Zifan Shi
Zifan Shi has a total of 38 co-authors. Here are the top co-authors by number of shared publications:
- Hao Tan
Adobe Research, San Jose, USA
2 shared publications - Baihan Yang
University of California San Diego, La Jolla, USA
1 shared publication - Beverly Feng
Department of Chemistry and Biochemistry
1 shared publication - Chen Wang
University of Pennsylvania, Philadelphia, USA
1 shared publication - D.L. Yu
Southwestern Institute of Physics, PO Box 432, Chengdu 610041, China
1 shared publication H.S. Xie
Institute for Fusion Theory and Simulation, Zhejiang University, Hangzhou 310027, People's Republic of China
1 shared publication
Latest Publications
- Open Access
- 19 July 2026
- Published by Association for Computing Machinery (ACM)
- p. 1-12
OmniRoam: World Wandering via Long-Horizon Panoramic Video Generation
- 19 July 2026
- Published by Association for Computing Machinery (ACM)
- p. 1-12
Open AccessOmniRoam: World Wandering via Long-Horizon Panoramic Video Generation
- Top Cited PapersOpen Access
- 27 July 2025
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Graphics
- Vol. 44 (4) , 1-12
RigAnything: Template-Free Autoregressive Rigging for Diverse 3D Assets
We present RigAnything , a novel autoregressive transformer-based model, which makes 3D assets rig-ready by probabilistically generating joints and skeleton topologies and assigning skinning weights in a template-free manner. Unlike most existing auto-rigging methods, which rely on predefined skeleton templates and are limited to specific categories like humanoid, RigAnything approaches the rigging problem in an autoregressive manner, iteratively predicting the next joint based on the global input shape and the previous prediction. While autoregressive models are typically used to generate sequential data, RigAnything extends its application to effectively learn and represent skeletons, which are inherently tree structures. To achieve this, we organize the joints in a breadth-first search (BFS) order, enabling the skeleton to be defined as a sequence of 3D locations and the parent index. Furthermore, our model improves the accuracy of position prediction by leveraging diffusion modeling, ensuring precise and consistent placement of joints within the hierarchy. This formulation allows the autoregressive model to efficiently capture both spatial and hierarchical relationships within the skeleton. Trained end-to-end on both RigNet and Objaverse datasets, RigAnything demonstrates state-of-the-art performance across diverse object types, including humanoids, quadrupeds, marine creatures, insects, and many more, surpassing prior methods in quality, robustness, generalizability, and efficiency. It achieves significantly faster performance than existing auto-rigging methods, completing rigging in under a few seconds per shape.
- 27 July 2025
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Graphics
- Vol. 44 (4) , 1-12
Top Cited PapersOpen AccessRigAnything: Template-Free Autoregressive Rigging for Diverse 3D Assets
We present RigAnything , a novel autoregressive transformer-based model, which makes 3D assets rig-ready by probabilistically generating joints and skeleton topologies and assigning skinning weights in a template-free manner. Unlike most existing auto-rigging methods, which rely on predefined skeleton templates and are limited to specific categories like humanoid, RigAnything approaches the rigging problem in an autoregressive manner, iteratively predicting the next joint based on the global input shape and the previous prediction. While autoregressive models are typically used to generate sequential data, RigAnything extends its application to effectively learn and represent skeletons, which are inherently tree structures. To achieve this, we organize the joints in a breadth-first search (BFS) order, enabling the skeleton to be defined as a sequence of 3D locations and the parent index. Furthermore, our model improves the accuracy of position prediction by leveraging diffusion modeling, ensuring precise and consistent placement of joints within the hierarchy. This formulation allows the autoregressive model to efficiently capture both spatial and hierarchical relationships within the skeleton. Trained end-to-end on both RigNet and Objaverse datasets, RigAnything demonstrates state-of-the-art performance across diverse object types, including humanoids, quadrupeds, marine creatures, insects, and many more, surpassing prior methods in quality, robustness, generalizability, and efficiency. It achieves significantly faster performance than existing auto-rigging methods, completing rigging in under a few seconds per shape.
- 19 February 2016
- Published by IOP Publishing in Nuclear Fusion
- Vol. 56 (3) , 036018
Core-localized Alfvénic modes driven by energetic ions in HL-2A NBI plasmas with weak magnetic shears
Recent experimental results that are associated with the core-localized (i.e. normalized radius ) Alfvénic modes in HL-2A neutral beam injection (NBI) plasmas with weak magnetic shears are reported. In the different plasma parameter regions, the energetic ions produced by the NBI drive multiple Alfvénic instabilities, such as the toroidal Alfvén eigenmode (TAE), beta-induced Alfvén eigenmode, reversed shear Alfvén eigenmode (RSAE) and fishbone and energetic particle mode (EPM). Here, we focus on the high-frequency RSAE (HFRSAE) and resonant kinetic ballooning mode (rKBM). A group of downward-sweeping frequency coherent modes (HFRSAEs) with 100 < f < 500 kHz and n = 3–7 are often observed with an increase in the edge safety factor, q a . Their measured frequency is more than that of the TAEs, and . The analysis suggests that these modes localize inside the high-order Alfvén eigenmode (AE) gap of the Alfvénic continuum, and their eigenfrequency and eigenfunction depend on the and q-profile. When the core plasma density is more than m−3 and the impurity or supersonic molecular beam enters the bulk plasma, the profiles of the plasma density/pressure peak, and the magnetic shear is weak or negative. In this case, a group of multi-harmonic coherent modes (rKBMs) with 30 < f < 150 kHz and n = 2–9 are observed through multiple diagnostic techniques, and , where is the diamagnetic drift frequency of the thermal ion. It is found that the HFRSAEs can transit into the rKBMs when the density profile suddenly peaks. Neutron monitoring outside the vacuum chamber demonstrates that the HFRSAE and rKBM both degrade the confinement of the energetic ions. The rKBM instabilities also affect the bulk plasma performance.
- 19 February 2016
- Published by IOP Publishing in Nuclear Fusion
- Vol. 56 (3) , 036018
Core-localized Alfvénic modes driven by energetic ions in HL-2A NBI plasmas with weak magnetic shears
Recent experimental results that are associated with the core-localized (i.e. normalized radius ) Alfvénic modes in HL-2A neutral beam injection (NBI) plasmas with weak magnetic shears are reported. In the different plasma parameter regions, the energetic ions produced by the NBI drive multiple Alfvénic instabilities, such as the toroidal Alfvén eigenmode (TAE), beta-induced Alfvén eigenmode, reversed shear Alfvén eigenmode (RSAE) and fishbone and energetic particle mode (EPM). Here, we focus on the high-frequency RSAE (HFRSAE) and resonant kinetic ballooning mode (rKBM). A group of downward-sweeping frequency coherent modes (HFRSAEs) with 100 < f < 500 kHz and n = 3–7 are often observed with an increase in the edge safety factor, q a . Their measured frequency is more than that of the TAEs, and . The analysis suggests that these modes localize inside the high-order Alfvén eigenmode (AE) gap of the Alfvénic continuum, and their eigenfrequency and eigenfunction depend on the and q-profile. When the core plasma density is more than m−3 and the impurity or supersonic molecular beam enters the bulk plasma, the profiles of the plasma density/pressure peak, and the magnetic shear is weak or negative. In this case, a group of multi-harmonic coherent modes (rKBMs) with 30 < f < 150 kHz and n = 2–9 are observed through multiple diagnostic techniques, and , where is the diamagnetic drift frequency of the thermal ion. It is found that the HFRSAEs can transit into the rKBMs when the density profile suddenly peaks. Neutron monitoring outside the vacuum chamber demonstrates that the HFRSAE and rKBM both degrade the confinement of the energetic ions. The rKBM instabilities also affect the bulk plasma performance.