Tobias Pfaff
  • About
  • Publications
  • Projects
  • Publications
    • NVIDIA OmniDreams: Real-Time Generative World Model for Closed-Loop Autonomous Vehicle Simulation
    • A review of learning-based dynamics models for robotic manipulation
    • Cosmos-Drive-Dreams: Scalable synthetic driving data generation with world foundation models
    • Motion prompting: Controlling video generation with motion trajectories
    • A Review of Graph Neural Network Applications in Mechanics-related Domains
    • Imagen 3
    • Learning rigid-body simulators over implicit shapes
    • Learning rigid dynamics with face interaction graph networks
    • Graph network simulators can learn discontinuous, rigid contact dynamics
    • Learning 3D Particle-based Simulators from RGB-D Videos
    • Multiscale meshgraphnets
    • Physical Design using Differentiable Learned Simulators
    • Predicting Physics in Mesh-reduced Space with Temporal Attention
    • Constraint-based graph network simulator
    • Learned Coarse Models for Efficient Turbulence Simulation
    • Learning ground states of quantum hamiltonians with graph networks
    • Learning mesh-based simulation with graph networks
    • Combining q-learning and search with amortized value estimates
    • Learning to simulate complex physics with graph networks
    • Grandmaster level in StarCraft II using multi-agent reinforcement learning
    • One-shot high-fidelity imitation: Training large-scale deep nets with rl
    • Playing hard exploration games by watching youtube
    • Adaptive tearing and cracking of thin sheets
    • Folding and crumpling adaptive sheets
    • Lagrangian vortex sheets for animating fluids
    • Scalable fluid simulation using anisotropic turbulence particles
    • Field-scale apparent hydraulic parameterisation obtained from TDR time series and inverse modelling
    • Synthetic turbulence using artificial boundary layers
  • Projects
    • OmniDreams
    • Imagen 3
    • Veo
    • Grim Fandango Remastered
    • ARCSim
    • MantaFlow

Imagen 3

Aug 1, 2024 · 1 min read
Go to Project Site

Imagen 3 is Google’s highest-quality generative image model. It can generate images from text, or image-to-image stylization. Imagen is available as part of ImageFX and also integrated into Gemini.

Last updated on Aug 1, 2024
Google
Tobias Pfaff
Authors
Tobias Pfaff
Research Scientist
← OmniDreams Jun 1, 2026
Veo May 1, 2024 →

Published with Hugo Blox Builder — the free, open source website builder that empowers creators.