Search

Recent advancements in video generation models, like Stable Video Diffusion, show promising results, but primarily focus on short, single-scene videos. These models struggle with generating long videos that involve multiple scenes, coherent narratives, and consistent characters. Furthermore, there is no publicly available dataset tailored for the analysis, evaluation, and training of long video generation models. In this paper, we present MovieBench: A Hierarchical Movie-Level Dataset for Long Video Generation, which addresses these challenges by providing unique contributions: (1) movie-length videos featuring rich, coherent storylines and multi-scene narratives, (2) consistency of character appearance and audio across scenes, and (3) hierarchical data structure contains high-level movie information and detailed shot-level descriptions. Experiments demonstrate that MovieBench brings some new insights and challenges, such as maintaining character ID consistency across multiple scenes for various characters. The dataset will be public and continuously maintained, aiming to advance the field of long video generation. Data can be found at: https://weijiawu.github.io/MovieBench/., Comment: The project website is at: https://weijiawu.github.io/MovieBench/. Code: https://github.com/showlab/MovieBecnh

Concepts involved in long-form videos such as people, objects, and their interactions, can be viewed as following an implicit prior. They are notably complex and continue to pose challenges to be comprehensively learned. In recent years, generative pre-training (GPT) has exhibited versatile capacities in modeling any kind of text content even visual locations. Can this manner work for learning long-form video prior? Instead of operating on pixel space, it is efficient to employ visual locations like bounding boxes and keypoints to represent key information in videos, which can be simply discretized and then tokenized for consumption by GPT. Due to the scarcity of suitable data, we create a new dataset called \textbf{Storyboard20K} from movies to serve as a representative. It includes synopses, shot-by-shot keyframes, and fine-grained annotations of film sets and characters with consistent IDs, bounding boxes, and whole body keypoints. In this way, long-form videos can be represented by a set of tokens and be learned via generative pre-training. Experimental results validate that our approach has great potential for learning long-form video prior. Code and data will be released at \url{https://github.com/showlab/Long-form-Video-Prior}.

This work examines the coalescence of two unequal-size spherical liquid droplets in the inviscid regime, with an emphasis on exploring the scaling of the liquid bridge evolution. Our experiment suggests that the classical 1/2 power-law scaling for equal-size droplets still holds for the unequal-size situation of small size ratios, but it diverges as the size ratio increases. Employing an energy balance analysis, we develop a theoretical model to collapse the experimental data of different droplet size ratios. The model reveals an exponential dependence of the bridge's radial growth on time, implying an intrinsic breaking of scaling law. The scale-free evolution behavior is evident only at late coalescence time and large size ratio, which can be explained using the length and time scales obtained from the theory.

When conducting autonomous scanning for the online reconstruction of unknown indoor environments, robots have to be competent at exploring scene structure and reconstructing objects with high quality. Our key observation is that different tasks demand specialized scanning properties of robots: rapid moving speed and far vision for global exploration and slow moving speed and narrow vision for local object reconstruction, which are referred as two different scanning modes: explorer and reconstructor, respectively. When requiring multiple robots to collaborate for efficient exploration and fine-grained reconstruction, the questions on when to generate and how to assign those tasks should be carefully answered. Therefore, we propose a novel asynchronous collaborative autoscanning method with mode switching, which generates two kinds of scanning tasks with associated scanning modes, i.e., exploration task with explorer mode and reconstruction task with reconstructor mode, and assign them to the robots to execute in an asynchronous collaborative manner to highly boost the scanning efficiency and reconstruction quality. The task assignment is optimized by solving a modified Multi-Depot Multiple Traveling Salesman Problem (MDMTSP). Moreover, to further enhance the collaboration and increase the efficiency, we propose a task-flow model that actives the task generation and assignment process immediately when any of the robots finish all its tasks with no need to wait for all other robots to complete the tasks assigned in the previous iteration. Extensive experiments have been conducted to show the importance of each key component of our method and the superiority over previous methods in scanning efficiency and reconstruction quality., Comment: 13pages, 12 figures, Conference: SIGGRAPH Asia 2022

This paper presents an experimental study on the emergence of the central recirculation zone (CRZ) in a V-shaped premixed swirling flame, using simultaneous measurement of particle image velocimetry (PIV) and CH* chemiluminescence. The results show that either increasing the Reynolds number (Re) or decreasing the equivalence ratio ({\Phi}) would facilitate the emergence of CRZ. Further analysis demonstrates that the CRZ characteristics and its emergence are strongly influenced by the inner shear layer (ISL) surrounding the CRZ, while the swirl intensity remains unchanged. Dimensional analysis is performed to understand the underlying mechanism, suggesting the CRZ emergence is controlled by a non-dimensional parameter, Re_s=|{\gamma}|_max D/{\nu}_s, defined based on the maximum ISL intensity (|{\gamma}|_max), the exit diameter (D), and the kinematic viscosity ({\nu}_s) of the burnt gas. By estimating the temperature and viscosity with a simple heat-loss model, we show in the |{\gamma}|_max D-{\nu}_s regime diagram that the cases with and without CRZ are separated by a single boundary line, corresponding to a critical Re_s of about 424. This verifies the applicability of the proposed Re_s criterion to lean-premixed V-shaped swirling flames under various conditions. Unlike most previous works that attribute the CRZ of swirling flames to vortex breakdown, the present work reveals the non-negligible effect of the ISL, especially the CRZ suppression when the ISL is weakened by flame heating.

This letter presents a theory on the coalescence of two spherical liquid droplets that are initially stationary. The evolution of the radius of a liquid neck formed upon coalescence was formulated as an initial value problem and then solved to yield an exact solution without free parameters, with its two asymptotic approximations reproducing the well-known scaling relations in the viscous and inertial regimes. The viscous-to-inertial crossover observed by Paulsen et al. [Phys. Rev. Lett. 106, 114501 (2011)] is also recovered by the theory, rendering the collapse of data of different viscosities onto a single curve., Comment: 5 pages; 4 figures

The off-center collision of binary bouncing droplets of equal size was studied numerically by a volume-of-fluid (VOF) method with two marker functions, which has been validated by comparing with available experimental results. A non-monotonic kinetic energy recovery with varying impact parameters was found based on the energy budget analysis. This can be explained by the prolonged entanglement time and the enhanced internal-flow-induced viscous dissipation for bouncing droplets at intermediate impact parameters, compared with those at smaller or larger impact parameters. The universality of this non-monotonicity was numerically verified, and thereby an approximate fitting formula was proposed to correlate the kinetic energy dissipation factor with the impact parameter for various Weber numbers and Ohnesorge numbers. From the vortex dynamics perspective, a helicity analysis of droplet internal flow identifies a strong three-dimensional interaction between the "ring-shaped" vortices and the "line-shaped" shear layers for off-center collisions. Furthermore, we demonstrated theoretically and verified numerically that the equivalence between the total enstrophy and the total viscous dissipation, which holds for a single-phase flow system confined by stationary boundaries, is not generally satisfied for the two-phase flow system containing gas-liquid interfaces. This is attributed to the work done by the unbalanced viscous stresses, which results from the interfacial flow and the vorticity associated with the movement of the oscillating interface., Comment: 41 Pages, 12 figures

To carry out autonomous 3D scanning and online reconstruction of unknown indoor scenes, one has to find a balance between global exploration of the entire scene and local scanning of the objects within it. In this work, we propose a novel approach, which provides object-aware guidance for autoscanning, for exploring, reconstructing, and understanding an unknown scene within one navigation pass. Our approach interleaves between object analysis to identify the next best object (NBO) for global exploration, and object-aware information gain analysis to plan the next best view (NBV) for local scanning. First, an objectness-based segmentation method is introduced to extract semantic objects from the current scene surface via a multi-class graph cuts minimization. Then, an object of interest (OOI) is identified as the NBO which the robot aims to visit and scan. The robot then conducts fine scanning on the OOI with views determined by the NBV strategy. When the OOI is recognized as a full object, it can be replaced by its most similar 3D model in a shape database. The algorithm iterates until all of the objects are recognized and reconstructed in the scene. Various experiments and comparisons have shown the feasibility of our proposed approach., Comment: 12 pages, 17 figures

Anti-phase and in-phase flickering modes of dual buoyant diffusion flames were numerically investigated and theoretically analyzed in this study. Inspired by the flickering mechanism of a single buoyant diffusion flame, for which the deformation, stretching, or even pinch-off of the flame surface result from the formation and evolution of the toroidal vortices, we attempted to understand the anti-phase and in-phase flickering of dual buoyant diffusion flames from the perspective of vortex dynamics. The interaction between the inner-side shear layers of the two flames was identified to be responsible for the different flickering modes. Specifically, the transition between anti-phase and in-phase flickering modes can be predicted by a unified regime nomogram of the normalized flickering frequency versus a characteristic Reynolds number, which accounts for the viscous effect on vorticity diffusion between the two inner-side shear layers. Physically, the transition of the vortical structures from symmetric (in-phase) to staggered (anti-phase) in a dual-flame system can be interpreted as being similar to the mechanism causing flow transition in the wake of a bluff body and forming the Karman vortex street., Comment: 32 pages, 14 figures. V2, in which new computations and theoretical analysis are added, has additional figures, references, and an appendix

Previous studies have found that the flickering of buoyant diffusion flames is associated with the periodic shedding of a toroidal vortex, which is formed under gravity-induced shearing at the flames. Moreover, numerous experimental investigations have reported the scaling relation, $f \propto D^{-1/2}$, where $f$ is the flickering frequency and $D$ is the diameter of the fuel inlet. However, the connection between the toroidal vortex mechanism and the scaling relation has not been clearly understood. The current study theoretically revisits this problem from the perspective of vortex dynamics. The theory incorporates a recent finding in the vortex dynamics community: the detachment of a continuously growing vortex ring is inevitable and can be dictated by a universal constant that is essentially a non-dimensional circulation of the vortex. By calculating the total circulation of the toroidal vortex and applying the vortex ring detachment criterion, we mathematically established the connection between the lifetime of a toroidal vortex and the flickering of a buoyant diffusion flame, resulting in a general flickering frequency formulation, which is validated by comparing with existing experimental data of pool and jet diffusion flames.