Papers

Interactive control for video generation is moving from coarse prompts toward fine-grained, physically meaningful manipulation of dynamic scenes. Yet existing controllable methods either require the full control schedule before generation starts, or use pixel-space signals that dictate object positions rather than physical dynamics. To address these limitations, we propose PhysStream, an autoregressive model for physics-grounded image-to-video synthesis that incorporates structured scene memory—positional maps and object tracking maps derived online from previously generated frames—and supports fine-grained motion control via sparse velocity-increment signals that encode physical quantities, letting the model learn the underlying dynamics. We train our model in two stages: a bidirectional model is first finetuned with motion-control conditioning, then a causal autoregressive model is trained with additional structured scene memory, further improving physical consistency. PhysStream enables interactive, mid-generation control over multi-object tabletop rigid-body scenes—a capability not supported by prior methods—reducing motion distribution distance (FVMD) by 33% and trajectory error by 12% over the strongest baselines on synthetic benchmarks, and is preferred by human evaluators in over 85% of in-the-wild comparisons. Please check our website for more details: https://czzzzh.github.io/PhysStream

We present an automated pipeline that converts hair-card models into high-quality strand-based hairstyles. Given a collection of textured triangular or quad strips as input, our method produces a strand-based representation that preserves the original hairstyle while enriching it with fine-scale geometric detail and adhering to standard production requirements: strands originate from the scalp, roots are uniformly distributed, and the hair volume is plausibly filled. The resulting assets are directly compatible with strand-based rendering, physics-based simulation, and common grooming modifiers (e.g., clumping, curling, noise) for enhanced realism and artistic control. We validate our approach on a large and diverse set of hairstyles, including short and long hair, curly styles, and complex styles such as buns and ponytails.

We propose a decoupled reconstruction framework that temporarily decomposes a NURBS representation into independent local Active Sections, enabling arbitrary local knot insertion, degree elevation, and basis modifications while preserving exact CAD geometry. However, these independent modifications inevitably violate inter-patch continuity, requiring a robust algebraic recovery of global admissibility. We introduce a novel reconstruction methodology that constructs a positive Hybrid Reconstruction Operator: anchor degrees of freedom are identified via QR pivoting, after which a sequence of linear programming problems with distance-based regularization generates strictly non-negative nullspace vectors, while a subsequent non-negative least-squares solve enforces a precise partition of unity. Crucially, to ensure computational efficiency and preserve locality, we implement a hierarchical pairwise condensation strategy that freezes columns already satisfying new interface constraints, confining the optimization exclusively to the active degrees of freedom at each merge step. The resulting hybrid basis is, by construction, strictly non-negative, spans the exact constrained nullspace, preserves partition of unity to machine precision, and reproduces the original geometry exactly. Numerical benchmarks, including a nonlinear diffusion problem on multi-patch curves with heterogeneous local polynomial degrees, confirm optimal convergence rates and demonstrate the framework's ability to seamlessly combine local geometric flexibility with globally consistent approximation spaces. By entirely decoupling local spline editing from the enforcement of continuity, this methodology provides a highly efficient, mathematically principled, and universally applicable tool for adaptive Isogeometric Analysis.

Interocular differences in visual perception can induce a variety of effects when fused by the brain. For example, prior works have found that carefully crafted binocular differences in local detail can improve contrast. It has also been found that when the frequency content of two stimuli are slightly different, blur suppression leads to a fused percept that is typically dominated by the sharper image. In this paper, we develop a psychophysical framework to measure the perception of natural image stimuli with interocular frequency differences across the visual field. To this end, we study the effect of dichoptic foveation, which we define as the application of blur to one eye and a simultaneous sharpening filter to the other. Stimuli were viewed in a virtual reality (VR) head-mounted display (HMD) and placed at different retinal eccentricities. Study data were scaled to a perceptual just objectionable difference (JOD) scale, and a 4D model was fit to it; our results suggest that interocular frequency differences can be well described by a simple computational model. We applied the model in a realistic VR scenario with free exploration of 360{\deg} videos to improve a base foveated rendering system by enhancing high frequency information dichoptically.

We present a real-time rendering method for scenes perceived through signals of finite speed $c$ in a non-relativistic setting: $c$ is a property of the imaging signal, not the causal speed limit, so bodies may move faster than $c$. A superluminal body presents several simultaneous images; pairs are created and annihilated in caustic flashes, and some branches play backward in time. Rather than approximating these effects, we enumerate them as roots of an emission condition posed against recorded state history, exact at every point solved, up to one stated approximation for moving observers. Because a fixed-step simulation records piecewise-linear history, the emission condition restricted to one history segment is a quadratic whose discriminant detects image-pair creation and whose slope classifies each image's playback direction, rate, and brightness. Perception is not tied to a single reference point: a global perceived state is built for an arbitrary finite set of observation events, and the delay landscape is the upper envelope of their backward light cones. A monotonic emission clamp guarantees the picture never regresses to images older than those already shown. Per-vertex solves shear bodies straddling delay gradients, and pre-generated frame-sequence assets are adapted by treating them as $(x,y,t)$ volumes sliced by the solved emission surface, so playback rate, reversal, and intra-body de-phasing arise with no animation-specific code. The exposition and implementations are planar: the emission condition and its per-segment solve are norm conditions, independent of dimension, but the visibility and occlusion questions a 3D renderer must answer are out of scope. The method is deployed in a released real-time strategy game; we describe the optimizations that make it run at interactive rates, and measure the algorithms through independent reference implementations.

Automatic facial rigging across heterogeneous mesh topologies remains challenging because high-quality expression supervision is often tied to canonical templates, while deformation transfer to arbitrary meshes can introduce geometric artifacts and correspondence errors. We present TopoRig, a topology-agnostic facial rigging framework that predicts FACS-conditioned deformations directly on input mesh vertices while preserving the original topology. Starting from the ICT FaceKit expression model, we construct complementary supervision from accurate but template-biased common-topology rigs, topology-diverse but noisier transferred rigs, and targeted image-based cues for controls poorly captured by geometric transfer. TopoRig combines local surface geometry, landmark-relative semantic features, global shape context, and FACS controls to predict per-vertex displacements. We train on 3,496 generated identities using 45 non-gaze expression controls from the 53-control ICT FaceKit vocabulary. On held-out identities and unseen mesh topologies, TopoRig more faithfully reproduces the reference expression space than prior neural facial-rigging methods, while qualitative results show consistent localized deformations across diverse character geometries. Ablations demonstrate that semantic landmark features and complementary supervision improve cross-identity and cross-topology generalization. Overall, TopoRig amortizes heterogeneous and imperfect expression supervision into a single topology-preserving deformation model.

We address the problem of color attribute compression for 3D splats. We show that all images generated by 3D splats are linear in the coefficients for each color channel, each spherical harmonic, and each splat, and we identify a basis for the space of all such images. We identify an inner product for the coefficient space that induces the squared error loss on images. We show that orthonormalizing the coefficients with respect to this innner product before coding can yield over 2 dB gain.

Native 3D generators turn one image into a single mesh. TRELLIS.2 and its peers deliver high-fidelity non-watertight geometry with materials, but the output is one fused object, while downstream work such as editing, rigging and simulation operates on part-level assets. A naive idea is to run a 3D segmentation network on the fused mesh that TRELLIS.2 generates, but such pipelines are slow and bounded by the accuracy of the segmentation. We want a simple way to extend an existing native 3D generator to the part level. But we face a critical problem: the O-Voxel grid stores one sheet of surface per voxel, so a single volume cannot represent the interface where two parts touch, at any resolution. We introduce a dual-volume representation to solve this problem and put forward KaiNinja, a part-level extension of TRELLIS.2 built on a dual-volume form of its O-Voxel representation. KaiNinja keeps the generation speed and quality of TRELLIS.2 while extending it to the part level, with no mask or segmenter in the pipeline. Its training data come from sources of many kinds, including CAD models and assets authored by an LLM-driven agent; to our knowledge it is the first 3D generative model trained on agent-authored part data. Surprisingly, we also find that whole-object fidelity improves over the same backbone fine-tuned on the same dataset. Against part generation pipelines of different paradigms, it lowers whole-object Chamfer distance by 40% and raises strict part F-score by 16%.

Deep learning models have become the standard tool for document rectification and illumination correction, yet their performance is fundamentally bound by their training data. For nearly a decade, the community has heavily relied on Doc3D, a pioneering but increasingly limited document unwarping dataset in terms of scale and quality. To address this bottleneck, we introduce SyntheticDoc, a massive, high-quality dataset designed to push the boundaries of document unwarping. SyntheticDoc is composed of 1,000,000 high-resolution procedurally generated training samples, alongside extensive validation and test sets. Each sample is paired with rich, pixel-perfect annotations, including UV maps, normal maps, albedo and shading. To ensure physical accuracy and photorealism, the paper geometries are generated via a physics-based simulator and rendered using a path tracer. To demonstrate the benefit of our dataset, we train a simple baseline model on SyntheticDoc and report on its performance in comparison to state-of-the-art methods on both document unwarping and illumination correction tasks. Our dataset is available at https://igl.ethz.ch/projects/SyntheticDoc/ and the code used to generate it at https://github.com/tanguymagne/SyntheticDoc .

Recent progress in image and 3D scene generation has enabled increasingly realistic static environments, yet most methods remain confined to such static configurations. Generating dynamic scenes from natural language is fundamentally challenging: it requires joint reasoning over scene structure, temporal evolution, and physical feasibility, while ensuring reliable execution in modern physics engines. We present a unified framework for generating physically consistent dynamic 3D scenes from text, with outputs directly executable in Unreal Engine. Central to our approach is the Evolutive Scene Graph (ESG), which specifies entities with physical attributes, spatial relations, and event-driven timelines in a machine-checkable form. Given a prompt, a large language model constructs and validates a complete ESG; spatial layouts are grounded via energy-minimized gradient optimization; timeline-constrained physical parameters are then optimized through differentiable simulation to satisfy user-specified events; and the resulting scene is compiled into an engine-executable class. Experiments on 10 scenes across three complexity levels show that our method achieves $16.4/18$ mean event completion, outperforming Scene Language, the strongest engine-executable baseline (SimWorld), and our ablation without physical optimization by a clear margin in event completion and parameter accuracy.

Web graphics travel with content across pages and themes, where changing backgrounds can require recoloring and maintenance. Opacity already makes a fixed object's appearance depend on its background, yet is usually understood only as how much the object obscures it. In fact, opacity controls the scaling of the object-background color difference; transparency is only one effect of this relationship. Zero places the output at the background and one at the source color, but difference scaling need not stop at either position. We retain the compositing expression and extend the coefficient domain from $[0,1]$ to the real numbers: negative values reverse the difference, whereas values above one expand it in the same direction. We focus on same-direction expansion for reusing Web graphics across backgrounds. Each object carries a fixed source color and coefficient, while the actual background determines the enhancement direction. Background-adaptive contrast enhancement thus becomes part of the object's compositing properties, reducing the design and maintenance of separate color variants. The implementation reuses the original equation without increasing the per-pixel arithmetic operation count within the same pipeline. Enumerating all 8-bit sRGB source colors on 16 predefined light and dark canvases, a fixed $\alpha=1.1$ increases the contrast ratio in 99.8145% of combinations. Without changing source colors, 4.8346% of all combinations newly reach the $3:1$ contrast threshold. Output validation and timing across three browsers demonstrate implementation in the same WebGL pipeline, with no sustained additional runtime observed.

Medical volume visualization requires selecting regions of interest (ROIs) and carefully controlling their relative visual emphasis according to a given clinical intent. Implementing these decisions in conventional workflows demands substantial clinical and visualization expertise and often involves trial-and-error optimization. Recent agentic systems have introduced natural-language interaction and autonomous visualization operations but largely rely on MLLM-based inference throughout the workflow. Although MLLMs encode broad medical knowledge and provide strong reasoning capabilities, such inference may be suboptimal for medical volume visualization, potentially leading to clinically incomplete interpretations of user requests and unreliable ROI identification and visualization optimization. In this work, we present MedVA, an end-to-end neuro-symbolic agentic system for medical volume visualization that addresses these limitations through three complementary agents. The neuro-symbolic intent formulation agent refines MLLM-based interpretations of natural-language requests through symbolic reasoning over established clinical knowledge, which provides more complete, clinically grounded ROI specifications than MLLM-only reasoning. The multi-model ROI identification agent directly identifies semantically specified ROIs in the original volume by leveraging complementary large-scale pretrained medical segmentation models. The objective-driven visualization optimization agent explicitly evaluates ROI visibility and occlusion in the original volume using a volume-based visibility objective. Extensive agent-level and system-level evaluations across diverse medical datasets and interaction scenarios support the effectiveness of the individual agents. A formative user study further indicates high usability and practical value among users with different levels of expertise.

Automatically generating high-quality conforming hexahedral (hex) meshes from general input boundaries remains a challenging problem, despite the numerical advantages hex elements offer in simulation. Grid-based adaptive refinement, followed by hanging-node removal, is the most robust fully automatic choice. Among two-refinement methods, primal templates have better mesh quality than dual templates but depend on a strong balancing condition that over-refines the grid; neither family guarantees planar quadrilateral (quad) faces. This paper generalizes recent primal three-refinement templates to a two-refinement scheme under a moderate balancing condition, a relaxation of the strong one. 32 special and five fundamental patterns resolve all 144 symmetry-reduced configurations, followed by either a fast greedy algorithm or an integer linear programming (ILP) algorithm that trades extra runtime for a further reduction in element count. The scheme is the first two-refinement method in which all hexes have planar faces, and its meshes feed directly into a recent quality-guaranteed hex mesh reconstruction algorithm. At comparable element counts, it attains a similar Hausdorff ratio (HR) and a higher minimum scaled Jacobian (min SJ) than its three-refinement counterpart.

We present a theory of light scattering that connects Gaussian Process Implicit Surfaces (GPISes) and participating media in both directions. Applying the Kac–Rice level-crossing formula under a local-conditioning approximation yields a complete anisotropic radiative transfer equation (RTE) directly from pointwise GPIS statistics. A shared projected area couples extinction and scattering, ensuring geometric consistency between the GPIS and its volumetric representation. The framework spans rough surfaces, porous and non-height-field geometries, and participating media. From the same statistical structure, we derive full-sphere Beckmann and GGX normal distribution functions supporting in-plane and out-of-plane anisotropy. These families provably recover SGGX, Beckmann, and GGX as special cases and admit exact visible-normal importance sampling. We also derive analytic masking–shadowing functions and single-scattering surface models for specular microsurfaces, with extensions to multiple scattering. In the height-field limit, we prove that the local-conditioning approximation reduces to Smith's independence assumption. Our realization-free approach improves rendering efficiency over realization-based methods and can be implemented within a standard volume renderer. In the inverse direction, we characterize families of GPISes corresponding to compatible RTE parameters and develop practical lifts for heterogeneous density fields. Existing volumetric assets thereby become renderable as GPISes, while trained radiance-field reconstructions yield surface geometry and shading normals without mesh extraction and provide a density-based representation of geometric uncertainty.

A chart may be declarative; authoring it is not. Visualization grammars often describe charts as finished specifications, whereas people construct them through a sequence of authoring actions. This mismatch can make visualization code difficult for humans to interpret and for machines to generate from human intent. ggaction addresses this gap by modeling the chart authoring process itself. In ggaction, individual authoring actions are abstracted as functions, and the authoring process is expressed as a chain of these functions. This representation more closely aligns chart designers' authoring intent with code specifications, making the code easily understandable to both humans and machines, including language models. Through a series of evaluations, we show that ggaction is sufficiently expressive to capture common chart authoring intents and outperforms widely used visualization grammars, including Vega-Lite and ggplot2, in both human and machine interpretability. ggaction is available at github.com/ggaction/ggaction.

Swept surface construction is widely used in computer-aided design. We propose a novel optimization method using non-uniform B-splines to improve the approximate accuracy of swept surfaces. First, discrete points on the swept shape are computed, and geometric properties such as surface area, discrete curvature, first-order derivatives, and their rotation angles are used to derive a distribution function representing surface irregularity, with weights adjusted from samples. Then, feature points are selected based on the distribution function to determine control points for the approximate non-uniform B-spline surface via inverse calculation, producing an optimized approximation. Finally, the number of feature points is adjusted based on the estimated approximation error. Experiments on 969 randomly generated sweep samples and 1 pipe example show that the proposed algorithm achieves similar accuracy with fewer control points, reducing them by about 15.85% at a specified accuracy of 0.01. Moreover, with ample sampling points, it reduces the average error by approximately 51.35% when the feature point multiple is 10 times the path control points, outperforming comparable methods.

Heat Kernel Textures (HKTex) represent surface appearance with intrinsic anisotropic kernels, but evaluate them using 50 global Laplace-Beltrami eigendecompositions and a resident basis of shape [50,V,256]. We study two complementary ways to remove this bottleneck while leaving the trainer, GeodesicOpt, density control, and compositing unchanged. LocalHK exploits the measured locality of trained kernels and replaces spectral evaluation by radius-bounded hinge unfolding and an analytic log-map kernel. On 10 Objaverse meshes and an 8-mesh low-poly holdout, it changes mean view PSNR from 31.35 to 32.00 and from 29.82 to 30.76 dB, respectively, while reducing initialization by 40.5 times and enabling a 749,570-vertex proxy-backed run where the spectral baseline fails. ThermalRF instead preserves the discrete anisotropic heat semigroup: GPU sparse Chebyshev actions and randomized range finding construct global low-rank heat factors without mesh-sized eigenvectors, and a compiled evaluator mixes four neighboring thermal responses. On spot and a thin-stem challenge, ThermalRF reduces end-to-end preprocessing, initialization, and 5,000-step optimization by 29.3% and 24.4%, with every surface, atlas, or view PSNR change within 0.12 dB and training allocation reduced by about 90%. The two routes expose a useful design choice: maximal locality and scale versus fidelity to the thermal PDE. Broader thermal-feature evaluation and real large scenes remain future work.

A contiguous cartogram is a map whose area is proportional to a quantity such as population. The defining condition, that the Jacobian determinant of the deformation equals the density, is one equation for two unknown functions, so every cartogram method adds a tie-breaker, usually implicitly. Optimal transport makes the tie-breaker explicit: among all density-equalising maps of the frame onto itself, take the one that moves the population least in the mean-square sense. By Brenier's theorem that map is the gradient of a convex potential, so it has no local rotation anywhere and cannot fold. We compute this map for the world population of 2025 on a 4096 by 4096 Mercator grid (10 km cells) from the GHS-POP raster, using the fixed-point Monge-Ampère iteration of Benamou, Froese and Oberman with spectral Poisson solves, continuation in the population share and an ocean-only buffer, on a laptop GPU. Against a Gastner-Newman diffusion cartogram of the same density we find the same transport cost to within 0.4 per cent and density errors of a few per cent for both, but a median local rotation of 8.8 degrees for diffusion against 0.01 for transport, and a median anisotropy of 6.30 against 3.93. The construction extends to local refinement, by transporting a city's 100 m population onto the area measure the global map assigns to it, and to a semi-discrete counterpart: 8,192 Laguerre cells of 1.00 million people each. Code, data provenance and every figure's inputs are public.

We study how to update a pre-built aerial scene with a newly captured, unposed ground-view sequence. The aerial scene already contains a reliable metric Structure-from-Motion (SfM) reconstruction and a pre-trained 3D Gaussian Splatting (3DGS) model, whereas the ground-view sequence is collected later to add street-level appearance but has unknown camera poses and global scale. Registering this sequence to the aerial SfM reconstruction is challenging because single-image cross-view localization is brittle and long trajectories are prone to drift. To address these challenges, we present SkyAnchor, which treats the existing aerial scene as a fixed scaffold for ground-view registration and scene update instead of jointly reconstructing aerial and ground imagery from scratch. It first localizes short groups of consecutive ground frames against geometrically verified aerial support, producing sparse anchor poses. It then recovers the full ground trajectory with anchor-constrained submaps, fixing the front and rear anchor poses during incremental registration and bundle adjustment. Finally, it inserts filtered ground Gaussians while preserving the aerial view, followed by lightweight joint refinement. Experiments on seven real aerial–ground scenes show accurate metric ground trajectories and updated 3D Gaussian scenes with strong aerial- and ground-view rendering quality.

Deploying reconstructed 3D environments on standalone VR headsets are constrained by limited compute and memory, and conventional level-of-detail policies optimize for visibility without accounting for the user's explicit inspection intent. We present GaussAnything, a native OpenXR system for intent-conditioned reallocation and progressive publication of evolving semantic Gaussian+SDF scenes. GaussAnything resolves class- or instance-level queries to persistent 3D objects and reallocates a fixed Gaussian resident budget toward the selected object while retaining global context, applying incremental, stable-identity updates coordinated with the TSDF-derived mesh through a source-epoch mechanism. Across eight scenes, an object query concentrates 88-90% of the fixed client budget onto the queried object without enlarging it, on-device rendering reproduces the host render to within a small margin (up to 36.7 dB), and the standalone client renders each stereo frame at a steady-state GPU cost of roughly 10 ms within the frame budget of standard standalone panels.

We present the Variational Incompressible Optimal Transport (VIOT) operator, a generative neural operator for amortized incompressible density transport. Given a new source-target density pair, VIOT predicts a divergence-free velocity field and generates the full transport trajectory by feed-forward inference, replacing the hour-scale per-pair optimization used by adjoint fluid solvers and differentiable simulation baselines. The system consists of three components: a stream-function or vector-potential representation that enforces incompressibility by construction, a regularized incompressible transport objective that balances endpoint accuracy and flow smoothness, and a Fourier Neural Operator backbone that amortizes the solve across new pairs and grid resolutions. Together, these components make incompressible transport a reusable neural operator that facilitates various transport processes. Further, the generative capability extends beyond the training distribution, with VIOT producing incompressible transports for user-drawn source-target pairs in a real-time interactive system. We demonstrate VIOT on 2D and 3D density-transport benchmarks. Both 2D and 3D rollouts complete in seconds per pair, while per-instance baselines in our 2D comparisons optimize each new pair from scratch and require on the order of an hour, a roughly $10^4\times$ online speedup.

The mechanical memory of fabrics often leads to persistent wrinkles, which reflect key physical properties and habitual wear patterns. Simulating these wrinkles accurately is essential for visual plausibility in digital garments, yet no dedicated approach exists for inferring the parameters that govern their formation due to the lack of precise datasets and estimation methods. We introduce Fabric-101, an inclusive, accurate, and extendable fabric dataset comprising over 101 common fabrics following textile standards. Unlike existing datasets, it captures three physically distinct deformation components (i.e., self-recoverable (elastic), recoverable (friction-driven), and unrecoverable (plastic)), from cyclic loading-unloading measurements. Building on this data, we propose a differentiable cloth simulator combining an elasto-plastic model with friction, designed to capture recoverable and unrecoverable wrinkle formation. Our simulator is differentiable and uses adjoint method to learn fabric physical parameters from the measured hysteresis curves, learning fabric-specific wrinkle behaviors. Through extensive experiments, we demonstrate that our model reproduces persistent wrinkles that are visually and physically similar to real fabrics across diverse materials and motions. Dataset and code are available in https://github.com/GongDeshan/Fabric_101_for_Wrinkles.

Given a grid-sampled 1-Lipschitz function $f:\mathbb{R}^3 \rightarrow \mathbb{R}$ (such as a signed distance function) sampled on a regular grid, we determine the sharp infimum $\sigma_\star$ such that for every $\sigma > \sigma_\star$ the true zero-level set $f^{-1}(0)$ is contained in the marching cubes mesh for $f = \sigma$. For grids with spacing $h$, the sharp infimum is $\sigma_\star = \frac{\sqrt{3}}{2} h$. This result is proven despite the planar marching cube faces potentially lying on either side of the level set of the trilinearly interpolated approximation of $f$ (which shares the same infimum). The proof instead constructs a convex lift of the corner samples and shows that the high-side portion of each marching-cubes cell lies in the projection of its $\sigma$-superlevel portion, which cannot contain a point of $f^{-1}(0)$.

Part-aware 3D asset generation enables applications such as editing, articulation, simulation, and fabrication, yet existing methods can generate visually complete individual parts without ensuring that they form a valid physical assembly. Consequently, generated neighboring parts may interpenetrate, lack valid connections, or collapse under gravity. We propose a physics-guided framework for improving single-image part-aware 3D generation with physically compatible geometry and stable connections. Our method resolves inter-part penetration, recovers a contact graph between neighboring parts, and introduces parameterized connectors at their contact surfaces. Using feedback from physical simulation, we refine connector placement, orientation, and dimensions to improve assembly stability while preserving the generated geometry. We further introduce a physics-based evaluation protocol that complements conventional geometric metrics by directly testing assembly validity and stability under gravity. Experiments comparing against multiple part-aware 3D generators show substantial improvements in physical realizability and stability while maintaining geometric quality. We additionally validate the resulting parts through 3D printing and real-world assembly.

We present an approach to progressively extend the highlights of an image. Instead of reconstructing the full dynamic range of a complex scene directly, we learn a simpler task first: We extend the dynamic range of an input image by a single exposure value. Once this is mastered, we retrieve the full HDR image for the scene by executing our network recurrently, progressively increasing the dynamic range of the input. Our formulation is agnostic to the input dynamic range and targets a bounded output domain. This enables us to use widely available RAW images for the reconstruction task and adapt adversarial losses to construct realistic images. By incorporating Memory Replay for backpropagation, we can train our network recurrently over multiple inference stages and reduce reconstruction errors. As a consequence, our system reconstructs challenging long-tailed HDR scenes robustly and shows powerful recovery of bright light sources and highlights.

The conditional generation of 3D motion has emerged as a key research topic due to its wide applicability across robotics, AR/VR, gaming, and content creation. However, extending recent advances in text-driven human motion generation to the animal domain remains challenging due to two core limitations. First, animals exhibit highly diverse skeletal topologies, unlike the standard human structure, making unified modeling across species difficult and leading to inefficient per-species models. Second, existing animal motion datasets suffer from limited scale and annotation quality, constraining model performance. To address these challenges, we propose UniMo, a unified point cloud-based motion generation framework that bypasses topological discrepancies by converting parametric skeletons into unparametric representations, further enhanced by dynamic sampling that allocates more points to active joints. Additionally, we present UniML3D, a large-scale motion-language dataset spanning both human and animal categories, containing 145,907 motion sequences and 433,388 captions-over 102x larger than existing animal datasets. Our method achieves state-of-the-art results on UniML3D and three public benchmarks including HumanML3D, KIT-ML, and AnimalML3D, demonstrating the feasibility and effectiveness of unified human-animal motion generation. Website: https://steve-zeyu-zhang.github.io/UniMo.

Many scientific applications require modeling how diffusive systems evolve over time, not merely their eventual steady states. While conventional steady-state analysis of partial differential equations (PDEs) on complex geometries is already hindered by costly volumetric meshing, transient analysis further requires sequential time stepping and careful step size selection. Grid-free Monte Carlo solvers such as walk on spheres (WoS) and walk on stars (WoSt) avoid this meshing bottleneck but remain largely limited to steady-state problems. We generalize WoS, for pure Dirichlet problems, and WoSt, for mixed Dirichlet–Neumann problems, to heat equations with initial conditions and time-dependent source and boundary data. We equip each random walk with a finite time budget and sample an exit time at every spatial step. If the exit time exceeds the remaining budget, the walk samples an interior point and evaluates the initial condition; otherwise, it continues with a reduced budget, accumulating source and boundary contributions. Our main technical contribution is a suite of kernel sampling and variance reduction techniques, including a low-bias, tabulation-free exit time sampler and efficient rejection samplers. Unlike grid-based transient solvers, our method directly estimates the solution at any requested time without volumetric meshing or sequential time marching. It also retains the parallel, progressive, and output-sensitive evaluation of WoS and WoSt while eliminating time step selection and temporal discretization bias entirely. Finally, we show how sharing walks enables efficient estimates at multiple target times.

In scientific simulation, regular grids, unstructured meshes, and particle-based formats are chosen to represent field data for computational efficiency, geometry/adaptive flexibility, and following motion/deformation, respectively. Each of these field data formats is often handled through separate data-specific processing pipelines. We present a unified sample-based Gaussian encoding method that represents these data forms under a single fixed-budget formulation. The method initializes and refines Gaussian primitives directly from the input samples while preserving a prescribed primitive count and encoded size to achieve a desired level of data reduction. Across structured, unstructured, and particle data, the sample-based formulation improves reconstruction accuracy with measurably fewer primitives in comparison to prior formulations, achieving up to 4.8 dB higher PSNR with an approximate 44x reduction in primitive count. For time-varying data, warm-starting from the previous timestep reduces the optimization required to reach independently trained reconstruction quality. Together, these results demonstrate a unified fixed-budget Gaussian encoding framework for structured, particle, unstructured, and time-varying scientific data with predictable storage, higher reconstruction accuracy, and improved temporal encoding efficiency.

Solving novel view synthesis (NVS) for millimeter-wave (mmWave) radar requires a renderer that is physically faithful, complex-valued, and multi-viewpoint-tractable. No prior method achieves these three properties simultaneously. Differentiable Monte Carlo (MC) ray tracers implement the radar forward model directly with explicit material modeling and complex outputs, but do not scale to the multi-view optimization NVS demands. Optical-NVS ports of NeRF, hash grids, and 3D Gaussians train fast but discard phase and replace explicit material modeling with opaque learned features, restricting them to power-only range-azimuth (RA) magnitudes. We propose 3D Point Splatting (3DPS), the first differentiable point renderer for radar, derived directly from the standard solid-angle form of the radar equation. Each oriented 3D point carries an ITU-R P.2040 material model, evaluated in closed form, with the resulting complex phasor splatted into range bins through a precomputed point spread function (PSF). The complex-valued output makes the renderer product-agnostic. The same optimized scene yields analog-to-digital converter (ADC), complex range profile (CRP), and RA outputs through standard fast Fourier transform (FFT) pipelines without retraining for each format. On six outdoor ColoRadar scenes, 3DPS reaches 0.587 mean Pearson correlation on held-out RA images. This is between 1.7x and 5.2x the three optical-NVS baselines (RadarSplat, Radar Fields, DART). Training takes approximately 3 minutes per scene on a single RTX 4090.

Muscle-driven locomotion provides a physically grounded approach to generating realistic human movement. However, achieving both physiological plausibility and adaptability to changes in musculoskeletal capacity and external disturbances remains a fundamental challenge. To address this limitation, we propose a Reflex-Informed Neuromuscular Reinforcement Learning framework for muscle-driven locomotion. Within this framework, a fixed phase-dependent reflex controller serves as the underlying neuromuscular control mechanism, while the reinforcement learning policy produces four biomechanically meaningful residual parameters to modulate key reflex gains and thresholds associated with hip swing, knee support, and ankle propulsion according to the current state. Experimental results demonstrate that the proposed framework generates physiologically plausible locomotion with improved kinematic accuracy and dynamic consistency, as well as better bilateral symmetry and stride-to-stride consistency under nominal walking conditions. The learned policy remains robust under muscle weakness and external perturbations without retraining.