: Evaluating Atomic Visual Perception in Multimodal Large Language Models
- RefCaptioner: Multi-Reference Image-Grounded Video Captioning
- LongAV-Compass: Towards Unified Evaluation of Minute-Scale Audio-Visual Generation Across T2AV, I2AV, and V2AV
- Rubrics from the Task, Verdicts from Evidence: Pointwise Rubric Rewards with Tool-Augmented Judges
Related papers
- TEA: Text Encoder Alignment for Robust Concept Erasure in Text-to-Image Models
- Do Egocentric Video-Language Models Capture Both Hand- and Object-Centric Cues?
- Aura: Consistent Multi-Subject Video Generation via VLM-Grounded Semantic Alignment
- PVRF: All-in-one Adverse Weather Removal via Prior-modulated and Velocity-constrained Rectified Flow
- DLWM: Diverse Latent World Models for Efficient Multimodal Reasoning
- Out of Sight, Still in Mind: Token Compression for Omni-LLMs