arXiv:2602.24043v2 Announce Type: replace Abstract: Reconstructing 3D clothed humans from monocular images and videos is a fundamental problem with applications in virtual try-on, avatar creation, an...
OmniFabric is a new method for creating high‑quality, globally coherent texture maps for 3D garment reconstruction from a single image. It first generates a coarse texture initialization on the garment’s sewing pattern using a 3D mesh and Vision‑Language Model priors, then refines this in the UV domain with a diffusion transformer conditioned on 3D positional features. The approach removes distortion and baked‑in artifacts, producing photorealistic 3D garments that outperform existing baselines.
The paper introduces HyperBones, a real‑time garment simulation framework that combines a reduced‑space neural dynamics simulator with a lightweight neural network correcting Linear Blend Skinning (LBS) at a coarse level, and a convolutional MLP for fine‑scale wrinkle recovery in UV space. By decoupling identity‑specific computation from shape conditioning through a hypernetwork, the method achieves high performance without an offline simulator, delivering physically plausible dynamics across diverse motions and unseen body shapes. Experiments demonstrate a speedup of over 30× compared to state‑of‑the‑art autoregressive neural simulators, reaching interactive inference at roughly 1 ms per frame on a consumer GPU.
arXiv:2607. 23189v1 Announce Type: cross Abstract: AI-generated content (AIGC) has made significant progress, with 2D generative models becoming ready-to-use tools for the digital fashion industry.
OmniFabric is a new method for creating production‑ready 3D garment assets from a single image. It generates globally coherent texture maps directly in the 2D sewing pattern (UV) space, using a coarse initialization from Vision‑Language Models and refining it with a diffusion transformer conditioned on 3D positional features. The approach removes distortion and baked‑in artifacts, producing photorealistic 3D garments with high‑quality textures that outperform current state‑of‑the‑art baselines.
arXiv:2511. 18765v3 Announce Type: replace-cross Abstract: Existing industrial 3D garment meshes already cover most real-world clothing geometries, yet their texture diversity remains limited.
arXiv:2501. 13692v2 Announce Type: replace-cross Abstract: Diffusion models have recently unlocked new possibilities in editing images of real-world objects.
The paper introduces AvaImg, a multi‑stage optimization pipeline that achieves high‑fidelity SMPL(-X)+D registrations with UV texture for arbitrary clothed scans. By enforcing a body‑inside‑clothing constraint through signed winding numbers and employing a three‑level efficiency cascade, AvaImg significantly reduces runtime and storage while recovering fine surface detail via coarse‑to‑fine displacement optimization. The resulting textured registrations are nearly indistinguishable from scans, and encoding the UV maps with a frozen FLUX VAE demonstrates compatibility with 2D generative models, enabling 3D avatar generation using image‑based priors.
Creating photorealistic and temporally coherent animatable human avatars from RGB videos remains challenging. Current methods struggle to capture realistic cloth dynamics, producing over-smoothed appearance or severe artifacts on out-of-distribution poses.
FitControler introduces a fit-aware virtual try‑on system that adds garment fit control to existing VTON models. It uses a fit‑aware layout generator and a multi‑scale fit injector to redraw body‑garment layouts and render garments that match those layouts. The authors also release a new Fit4Men dataset of 13,000 body‑garment pairs and two fit consistency metrics to evaluate fit quality.
arXiv:2607. 10984v1 Announce Type: cross Abstract: Existing Stochastic 3D Human Motion Prediction models are fundamentally constrained by hard-coding the skeleton kinematics, severely limiting generalization, preventing cross-dataset training, and requiring complex data retargeting.
arXiv:2606. 02000v1 Announce Type: cross Abstract: Diffusion models have shown remarkable success in video generation.