Skip to content

Latest commit

 

History

History
66 lines (55 loc) · 6.1 KB

diffusion_model.md

File metadata and controls

66 lines (55 loc) · 6.1 KB
Raw

Diffusion Model

Controllable

  • ICCV 2023, Unsupervised Compositional Concepts Discovery with Text-to-Image Generative Models, Website
  • arXiv 2023.02, ControlNet: Adding Conditional Control to Text-to-Image Diffusion Models, arXiv / Github
  • CVPR 2023, DreamBooth: Fine Tuning Text-to-Image Diffusion Models for Subject-Driven Generation, Website
  • ICLR 2023, Prompt-to-Prompt Image Editing with Cross-Attention Control, arXiv / Website
  • ICLR 2023, An Image is Worth One Word: Personalizing Text-to-Image Generation using Textual Inversion, Website

Applications

  • ICLR 2023, MDM: Human Motion Diffusion Model, Website
  • arXiv 2022.05, Diffusion-LM Improves Controllable Text Generation, arXiv

High Resolution

  • arXiv 2023.10, MDM: Matryoshka Diffusion Models, arXiv

  • arXiv 2023.10, Latent Consistency Models: Synthesizing High-Resolution Images with Few-step Inference, Website

  • arXiv 2022.10, Imagen Video: High Definition Video Generation with Diffusion Models, arXiv / Website

  • NIPS 2022, k-diffusion: Elucidating the Design Space of Diffusion-Based Generative Models, arXiv / Github

  • CVPR 2022, Stable Diffusion: High-Resolution Image Synthesis with Latent Diffusion Models, arXiv

  • NIPS 2022, Imagen: Photorealistic Text-to-Image Diffusion Models with Deep Language Understanding, arXiv

  • arXiv 2022.07, Classifier-Free Diffusion Guidance, arXiv

  • NIPS 2021, Guided Diffusion: Diffusion Models Beat GANS on Image Synthesis, arXiv / Github

  • ICLR 2021, Score-Based Generative Modeling through Stochastic Differential Equations, arXiv

  • ICML 2021, iDDPM: Improved Denoising Diffusion Probabilistic Models, arXiv / Github

  • NIPS 2020, DDPM: Denoising Diffusion Probabilistic Models, Website

Feature Utilization

  • arXiv 2023.10, FreeU: Free Lunch in Diffusion U-Net, Website
  • arXiv 2023.06, Emergent Correspondence from Image Diffusion, Website
  • CVPR 2023, ODISE: Open-Vocabulary Panoptic Segmentation with Text-to-Image Diffusion Models, Website

Sampling

  • ICLR 2023 spotlight, gDDIM: Generalized denoising diffusion implicit models, arXiv
  • ICLR 2023, DEIS: Fast Sampling of Diffusion Models with Exponential Integrator, arXiv / Github
  • ICLR 2022, Progressive Distillation for Fast Sampling of Diffusion Models, Openreview / Github
  • ICLR 2022 outstanding paper award, Analytic-DPM: an Analytic Estimate of the Optimal Reverse Variance in Diffusion Probabilistic Models , arXiv
  • NeurIPS 2022 oral, DPM-Solver: A Fast ODE Solver for Diffusion Probabilistic Model Sampling in Around 10 Steps , arXiv
  • ICLR 2021, DDIM: Denoising Diffusion Implicit Models, arXiv / Github

3D

  • CoRL 2023, ChainedDiffuser: Unifying Diffusion Models with Action Detection Transformers for Multi-task Robotic Manipulation, OpenReview
  • arXiv 2023.08, Diffusion with Forward Models: Solving Stochastic Inverse Problems Without Direct Supervision, Website
  • arXiv 2023.07, NIFTY: Neural Object Interaction Fields for Guided Human Motion Synthesis, Website
  • arXiv 2023.03, Let 2D Diffusion Model Know 3D-Consistency for Robust Text-to-3D Generation, arXiv / GitHub / Website
    • main intuition is to incorporate 3D information into 2D diffusion model
    • and still, use 2D diffusion model to optimize nerf
  • ICLR 2023 outstanding paper award, DreamFusion: Text-to-3D using 2D Diffusion, OpenReview
  • NIPS 2022, LION: Latent Point Diffusion Models for 3D Shape Generation, Website
  • CVPR 2022, Zero-Shot Text-Guided Object Generation with Dream Fields, Website / GitHub
  • arXiv 2022.12, Point-E: A System for Generating 3D Point Clouds from Complex Prompts, arXiv / GitHub
    • text -> image -> point cloud
    • use transformer to generate point cloud
    • point cloud diffusion not condition on language
  • arXiv 2022.11, NFD: 3D Neural Field Generation using Triplane Diffusion, arXiv / Website
    • two-stage model
    • use 2D diffusion model to generate three planes
    • use occupancy network to generate 3D point cloud based on three plane interpolation
  • arXiv 2022.08, 3DiM: Novel View Synthesis with Diffusion Models, Website