Related Experiment Video
Updated: Aug 12, 2026

Subject-specific Musculoskeletal Model for Studying Bone Strain During Dynamic Motion
Published on: April 11, 2018
JointConn-v2: Learning a joint vector field with diffusion transformers for cross-modal connectivity and
Honggang Zhao1, Yi-Jun Yang1, Wei Zeng2
1School of Computer Science and Technology, Xi'an Jiaotong University, Xi'an, 710049, Shaanxi, China.
This study introduces JointConn-v2, improving diffusion Transformers for image-depth synthesis by enhancing cross-modal attention and stabilizing depth prediction. It balances edge control and consistency for better geometric and semantic alignment.
Area of Science:
- Computer Vision
- Deep Learning
- Generative Models
Background:
- Diffusion Transformers are used for image-depth synthesis.
- Existing methods face challenges with geometric distortions and over-coupling between semantic and depth information.
Purpose of the Study:
- To develop a unified framework, JointConn-v2, for relative-depth-conditioned and joint image-depth synthesis.
- To address bottlenecks in cross-modal attention and depth branch stability.
Main Methods:
- Proposed JointConn-v2 framework with Gated Cross-Modal Weighted Flow Matching (GCM-WFM).
- Incorporated Swap-Q cross-attention, Geometric Mask Bias, Regional Routing, and Content Gate.
- Introduced GCM-WFM for training, regressing a joint vector field with temporal, geometric, gating, and routing terms.
Main Results:
- Achieved improved balance between edge controllability and cross-modal consistency.
- Reduced geometric distortions around edges and structural regions.
- Enhanced stability in joint image-depth synthesis.
Conclusions:
- JointConn-v2 effectively improves relative-depth-conditioned and joint image-depth synthesis.
- The proposed methods mitigate key limitations of previous diffusion Transformer approaches.
- The framework offers better control over geometric details and semantic consistency.
Related Concept Videos
Bone Structure
Functional Classification of Joints
The functional classification of joints is determined by the amount of mobility between the adjacent bones. Joints are functionally classified as a synarthrosis or immobile joint, an amphiarthrosis or slightly moveable joint, or as a diarthrosis, a freely moveable joint. Fibrous and cartilaginous joints can be functionally classified as either synarthroses or amphiarthroses, whereas all synovial joints are classified as diarthroses.
Synarthrosis
An immobile...
Structural Classification of Joints
A fibrous joint is where the adjacent bones are united by fibrous connective...
Virtual Work for a System of Connected Rigid Bodies
Next,...
Deformations in a Transverse Cross Section
As the material stretches, it expands or contracts in orthogonal directions to the load. This phenomenon varies...
Modeling with Differential Equations

