Research News
The latest Research coverage curated for game developers.
arXiv cs.GR details CuACD, a fully GPU-resident approximate convex decomposition pipeline that turns triangle meshes into convex parts much faster. The authors say it can cut preprocessing from tens of …
arXiv cs.GR details OREO, a 3D generation alignment method that uses on-the-fly edited renderings to push models toward more realistic assets. The system feeds refined 2D views back as pseudo-targets, …
SIGGRAPH has unveiled the keynote lineup for SIGGRAPH Asia 2026, putting Markus Gross, Ran Huang, and Andrea Vedaldi on stage. For graphics programmers and technical artists, that usually signals a …
SIGGRAPH says Don Greenberg, 92, headlined the Pioneers featured talk in a hybrid session at SIGGRAPH 2026, putting a long-view spotlight on one of computer graphics' early figures. For developers, …
arXiv cs.GR details VideoX-Qwen, a data-centric video editing pipeline that pairs instruction data with a unified editor. The system builds 1.2 million directional edit records and reports strong results against …
arXiv cs.GR details MoSAT, a motion-synthesis model that uses spatial audio plus text to drive full-body animation. The team pairs it with STAM, a new dataset of motion sequences, directional …
arXiv cs.GR details a new diffuse-like BRDF that evaluates and importance-samples in closed form while accounting for all scattering orders under Smith shadowing. For rendering teams, the practical draw is …
arXiv cs.GR details LYRIC, a physics-based character controller that turns free-form language into contact-rich whole-body interactions. For game teams, the interesting part is the split between a task planner and …
arXiv cs.GR details PART, a transformer-based system that can retrieve 3D parts from a library and assemble them into a target shape. The approach tackles variable part counts and 6-DoF …
arXiv cs.GR details PointGrade, a machine-learning method that grades MoonBoard bouldering problems by sampling point clouds from pre-scanned hold meshes. By folding geometric priors into a sequence-based predictor, the model …
arXiv cs.GR details DSD, a diffusion-based skill discovery method that learns a wider set of reusable humanoid motor skills. For game teams building animation or AI-driven characters, the pitch is …
Researchers have pushed non-line-of-sight imaging beyond the usual three-bounce assumption, using higher-order light paths to reconstruct hidden objects around one or even two corners. The technique combines ultrafast laser scanning …
A new real-time rendering method handles superluminal motion by solving for delayed-light images directly from recorded simulation history. It can produce multiple simultaneous views, pair creation and annihilation flashes, and …
arXiv cs.GR details PhysStream, a video-generation model that lets creators steer physics-driven motion mid-generation instead of locking in a full control schedule up front. For game teams, the interesting bit …
A new pipeline turns text prompts into physically consistent dynamic 3D scenes that can run in Unreal Engine. The system builds an Evolutive Scene Graph, then optimizes layout and motion …
A new HKTex acceleration method drops the mesh eigensystem bottleneck that has made heat-kernel texture workflows expensive to initialize. Two variants, LocalHK and ThermalRF, keep the trainer and compositing pipeline …
A new SIGGRAPH-era rendering technique brings physically accurate defocus to Gaussian splatting through closed-form convolution. For teams using splats in real-time pipelines, that means cleaner depth-of-field behavior without relying on …
ATOP is a new 3D part-articulation method that lets creators drive object motion from text and personalize it with motion examples. For game teams, that points to faster iteration on …
Researchers have introduced DMP, a motion-retargeting method that lets animators steer character movement more directly while preserving the source performance. The technique uses motion paraphrasing to adapt animation across different …
UMA is a new human-avatar pipeline aimed at producing ultra-detailed digital people through multi-level surface alignment. Published in ACM Transactions on Graphics in February 2027, it targets higher-fidelity body representation …