mlsanityData to 7 Oct 2026

MoCA: Mixture-of-Components Attention for Scalable Compositional 3D Generation

arXiv 2512.07628 ↗First posted 8 Dec 2025cs.CVcited by 5

Not on the map. Why?

Abstract (arXiv)

Compositionality is critical for 3D object and scene generation, but existing part-aware 3D generation methods suffer from poor scalability due to quadratic global attention costs when increasing the number of components. In this work, we present MoCA, a compositional 3D generative model with two key designs: (1) importance-based component routing that selects top-k relevant components for sparse global attention, and (2) unimportant components compression that preserve contextual priors of unselected components while reducing computational complexity of global attention. With these designs, MoCA enables efficient, fine-grained compositional 3D asset creation with scalable number of components. Extensive experiments show MoCA outperforms baselines on both compositional object and scene generation tasks. Project page: https://lizhiqi49.github.io/MoCA

Key facts. Cited by 5 papers in the map (data through 7 Oct 2026). Builds on 51 of its 55 references that are in the map. First posted on arXiv 8 Dec 2025.

Builds on

51 of 55 references.

References · 51

  1. 2023Segment Anythingcited by 8,158
  2. 2022Scalable Diffusion Models with Transformerscited by 5,931
  3. 2022Classifier-Free Diffusion Guidancecited by 5,717
  4. 2024Scaling Rectified Flow Transformers for High-Resolution Image Synthesiscited by 4,094
  5. 2022Flow Straight and Fast: Learning to Generate and Transfer Data with Rectified Flowcited by 3,550
  6. 2017Outrageously Large Neural Networks: The Sparsely-Gated Mixture-of-Experts Layercited by 3,413
  7. 2021Switch Transformers: Scaling to Trillion Parameter Models with Simple and Efficient Sparsitycited by 3,014
  8. 2024Mixtral of Expertscited by 1,785
  9. 2020GShard: Scaling Giant Models with Conditional Computation and Automatic Shardingcited by 1,601
  10. 2023LRM: Large Reconstruction Model for Single Image to 3Dcited by 834
  11. 2024DeepSeekMoE: Towards Ultimate Expert Specialization in Mixture-of-Experts Language Modelscited by 834
  12. 2024Structured 3D Latents for Scalable and Versatile 3D Generationcited by 742
38 more
  1. 2024LGM: Large Multi-View Gaussian Model for High-Resolution 3D Content Creationcited by 730
  2. 2023SyncDreamer: Generating Multiview-consistent Images from a Single-view Imagecited by 638
  3. 20233DShape2VecSet: A 3D Shape Representation for Neural Fields and Generative Diffusion Modelscited by 449
  4. 2024CLAY: A Controllable Large-scale Generative Model for Creating High-quality 3D Assetscited by 444
  5. 20203D-FRONT: 3D Furnished Rooms with layOuts and semaNTicscited by 375
  6. 2021ABO: Dataset and Benchmarks for Real-World 3D Object Understandingcited by 360
  7. 2025Hunyuan3D 2.0: Scaling Diffusion Models for High Resolution Textured 3D Assets Generationcited by 342
  8. 2022Rodin: A Generative Model for Sculpting 3D Digital Avatars Using Diffusioncited by 309
  9. 2024Direct3D: Scalable Image-to-3D Generation via 3D Latent Diffusion Transformercited by 210
  10. 2025TripoSG: High-Fidelity 3D Shape Synthesis using Large-Scale Rectified Flow Modelscited by 204
  11. 2024Auxiliary-Loss-Free Load Balancing Strategy for Mixture-of-Expertscited by 191
  12. 2023WonderJourney: Going from Anywhere to Everywherecited by 157
  13. 2025Hunyuan3D 2.1: From Images to High-Fidelity 3D Assets with Production-Ready PBR Materialcited by 157
  14. 2024CraftsMan3D: High-fidelity Mesh Generation with 3D Native Generation and Interactive Geometry Refinercited by 120
  15. 2024MIDI: Multi-Instance Diffusion for Single Image to 3D Scene Generationcited by 113
  16. 2024InstructScene: Instruction-Driven 3D Indoor Scene Synthesis with Semantic Graph Priorcited by 110
  17. 2025Direct3D-S2: Gigascale 3D Generation Made Easy with Spatial Sparse Attentioncited by 103
  18. 2025CAST: Component-Aligned 3D Scene Reconstruction from an RGB Imagecited by 101
  19. 2024SAMPart3D: Segment Any Part in 3D Objectscited by 95
  20. 2025Hi3DGen: High-fidelity 3D Geometry Generation from Images via Normal Bridgingcited by 95
  21. 2025Sparc3D: Sparse Representation and Construction for High-Resolution 3D Shapes Modelingcited by 84
  22. 2025PartCrafter: Structured 3D Mesh Generation via Compositional Latent Diffusion Transformerscited by 78
  23. 2025SparseFlex: High-Resolution and Arbitrary-Topology 3D Shape Modelingcited by 76
  24. 2024Scaling Diffusion Transformers to 16 Billion Parameterscited by 65
  25. 2024PartGen: Part-level 3D Generation and Reconstruction with Multi-View Diffusion Modelscited by 60
  26. 2024ComboVerse: Compositional 3D Assets Creation Using Spatially-Aware Diffusion Guidancecited by 59
  27. 2024Gen3DSR: Generalizable 3D Scene Reconstruction via Divide and Conquer from a Single Viewcited by 59
  28. 2025HoloPart: Generative 3D Part Amodal Segmentationcited by 57
  29. 2024Part123: Part-aware 3D Reconstruction from a Single-view Imagecited by 54
  30. 2025SceneGen: Single-Image 3D Scene Generation in One Feedforward Passcited by 49
  31. 2025Efficient Part-level 3D Object Generation via Dual Volume Packingcited by 47
  32. 2024REPARO: Compositional 3D Assets Generation with Differentiable 3D Layout Alignmentcited by 45
  33. 2025X-Part: high fidelity and structure coherent shape decompositioncited by 42
  34. 2024Zero-Shot Scene Reconstruction from Single Images with Deep Prior Assemblycited by 34
  35. 2025AutoPartGen: Autogressive 3D Part Generation and Discoverycited by 33
  36. 2025Decompositional Neural Scene Reconstruction with Generative Diffusion Priorcited by 30
  37. 2023BUOL: A Bottom-Up Framework with Occupancy-aware Lifting for Panoptic 3D Scene Reconstruction From A Single Imagecited by 20
  38. 2024Phidias: A Generative Model for Creating 3D Content from Text, Image, and 3D Conditions with Reference-Augmented Diffusioncited by 19

50 most cited of 51 All 51 (JSON)

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2026: 5.

5 papers.

  1. 2026HY3D-Bench: Generation of 3D Assetscited by 11
  2. 2026ISAP-3D: Identity-Slot Aligned Part-Aware 3D Generationcited by 0
  3. 2026Hunyuan3D-Buffalo 1.0: A Unified Multimodal Model for Scalable 3D Generation, Understanding, and Editingcited by 0
  4. 2026MultiCube: Compositional 3D Generation With Part-Level Semantic and Spatial Controlcited by 0
  5. 2026Seg3DParts: Segmentation-Grounded Controllable Part-Level 3D Generationcited by 0