HumanOrbit: 3D Human Reconstruction as 360° Orbit Generation

Keito Suzuki; Kunyao Chen; Lei Wang; Bang Du; Runfa Blark Li; Peng Liu; Ning Bi; Truong Nguyen

cs.CVFeb 27, 2026

HumanOrbit: 3D Human Reconstruction as 360° Orbit Generation

Authors: Keito Suzuki, Kunyao Chen, Lei Wang, Bang Du, Runfa Blark Li, Peng Liu, Ning Bi, Truong Nguyen

Organizations: University of California, San Diego · Qualcomm

Abstract

We present a method for generating a full 360° orbit video around a person from a single input image. Existing methods typically adapt image-based diffusion models for multi-view synthesis, but yield inconsistent results across views and with the original identity. In contrast, recent video diffusion models have demonstrated their ability in generating photorealistic results that align well with the given prompts. Inspired by these results, we propose HumanOrbit, a video diffusion model for multi-view human image generation. Our approach enables the model to synthesize continuous camera rotations around the subject, producing geometrically consistent novel views while preserving the appearance and identity of the person. Using the generated multi-view frames, we further propose a reconstruction pipeline that recovers a textured mesh of the subject. Experimental results validate the effectiveness of HumanOrbit for multi-view image generation and that the reconstructed 3D models exhibit superior completeness and fidelity compared to those from state-of-the-art baselines.

Figures & tables

Figure 1: Given an in-the-wild image, we treat the ill-posed problem of multi-view synthesis as orbit video generation for robust 3D human reconstruction. Our model creates identity preserving and view consistent frames, resulting in a high quality textured mesh.

Figure 2: The proposed HumanOrbit model. We finetune a DiT-based video diffusion model such that it directly generates a 360° orbit video given a single input image. While keeping much of the architecture frozen, the finetuned LoRAs learn to accurately conduct an orbit around the subject, generating consistent multi-view images.

Figure 3: The proposed mesh reconstruction framework. Given the generated multi-view images, we first apply a SfM method to obtain the point cloud and camera parameters for each view. We then estimate the normal maps for each frame. Finally, using the previously generated results, the textured mesh is reconstructed via an explicit mesh carving method.

Figure 4: Two examples of the camera pose and point cloud predicted by VGGT on our generated multi-view images. The original images are shown in Figure 1 .

Dataset	Method	CLIP Score [ 30 ] $\uparrow$	MEt3R [ 1 ] $\downarrow$	MVReward [ 44 ] $\uparrow$
Full body [ 51 ]	SV3D	0.7888	0.2966	0.2378
	MV-Adapter	0.7735	0.3721	0.6795
	PSHuman	0.8282	0.3576	0.6814
	Ours	0.8317	0.3175	0.8035
Head [ 17 ]	SV3D	0.6582	0.4745	0.4918
	MV-Adapter	0.6729	0.4826	0.4727

Table 1: Quantitative comparison of multi-view image generation results on in-the-wild human images. For full body and head image evaluation we use the CCP [ 51 ] and CelebA [ 17 ] dataset respectively. Results in bold show the best value.

Figure 5: Visual comparison of multi-view generation results on full body images from the CCP dataset [ 51 ] . Given the input image, we show three generated views (left, back, and right) from each method. Zoom in for more detail.

Figure 6: Visual comparison of multi-view generation results on head portrait photos from the CelebA dataset [ 17 ] . Given the input image, we show three generated views (front left, back, and front right) from each method.

Figure 7: Visual comparison of the appearance and geometry of 3D mesh reconstruction methods on body images.

Figure 8: Visual comparison of the appearance and geometry of 3D mesh reconstruction methods on a head portrait.

Figure 9: Comparison of 3D reconstruction when using COLMAP and VGGT for camera pose estimation.

Figure 10: Novel views generated by our model on non-human images such as a chair and a dog.

Multi-View Image Synthesis

See all 24 papers

PSHuman: Photorealistic Single-image 3D Human Reconstruction using Cross-Scale Multiview Diffusion and Explicit Remeshing
Sep 16, 2024Peng Li, Wangguandong Zheng, Yuan Liu +9Multi-View Diffusion Models 3D Reconstruction
HKUST · Southeast University · Tsinghua University +1
OrbitForge: Text-to-3D Scene Generation via Reconstruction-Anchored Video Synthesis
Jun 23, 2026Chenrui Fan, Paolo Favaro3D Gaussian Splatting Multi-View 3D Reconstruction
Computer Vision Group, Institute of Computer Science University of Bern, Switzerland
Human Interaction-Aware 3D Reconstruction from a Single Image
Apr 7, 2026Gwanghyun Kim, Junghun James Kim, Suh Yoon Jeon +23D Human Reconstruction Single-View 3D Reconstruction
Dept. of Electrical and Computer Engineering, Seoul National University, Republic of Korea · IPAI, Seoul National University, Republic of Korea · INMC, Seoul National University, Republic of Korea

Sep 16, 2024J/K move · Enter open · S save

PSHuman: Photorealistic Single-image 3D Human Reconstruction using Cross-Scale Multiview Diffusion and Explicit Remeshing

Peng Li, Wangguandong Zheng, Yuan Liu, Tao Yu+8

HKUST · Southeast University · Tsinghua University+1

Multi-View Diffusion Models 3D Reconstruction 3D Mesh Reconstruction 3D Human Reconstruction Single-View 3D Reconstruction Diffusion Models

Open paper AskPDF arXiv

Detailed and photorealistic 3D human modeling is essential for various applications and has seen tremendous progress. However, full-body reconstruction from a monocular RGB image remains challenging due to the ill-posed nature of the problem and sophisticated clothing topology with self-occlusions. In this paper, we propose PSHuman, a novel framework that explicitly reconstructs human meshes utilizing priors from the multiview diffusion model. It is found that directly applying multiview diffusion on single-view human images leads to severe geometric distortions, especially on generated faces. To address it, we propose a cross-scale diffusion that models the joint probability distribution of global full-body shape and local facial characteristics, enabling detailed and identity-preserved novel-view generation without any geometric distortion. Moreover, to enhance cross-view body shape consistency of varied human poses, we condition the generative model on parametric models like SMPL-X, which provide body priors and prevent unnatural views inconsistent with human anatomy. Leveraging the generated multi-view normal and color images, we present SMPLX-initialized explicit human carving to recover realistic textured human meshes efficiently. Extensive experimental results and quantitative evaluations on CAPE and THuman2.1 datasets demonstrate PSHumans superiority in geometry details, texture fidelity, and generalization capability.

Abstract

Multi-View Image Synthesis

Explore similar work

PSHuman: Photorealistic Single-image 3D Human Reconstruction using Cross-Scale Multiview Diffusion and Explicit Remeshing

OrbitForge: Text-to-3D Scene Generation via Reconstruction-Anchored Video Synthesis

Human Interaction-Aware 3D Reconstruction from a Single Image