Digital Humans News
The latest Digital Humans coverage curated for game developers.
arXiv cs.GR details Heartian, a Gaussian head-avatar system that bakes cardiac-cycle skin-color changes into relightable facial assets. The method uses synchronized contact PPG supervision to recover heart-rate signals while keeping …
arXiv cs.GR details DEAL-Grasp, a dexterous grasp generator that separates rigid alignment from finger articulation. The method uses alignment-space generation and flow matching to produce more stable, physically plausible hand-object …
A new co-speech gesture system aims to make digital humans move more naturally while cutting inference from dozens of steps to a faster flow-matching pipeline. GestureLSM models body-region interactions with …
A new strand-based hairstyle pipeline aims to turn a single image into production-ready hair geometry in minutes. It combines large reconstruction models, multimodal models, and classical geometry processing to handle …
A new multi-view avatar pipeline combines separate face and hand modeling with a parametric upper-body mesh, then renders the result with dynamic 3D Gaussians. The goal is to preserve subtle …
URHead proposes a UV-space bridge between mesh control and 3D Gaussian Splatting for animatable head avatars. The system jointly optimizes both representations so geometry stays controllable while fine facial detail …
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 state-of-the-art survey maps the current pipeline for controllable 3D digital humans, from learned priors to personalized avatars. It breaks the space into three stages—learning appearance and motion, building …
UE 5.8 brings another round of MetaHuman updates, which matters most if your pipeline depends on high-fidelity digital humans. Epic says the release is available now through the Epic Games …
A new arXiv paper tackles single-image human material estimation, a hard problem for relighting and digital human workflows because lighting, shape, and reflectance are tangled together. HAFMat uses hybrid priors …
A new Unreal Engine pipeline aims to make speech-driven facial animation usable in production, not just in papers. The authors convert MEAD into an ARKit-style blendshape dataset, retrain FaceDiffuser and …
A new multi-view capture pipeline can turn facial hair into editable strand assets instead of baking it into face geometry. For teams shipping avatars, that matters because beards, brows, lashes, …
Disney Research’s new head-relighting model removes a big capture bottleneck: you no longer need OLAT footage for every subject. It can learn from ordinary single- or multi-view images, then generalize …
A paper from arXiv claims it can train high-fidelity 3D Gaussian head avatars from random 2D images alone, skipping multi-view capture, 3D supervision, and intermediate view synthesis. For teams building …
For teams working on avatar rendering, this paper pushes Gaussian avatars beyond surface-bound templates by moving the representation into a volumetric canonical space. The practical upside is better handling of …
A new approach to facial appearance capture leverages a delighting network, significantly improving quality from casual video inputs. This innovation allows developers to create photorealistic digital humans without the need …
A small team has successfully integrated Unreal Engine and MetaHuman to create stylized digital humans for their project, Daughter of the Inner Stars. This innovative approach combines real-time production with …