mlsanityData to 7 Oct 2026

DiffSoup: Direct Differentiable Rasterization of Triangle Soup for Extreme Radiance Field Simplification

arXiv 2603.27151 ↗First posted 28 Mar 2026cs.GR cs.CVcited by 3

Not on the map. Why?

Abstract (arXiv)

Radiance field reconstruction aims to recover high-quality 3D representations from multi-view RGB images. Recent advances, such as 3D Gaussian splatting, enable real-time rendering with high visual fidelity on sufficiently powerful graphics hardware. However, efficient online transmission and rendering across diverse platforms requires drastic model simplification, reducing the number of primitives by several orders of magnitude. We introduce DiffSoup, a radiance field representation that employs a soup (i.e., a highly unstructured set) of a small number of triangles with neural textures and binary opacity. We show that this binary opacity representation is directly differentiable via stochastic opacity masking, enabling stable training without a mollifier (i.e., smooth rasterization). DiffSoup can be rasterized using standard depth testing, enabling seamless integration into traditional graphics pipelines and interactive rendering on consumer-grade laptops and mobile devices. Code is available at https://github.com/kenji-tojo/diffsoup.

Key facts. Cited by 3 papers in the map (data through 7 Oct 2026). Builds on 39 of its 62 references that are in the map. First posted on arXiv 28 Mar 2026.

Builds on

39 of 62 references.

References · 39

  1. 2015Deep Residual Learning for Image Recognitioncited by 51,809
  2. 2019PyTorch: An Imperative Style, High-Performance Deep Learning Librarycited by 20,160
  3. 20233D Gaussian Splatting for Real-Time Radiance Field Renderingcited by 6,141
  4. 2022Instant Neural Graphics Primitives with a Multiresolution Hash Encodingcited by 3,698
  5. 2013Estimating or Propagating Gradients Through Stochastic Neurons for Conditional Computationcited by 2,031
  6. 2021Mip-NeRF 360: Unbounded Anti-Aliased Neural Radiance Fieldscited by 2,013
  7. 2021Mip-NeRF: A Multiscale Representation for Anti-Aliasing Neural Radiance Fieldscited by 1,957
  8. 2021NeuS: Learning Neural Implicit Surfaces by Volume Rendering for Multi-view Reconstructioncited by 1,701
  9. 2021Plenoxels: Radiance Fields without Neural Networkscited by 1,607
  10. 20242D Gaussian Splatting for Geometrically Accurate Radiance Fieldscited by 1,061
  11. 2021Volume Rendering of Neural Implicit Surfacescited by 990
  12. 2020Accelerating 3D Deep Learning with PyTorch3Dcited by 781
27 more
  1. 2023Neuralangelo: High-Fidelity Neural Surface Reconstructioncited by 464
  2. 2020Modular Primitives for High-Performance Differentiable Renderingcited by 439
  3. 2021Neural Geometric Level of Detail: Real-time Rendering with Implicit 3D Shapescited by 426
  4. 2019Soft Rasterizer: A Differentiable Renderer for Image-based 3D Reasoningcited by 420
  5. 2022MobileNeRF: Exploiting the Polygon Rasterization Pipeline for Efficient Neural Field Rendering on Mobile Architecturescited by 308
  6. 2019Differentiable Surface Splatting for Point-based Geometry Processingcited by 238
  7. 2019Learning to Predict 3D Objects with an Interpolation-based Differentiable Renderercited by 221
  8. 2020Differentiable Rendering: A Surveycited by 123
  9. 2023Adaptive Shells for Efficient Neural Radiance Field Renderingcited by 69
  10. 20243D Convex Splatting: Radiance Field Rendering with 3D Smooth Convexescited by 50
  11. 2025Triangle Splatting for Real-Time Radiance Field Renderingcited by 50
  12. 2024Texture-GS: Disentangling the Geometry and Texture for 3D Gaussian Splatting Editingcited by 47
  13. 2022Multi-View Mesh Reconstruction with Neural Deferred Shadingcited by 44
  14. 2024Textured Gaussians for Enhanced 3D Scene Appearance Modelingcited by 40
  15. 2024GStex: Per-Primitive Texturing of 2D Gaussian Splatting for Decoupled Appearance and Geometry Modelingcited by 30
  16. 2022Differentiable Rendering of Neural SDFs through Reparameterizationcited by 26
  17. 2021Appearance-Driven Automatic 3D Model Simplificationcited by 24
  18. 2024BillBoard Splatting (BBSplat): Learnable Textured Primitives for Novel View Synthesiscited by 23
  19. 2022VectorAdam for Rotation Equivariant Geometry Optimizationcited by 17
  20. 2025StochasticSplats: Stochastic Rasterization for Sorting-Free 3D Gaussian Splattingcited by 17
  21. 2024Rasterized Edge Gradients: Handling Discontinuities Differentiablycited by 14
  22. 2024Gaussian Billboards: Expressive 2D Gaussian Splatting with Texturescited by 13
  23. 2023A Theory of Topological Derivatives for Inverse Rendering of Geometrycited by 6
  24. 2024Volumetric Surfaces: Representing Fuzzy Geometries with Layered Meshescited by 5
  25. 2025Triangle Splatting+: Differentiable Rendering with Opaque Trianglescited by 5
  26. 2024Many-Worlds Inverse Renderingcited by 3
  27. 2025Radiance Surfaces: Optimizing Surface Representations with a 5D Radiance Field Losscited by 3

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2026: 3.

3 papers.

  1. 2026Triangle Splatting SLAMcited by 0
  2. 2026Inverse Rendering for Modeling with Line Primitivescited by 0
  3. 2026MeshSplatBench: A Unified Benchmark for Triangle- and Mesh-Based Neural Renderingcited by 0