arXiv:2508.15774v2 Announce Type: replace Abstract: Visual diffusion models achieve remarkable progress, yet they are typically trained at limited resolutions due to the lack of high-resolution data...
arXiv:2605. 16223v2 Announce Type: replace-cross Abstract: Generative video models are increasingly used in design animation tasks, yet no standardized evaluation framework exists for this domain.
We explore large-scale training of generative models on video data. Specifically, we train text-conditional diffusion models jointly on videos and images of variable durations, resolutions and aspect ratios.
arXiv:2608.29123v1 Announce Type: new Abstract: Training a video-generation model from scratch is hard for reasons that precede model design. The feedback loop is long: a failure that appears only af...
The paper presents a framework for converting standard dynamic range (SDR) videos into high dynamic range (HDR) videos using large-scale generative video models. It introduces a Multi-Exposure Video Model (MEVM) that predicts exposure-bracketed linear SDR sequences from a single nonlinear SDR input, and a Video Merging Model (VMM) that fuses these predictions into a high-quality HDR sequence while preserving detail in shadows and highlights. Experiments, qualitative evaluation, and a user study demonstrate robust HDR conversion for casual consumer footage and iconic films, and the approach can be integrated into existing SDR generative video pipelines.
RefVideo-6M is a new large-scale reference-guided editing dataset that includes 5 million video editing samples and 1 million image editing samples, each paired with about 6 million visual references. The dataset is constructed to avoid artifacts by using real, artifact‑free videos as targets and filtering input conditions with multiple editing experts, thereby providing reliable supervision. It enables models to learn fine‑grained visual correspondence beyond text‑only instructions and supports the training of a reference‑guided video editing model, Ref‑MoT, which shows improved visual quality, controllability, and reference consistency.
AnyView is a diffusion-based video generation framework designed for dynamic view synthesis, requiring minimal inductive biases or geometric assumptions. It trains a generalist spatiotemporal implicit representation using diverse data sources—monocular, multi-view static, and multi-view dynamic—to produce zero-shot novel videos from arbitrary camera locations and trajectories. The authors evaluate AnyView on standard benchmarks, introduce a new challenging benchmark called AnyViewBench for extreme dynamic view synthesis, and demonstrate that AnyView outperforms baselines in maintaining realistic, plausible, and spatiotemporally consistent videos across diverse real-world scenarios.
arXiv:2608.29331v1 Announce Type: new Abstract: Video traffic constitutes a significant share of global web traffic. To reduce its volume, video codecs have been developed and continuously improved....
Music-driven dance video generation aims to synthesize expressive human motion that is temporally aligned with music while maintaining high visual fidelity. Despite recent progress, existing methods still face two key limitations: the lack of large-scale, high-quality dance video datasets, and the absence of principled frameworks for integrating music as a complementary conditioning signal into Video Generation Foundation Models.
arXiv:2606. 09056v1 Announce Type: cross Abstract: Video generative models have become increasingly powerful, but long-range consistency remains challenging to achieve because even a few dozen frames require impractically long transformer sequence lengths.
arXiv:2608. 19583v1 Announce Type: cross Abstract: Recent studies suggest that video generation models can exhibit certain forms of zero-shot visual reasoning through generated frames.