Blocks2World: Controlling Realistic Scenes with Editable Primitives
Not on the map. Why?
Abstract (arXiv)
We present Blocks2World, a novel method for 3D scene rendering and editing that leverages a two-step process: convex decomposition of images and conditioned synthesis. Our technique begins by extracting 3D parallelepipeds from various objects in a given scene using convex decomposition, thus obtaining a primitive representation of the scene. These primitives are then utilized to generate paired data through simple ray-traced depth maps. The next stage involves training a conditioned model that learns to generate images from the 2D-rendered convex primitives. This step establishes a direct mapping between the 3D model and its 2D representation, effectively learning the transition from a 3D model to an image. Once the model is fully trained, it offers remarkable control over the synthesis of novel and edited scenes. This is achieved by manipulating the primitives at test time, including translating or adding them, thereby enabling a highly customizable scene rendering process. Our method provides a fresh perspective on 3D scene rendering and editing, offering control and flexibility. It opens up new avenues for research and applications in the field, including authoring and data augmentation.
Key facts. Cited by 4 papers in the map (data through 14 Oct 2026). Builds on 38 of its 60 references that are in the map. First posted on arXiv 7 Jul 2023.
Builds on
38 of 60 references.
References · 38
- 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,744
- 2021High-Resolution Image Synthesis with Latent Diffusion Modelscited by 18,863
- 2021Diffusion Models Beat GANs on Image Synthesiscited by 8,255
- 2022Generative Adversarial Networkscited by 5,627
- 2023Adding Conditional Control to Text-to-Image Diffusion Modelscited by 5,503
- 2016Image-to-Image Translation with Conditional Adversarial Networkscited by 4,527
- 2022DreamFusion: Text-to-3D using 2D Diffusioncited by 3,025
- 2016Categorical Reparameterization with Gumbel-Softmaxcited by 2,739
- 2019DeepSDF: Learning Continuous Signed Distance Functions for Shape Representationcited by 2,689
- 2014Conditional Generative Adversarial Netscited by 2,162
- 2021Vision Transformers for Dense Predictioncited by 1,803
- 2023Zero-1-to-3: Zero-shot One Image to 3D Objectcited by 1,489
26 more
- 2023T2I-Adapter: Learning Adapters to Dig out More Controllable Ability for Text-to-Image Diffusion Modelscited by 1,351
- 2021Efficient Geometry-aware 3D Generative Adversarial Networkscited by 1,205
- 2023ZoeDepth: Zero-shot Transfer by Combining Relative and Metric Depthcited by 657
- 2022Score Jacobian Chaining: Lifting Pretrained 2D Diffusion Models for 3D Generationcited by 597
- 2021StyleNeRF: A Style-based 3D-Aware Generator for High-resolution Image Synthesiscited by 477
- 2023Make-It-3D: High-Fidelity 3D Creation from A Single Image with Diffusion Priorcited by 343
- 2017Photographic Image Synthesis with Cascaded Refinement Networkscited by 223
- 2023Drag Your GAN: Interactive Point-based Manipulation on the Generative Image Manifoldcited by 219
- 2019BSP-Net: Generating Compact Meshes via Binary Space Partitioningcited by 206
- 2016Learning Shape Abstractions by Assembling Volumetric Primitivescited by 181
- 2019CvxNet: Learnable Convex Decompositioncited by 174
- 2021Visual Room Rearrangementcited by 102
- 2018LayoutNet: Reconstructing the 3D Room Layout from a Single RGB Imagecited by 95
- 20173D-PRNN: Generating Shape Primitives with Recurrent Neural Networkscited by 77
- 2018PlaneNet: Piece-wise Planar Reconstruction from a Single RGB Imagecited by 77
- 2023LEGO-Net: Learning Regular Rearrangements of Objects in Roomscited by 76
- 2021StructFormer: Learning Spatial Structure for Language-Guided Semantic Rearrangement of Novel Objectscited by 59
- 2022Towards High-Fidelity Single-view Holistic Reconstruction of Indoor Scenescited by 44
- 2021NeRP: Neural Rearrangement Planning for Unknown Objectscited by 34
- 2022Efficient Multi-Task RGB-D Scene Analysis for Indoor Environmentscited by 33
- 2022User-Controllable Latent Transformer for StyleGAN Image Layout Editingcited by 28
- 2020General 3D Room Layout from a Single View by Render-and-Comparecited by 21
- 2021Monte Carlo Scene Search for 3D Scene Understandingcited by 19
- 2021Cuboids Revisited: Learning Robust 3D Shape Fitting to Single RGB Imagescited by 17
- 2022MonteBoxFinder: Detecting and Filtering Primitives to Fit a Noisy Point Cloudcited by 11
- 2023Dequantization and Color Transfer with Diffusion Modelscited by 4
Built on by
Citing papers per year: 2023: 2; 2024: 0; 2025: 2.
4 papers.
- 2023OBJECT 3DIT: Language-guided 3D-aware Image Editingcited by 63
- 2025Visual Jenga: Discovering Object Dependencies via Counterfactual Inpaintingcited by 11
- 2023Convex Decomposition of Indoor Scenescited by 7
- 2025PrITTI: Primitive-based Generation of Controllable and Editable 3D Semantic Urban Scenescited by 1
- 2025PrITTI: Primitive-based Generation of Controllable and Editable 3D Semantic Urban Scenes23 Jun 2025
- 2025Visual Jenga: Discovering Object Dependencies via Counterfactual Inpainting27 Mar 2025
- 2023OBJECT 3DIT: Language-guided 3D-aware Image Editing20 Jul 2023
- 2023Convex Decomposition of Indoor Scenes9 Jul 2023