mlsanityData to 7 Oct 2026

LAMP: Data-Efficient Linear Affine Weight-Space Models for Parameter-Controlled 3D Shape Generation and Extrapolation

arXiv 2510.22491 ↗First posted 26 Oct 2025cs.LG cs.CE cs.CVcited by 3

Not on the map. Why?

Abstract (arXiv)

Generating high-fidelity 3D geometries under explicit parameter constraints is central to engineering design, yet current methods often require large datasets and fail to provide reliable control beyond the training distribution. We introduce LAMP, a data-efficient framework for controllable and interpretable 3D generation that aligns signed distance function (SDF) decoders by overfitting each exemplar from a shared initialization, then generates new designs by solving a parameter-constrained affine mixing problem in the aligned weight space. To improve reliability, we propose a linearity-mismatch safety metric that detects when mixed decoders leave the valid local regime. We evaluate LAMP on DrivAerNet++, BlendedNet, and additional industry-level vehicle families, including sports cars, SUVs, and convertibles. LAMP enables controlled interpolation with as few as 50 samples, safe extrapolation up to 100% beyond training ranges, and performance-guided optimization under fixed parameters, outperforming conditional autoencoder and Deep Network Interpolation (DNI) baselines in extrapolation, data efficiency, and parameter fidelity. Our results demonstrate that LAMP advances controllable, data-efficient, and safe 3D generation for design exploration, dataset generation, and performance-driven optimization.

Key facts. Cited by 3 papers in the map (data through 7 Oct 2026). Builds on 43 of its 54 references that are in the map. First posted on arXiv 26 Oct 2025.

Builds on

43 of 54 references.

References · 43

  1. 2016PointNet: Deep Learning on Point Sets for 3D Classification and Segmentationcited by 5,223
  2. 2022DreamFusion: Text-to-3D using 2D Diffusioncited by 3,006
  3. 2019DeepSDF: Learning Continuous Signed Distance Functions for Shape Representationcited by 2,682
  4. 2018Occupancy Networks: Learning 3D Reconstruction in Function Spacecited by 2,037
  5. 2022Model soups: averaging weights of multiple fine-tuned models improves accuracy without increasing inference timecited by 1,366
  6. 2022Editing Models with Task Arithmeticcited by 1,076
  7. 2024Structured 3D Latents for Scalable and Versatile 3D Generationcited by 742
  8. 2019Differentiable Volumetric Rendering: Learning Implicit 3D Representations without 3D Supervisioncited by 699
  9. 2022Point-E: A System for Generating 3D Point Clouds from Complex Promptscited by 694
  10. 2021Merging Models with Fisher-Weighted Averagingcited by 613
  11. 2016Learning a Probabilistic Latent Space of Object Shapes via 3D Generative-Adversarial Modelingcited by 558
  12. 2022LION: Latent Point Diffusion Models for 3D Shape Generationcited by 552
31 more
  1. 2018AtlasNet: A Papier-Mach'e Approach to Learning 3D Surface Generationcited by 539
  2. 20233DShape2VecSet: A 3D Shape Representation for Neural Fields and Generative Diffusion Modelscited by 449
  3. 2022GET3D: A Generative Model of High Quality 3D Textured Shapes Learned from Imagescited by 446
  4. 2024CLAY: A Controllable Large-scale Generative Model for Creating High-quality 3D Assetscited by 444
  5. 2019SAL: Sign Agnostic Learning of Shapes from Raw Datacited by 389
  6. 2020Implicit Functions in Feature Space for 3D Shape Reconstruction and Completioncited by 343
  7. 2022SDFusion: Multimodal 3D Shape Completion, Reconstruction, and Generationcited by 287
  8. 20223D Neural Field Generation using Triplane Diffusioncited by 274
  9. 2021DeepCAD: A Deep Generative Network for Computer-Aided Design Modelscited by 266
  10. 2021CLIP-Forge: Towards Zero-Shot Text-to-Shape Generationcited by 254
  11. 2023Michelangelo: Conditional 3D Shape Generation based on Shape-Image-Text Aligned Latent Representationcited by 225
  12. 2022From data to functa: Your data point is a function and you can treat it like onecited by 183
  13. 2023OpenShape: Scaling Up 3D Shape Representation Towards Open-World Understandingcited by 183
  14. 2019SDFDiff: Differentiable Rendering of Signed Distance Fields for 3D Shape Optimizationcited by 159
  15. 2023HyperDiffusion: Generating Implicit Neural Fields with Weight-Space Diffusioncited by 152
  16. 2022Diffusion-SDF: Conditional Generative Modeling of Signed Distance Functionscited by 133
  17. 2019SDM-NET: Deep Generative Network for Structured Deformable Meshcited by 113
  18. 2021Deep Generative Models in Engineering Design: A Reviewcited by 103
  19. 2024DrivAerNet++: A Large-Scale Multimodal Car Dataset with Computational Fluid Dynamics Simulations and Deep Learning Benchmarkscited by 74
  20. 2020SketchGraphs: A Large-Scale Dataset for Modeling Relational Geometry in Computer-Aided Designcited by 72
  21. 2023SALAD: Part-Level Latent Diffusion for 3D Shape Generation and Manipulationcited by 66
  22. 2024CAD-Recode: Reverse Engineering CAD Code from Point Cloudscited by 63
  23. 2018Deep Network Interpolation for Continuous Imagery Effect Transitioncited by 47
  24. 2024Diffusion Soup: Model Merging for Text-to-Image Diffusion Modelscited by 25
  25. 2025BlendedNet: A Blended Wing Body Aircraft Dataset and Surrogate Model for Aerodynamic Predictionscited by 16
  26. 2022Learning from Randomly Initialized Neural Network Featurescited by 11
  27. 2025VideoCAD: A Dataset and Model for Learning Long-Horizon 3D CAD UI Interactions from Videocited by 9
  28. 2021GAN Cocktail: mixing GANs without dataset accesscited by 8
  29. 2024GetMesh: A Controllable Model for High-quality Mesh Generation and Manipulationcited by 8
  30. 2024VehicleSDF: A 3D generative model for constrained engineering design via surrogate modelingcited by 6
  31. 2021A Mechanism for Producing Aligned Latent Spaces with Autoencoderscited by 5

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2026: 3.

3 papers.

  1. 2026FLARE: A Data-Efficient Surrogate for Predicting Displacement Fields in Directed Energy Depositioncited by 0
  2. 2026GeoFunFlow-3D: A Physics-Guided Generative Flow Matching Framework for High-Fidelity 3D Aerodynamic Inference over Complex Geometriescited by 0
  3. 2026CertMix: Certified, Data-Efficient Metamaterial Design by Affine Mixing of Aligned Neural-Implicit Weight Spacescited by 0