analiza aiqbench
Discrete Voxel Diffusion introduces a novel framework that treats voxel occupancy as native discrete variables to enhance the generation and editing of sparse voxels within structured latent 3D pipelines. By leveraging predictive entropy for uncertainty estimation and proposing lightweight fine-tuning strategies, this method offers improved interpretability and efficiency without requiring additional model evaluations during sampling. This advancement represents a significant step forward in applying diffusion models specifically tailored for discrete spatial data rather than continuous image-like generation.
DVD: Discrete Voxel Diffusion for 3D Generation and Editing
arXiv:2605.07971v3 Announce Type: replace-cross Abstract: We introduce Discrete Voxel Diffusion (DVD), a discrete diffusion framework to generate, assess, and edit sparse voxels for SLat (Structured LATent) based 3D generative pipelines. Although discrete diffusion has not generally displaced continuous diffusion in image-like generation, we show that it can be an effective first-stage prior for sparse voxel scaffolds. By treating voxel occupancy as a native discrete variable, DVD avoids continuous-to-discrete thresholding and provides a simple framework for voxel generation, uncertainty estimation, and editing. Beyond quality gains, DVD provides more interpretable generation dynamics through explicit categorical modeling. Furthermore, we leverage the predictive entropy as a robust uncertainty metric to identify ambiguous voxel regions and complicated samples, facilitating tasks such as data filtering and quality assessment. Finally, we propose a lightweight fine-tuning strategy using block-structured perturbation patterns. This approach empowers the model to inpaint and edit voxels within a single sampling round, requiring negligible auxiliary computation and no additional model evaluations. Code is available at https://github.com/TeCai/DVD.
Abstract:We introduce Discrete Voxel Diffusion (DVD), a discrete diffusion framework to generate, assess, and edit sparse voxels for SLat (Structured LATent) based 3D generative pipelines. Although discrete diffusion has not generally displaced continuous diffusion in image-like generation, we show that it can be an effective first-stage prior for sparse voxel scaffolds. By treating voxel occupancy as a native discrete variable, DVD avoids continuous-to-discrete thresholding and provides a simple framework for voxel generation, uncertainty estimation, and editing. Beyond quality gains, DVD provides more interpretable generation dynamics through explicit categorical modeling. Furthermore, we leverage the predictive entropy as a robust uncertainty metric to identify ambiguous voxel regions and complicated samples, facilitating tasks such as data filtering and quality assessment. Finally, we propose a lightweight fine-tuning strategy using block-structured perturbation patterns. This approach empowers the model to inpaint and edit voxels within a single sampling round, requiring negligible auxiliary computation and no additional model evaluations. Code is available at this https URL.
Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
arXiv:2605.07971 [cs.CV]
(or arXiv:2605.07971v3 [cs.CV] for this version)
Text integral sub licența deschisă a publicației · 176 cuvinte
Subiecte conexe
Fast Diffusion-GAN Model for Generating Symbolic Music with Emotions
arXiv · 15 oct. 2026
DiffusionGemma: Google’s Diffusion-Based Open Model for Faster Text Generation
Analytics Vidhya · 11 iun. 2026
Exploring the Potential of Diffusion Large Language Models in Code Generation
arXiv · 16 sept. 2026
Google AI Releases DiffusionGemma, a 26B MoE Open Model Using Text Diffusion for Up to 4x Faster Generation
MarkTechPost · 10 iun. 2026