Skip to main content
GameDev.net gamedev.net

Motion Generation News

The latest Motion Generation coverage curated for game developers.

UniMo: Unifying Human and Animal Motion Generation

A new motion-generation system aims to cover both humans and animals with one model, sidestepping the topology mismatch that usually forces separate pipelines. UniMo pairs a point-cloud representation with dynamic …

Research arXiv cs.GR · 1 week, 2 days ago
Per-Stroke Temporal Control for Text-to-Motion via Action Units and Action-Detection Guidance

A new text-to-motion method adds per-stroke timing control, aiming to make prompted actions land on the right frames instead of just the right motion. It introduces Action Units that encode …

Research arXiv cs.GR · 2 months ago
Learn2Chat: Rethinking Dyadic Talking Heads via Interaction-Modulated Monologic Priors

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 …

Research arXiv cs.GR · 2 months, 1 week ago
ARDY: Autoregressive Diffusion with Hybrid Representation for Interactive Human Motion Generation

ARDY is a new streaming motion-generation system that aims to make real-time humanoid animation both controllable and high quality. It combines autoregressive diffusion with a hybrid motion representation, letting interactive …

Research arXiv cs.GR · 2 months, 1 week ago
MoGeFlow: Flowing Through Motion Codebook Geometry for Text-to-Motion Generation

A new text-to-motion paper argues that motion codebooks are not just arbitrary token IDs: their learned embeddings preserve local kinematic geometry. That matters for animation pipelines because it replaces flat …

Research arXiv cs.GR · 3 months, 1 week ago
AnimaSpark: A Feed-Forward Method for Animating Arbitrary 3D Objects

A new paper claims feed-forward 3D animation for arbitrary objects, which is interesting because it tries to remove a lot of the manual rigging and keyframing pain. The pipeline renders …

Research arXiv cs.GR · 3 months, 1 week ago
EnchantDance: Unveiling the Potential of Music-Driven Dance Movement

A new music-driven dance generation paper claims state-of-the-art results by moving the problem into a latent space and conditioning on music genre. The practical takeaway for animation and tools teams …

Research arXiv cs.GR · 3 months, 2 weeks ago
EMA: Effort Metric Attention for Anatomical Effort-Guided Human Motion Diffusion

A diffusion-based motion system now takes numeric effort signals instead of vague adverbs, which is a big deal for anyone trying to direct character intensity more precisely. The paper maps …

Research arXiv cs.GR · 3 months, 4 weeks ago
MotionDuet: Dual-Conditioned 3D Human Motion Generation with Video-Regularized Text Learning

MotionDuet tries to make 3D human motion generation less “text prompt only” and more grounded in real motion statistics by adding video cues during training and inference. For animation-heavy game …

Research arXiv cs.GR · 4 months ago
Stylized Text-to-Motion Generation via Hypernetwork-Driven Low-Rank Adaptation

A new lightweight framework for stylized text-to-motion generation is set to enhance realism in animations. By utilizing hypernetwork-driven low-rank adaptation, developers can achieve nuanced motion styles without extensive model fine-tuning, …

Research arXiv cs.GR · 4 months, 1 week ago
Stability-Driven Motion Generation for Object-Guided Human-Human Co-Manipulation

The introduction of a flow-matching framework for co-manipulation in human interactions is a significant advancement for developers focused on realistic motion generation. By addressing the challenges of synchronizing multiple users …

Research arXiv cs.GR · 5 months ago
A Unified Conditional Flow for Motion Generation, Editing, and Intra-Structural Retargeting

A single rectified-flow model can now handle text-to-motion generation, semantic editing, and same-topology retargeting in one pass. The system conditions on both text and skeleton, then uses joint-level tokenization and …

Research arXiv cs.GR · 5 months, 1 week ago
Pulp Motion: Framing-aware multimodal camera and human motion generation

The introduction of Pulp Motion marks a significant advancement in generating coherent human motion and camera trajectories. By integrating these two modalities through a shared latent space, developers can achieve …

Research arXiv cs.GR · 5 months, 3 weeks ago
ProAct: Harnessing Streaming Motion Generation and Agentic Reasoning for Real-Time Embodied Social Interaction

A new humanoid-robot framework, ProAct, splits real-time social interaction into two systems: one streams fluent motion and nonverbal behavior, while the other reasons about dialogue and visual context to decide …

Research arXiv cs.GR · 7 months ago