Character Animation News
The latest Character Animation coverage curated for game developers.
arXiv cs.GR details VISTA, a two-stage text-to-animation system that blends prompt-driven motion structure with style pulled from reference video. For teams building character animation tools, the interesting bit is controllable …
arXiv cs.GR details TopoRig, a facial rigging system that keeps a character’s original mesh topology while predicting FACS-driven deformations. For teams building custom characters, the pitch is fewer transfer artifacts …
ReCHOIR targets a stubborn retargeting problem: preserving object interaction when the same motion lands on different humanoid rigs. The method uses contact cues and a part-aware latent space to keep …
InstantMimic pushes physics-based character training into the seconds range by moving the entire RL loop onto the GPU. The system targets imitation-driven motion control, cutting out CPU bottlenecks and fragmented …
STyMo can learn a motion style from only seconds of paired data and train in one to two minutes, making style transfer far more practical for character animation. The system …
A new CVAE-based pipeline can generate controllable facial expressions for virtual humans from just 7,680 samples. Trained on six basic emotions at low and high intensity, it preserves recognizable emotional …
A new musculoskeletal RL controller can produce human-like sprinting with only a minimal task reward and about an hour of training. The ?bb-hold approach cuts the action space down to …
A new skinning prior called cut-cell skinning aims to make neural weight generation faster and more robust for production meshes. It approximates volumetric geodesic distances with a graph-based method, avoiding …
SubtleTalk tackles a long-standing problem in audio-driven 3D talking heads: believable motion above the mouth. The system adds controllable eyebrow, blink, head, and affect cues on top of lip sync …
ETHead is a new speech-to-animation method aimed at making 3D talking heads look less robotic and more emotionally alive. It uses self-distillation from large-scale 2D talking videos to train a …
StreamTalk tackles a familiar real-time animation problem: gesture drift in streaming co-speech generation. The system adds a key-pose anchor at the end of each clip, then retrieves and refines motion …
Researchers have introduced DynHair, a head-avatar system that tracks and animates explicit strand-based hair from video. It combines structured 3D Gaussian splatting with a temporal model that uses head motion, …
Learn2Chat tackles dyadic talking-head animation by separating speech-driven motion from interaction-driven behavior. The system reuses pretrained monologic motion priors, then modulates them with learned interaction latents to produce synchronized two-person …
A new hand-motion completion system uses a learned body-hand prior first, then adds control signals on top, aiming to make real-time animation both plausible and steerable. It targets the hard …
A new avatar simulation approach tries to solve a long-standing problem: making human characters physically interact without collapsing their pose fidelity. PIAvatar decouples kinematic velocity from deformation gradient, then runs …
A new paper proposes a practical way to reuse existing facial animation models for micro-expressions by magnifying tiny motions into a regular-expression range first, then scaling them back down after …
A fully automatic rigging pipeline now turns a static character mesh into an inner-mouth-aware facial rig, including teeth, gums, tongue, and up to 155 FACS blendshapes. For teams that still …
RigPAPR tackles a common failure mode in neural character animation: joint-boundary artifacts when Gaussian splats are driven by rigid skinning. The paper claims a static PAPR point cloud can be …
For teams building animation or embodied-agent systems, the interesting bit here is a control adapter that keeps a frozen text-to-motion transformer intact while adding trajectory guidance through attention memory. KV-Control …
A new motion-encoding paper tackles a real pain point for animation pipelines: skeletal topology changes across humans, animals, and other creatures. The authors say their semantic-aware approach learns a shared …