Score Distillation Sampling

Also known as SDS

Follow this topic Ask about this topic

Momentum

0 papers in the last four weeks, against 1 the four weeks before. 0.0% of all new papers.

Jul 13Week of Sep 28

Latest papers 11

All topics

PixSDS: Why Latent SDS Makes Noisy Pixels
Aug 13, 2026Vsevolod SkorokhodovDiffusion Model Guidance Latent Diffusion Models
EPFL, Lausanne, Switzerland
Anchoring and Steering Diffusion: Enhancing the Faithfulness of Text-to-Image Generation at Inference Time
Jul 29, 2026Xinyi Wang, Yuyang Huang, Yalin Su +4Diffusion Model Guidance Text-Image Alignment
Shanghai Jiao Tong University, Shanghai, China
Generative Video Compression with Adaptive Score Distillation
Jul 24, 2026Naifu Xue, Zhaoyang Jia, Haosen Li +7Video Diffusion Models Diffusion Model Distillation
Communication University of China, Beijing, China · Microsoft Research Asia · University of Science and Technology of China, Hefei, China +1
PoseAlign: Sculpting Pose-Consistent Meshes via Text-Guided Deformation
Jul 12, 2026Shijin Wang, Zichong Chen, Yang Zhou +1Diffusion Guidance Score Distillation Sampling
Guangdong Provincial Key Lab. of Visual Media and Multidim. Intelligence, CSSE, Shenzhen University, China.
Stratified Multi-View Aggregation for Score Distillation
Jun 29, 2026Marian Lupascu, Mihai-Sorin Stupariu, Wongi Park +1Multi-View Consistency Generative Modeling
Department of Computer Science, University of Bucharest, Romania · Adobe Research
Occlusion-Robust Multi-Object Decoupling for Physics-Based Robotic Interaction
Jun 28, 2026Xin Dong, Lihan Zhang, Tianru Dai +23D Reconstruction Sparse-View 3D Reconstruction
1Pengcheng Laboratory · 2Shenzhen International Graduate School, Tsinghua University
Inclusive Interactive Collisions for Multi-View Consistent Compositional 3D Generation
Jun 23, 2026Chang Liu, Mingwen Shao, Xiang Lv +5Multi-View Consistency 3D Gaussian Splatting
School of Computer Science and Technology, China University of Petroleum (East China), Qingdao, 266580, China · Artificial Intelligence Research Institute, Shenzhen University of Advanced Technology, Shenzhen, 518107, China · College of Computer Science, Northwest University, Xi’an, 710127, China
Single-Line Drawing Generation via Semantics-Driven Optimization
Jun 1, 2026Tanguy Magne, Alexandre Binninger, Ruben Wiersma +1Image Generation Score Distillation Sampling
ETH Zurich, Zurich, Switzerland
Probability-Flow Distillation: Exact Wasserstein Gradient Flow for High-Fidelity 3D Generation
May 9, 2026Rohith Ramanan, A. N. RajagopalanDiffusion Model Distillation Wasserstein Gradient Flows
Department of Physics Indian Institute of Technology Madras · Department of Electrical Engineering Indian Institute of Technology Madras
Robust 3D-Masked Part-level Editing in 3D Gaussian Splatting with Regularized Score Distillation Sampling
Jul 15, 2025Hayeon Kim, Ji Ha Jang, Se Young Chun3D Gaussian Splatting 3D Shape Editing
Dept. of Electrical and Computer Engineering, Seoul National University, Republic of Korea · INMC & IPAI, Seoul National University, Republic of Korea
DLM-One: Diffusion Language Models for One-Step Sequence Generation
May 30, 2025Tianqi Chen, Shujian Zhang, Mingyuan ZhouDiffusion Language Models Diffusion Model Distillation
The University of Texas at Austin, Austin, TX 78712, USA

Aug 13, 2026, cs.CVJ/K move · Enter open · S save

PixSDS: Why Latent SDS Makes Noisy Pixels

Vsevolod Skorokhodov

EPFL, Lausanne, Switzerland

Diffusion Model Guidance Latent Diffusion Models Score Distillation Sampling Text-to-3D Generation

Open paper AskPDF arXiv

Score Distillation Sampling (SDS) enables text-to-3D generation by optimizing rendered images with a pretrained diffusion prior, but latent SDS often produces structured color artifacts and high-frequency texture noise. We identify a failure mode of latent SDS caused by VAE-induced pixel drift: the optimized image can move along pixel-space directions that are weakly constrained by the VAE encoder, so its latent representation remains clean and semantically meaningful while the image itself accumulates visible artifacts. We support this diagnosis with controlled 2D SDS experiments, VAE-only optimization, and a simplified analysis showing that encoder-like latent objectives can amplify image-space noise when the inverse mapping to pixels is underconstrained. Motivated by this observation, we propose PixSDS, a lightweight VAE-consistent gradient repair method. PixSDS decodes a latent SDS lookahead step and uses the decoded image as a clean direction for pixel-space optimization, reducing motion in VAE-inconsistent directions without retraining the diffusion model, changing the renderer, or replacing the SDS objective. Experiments in 2D optimization and text-to-3D generation show that PixSDS substantially reduces structured artifacts while preserving semantic content. Code is publicly available at https://sevashasla.github.io/pixsds-webpage/.

Momentum

Related topics

Latest papers 11