Generative media

Generating images, video, 3D and audio: diffusion, flow matching, editing, generative pipelines

43 articles

All Agents Language models Reinforcement learning Vision Generative media Speech & audio Robotics & embodied AI Safety & security Evaluation & benchmarks Efficiency & systems Code & software AI for science & health Theory

Viewgrid list compact

Adaptive reference attention reuse speeds up conditioned diffusion transformers

Three-stage latent generator reaches 1.11 FID on ImageNet

Fine-grained routing makes sparse video attention faster without quality loss

High-fidelity 3D texture models can train on images alone

Projecting critic errors stabilizes four-step video diffusion distillation

A method that generates consistent video observations across multiple vehicles in a shared driving scene

Diffusion models benefit most from teacher features they still lack

Gaussian mixtures improve distributional training for one-step image generators

Keeping token uncertainty improves discrete diffusion generation and fast sampling

One Flow Model Can Reconstruct Scenes or Generate Missing Detail

Training-free wrapper accelerates causal video diffusion to 50 FPS without quality loss

Dynamic optimization loop aligns 3D generators with 2D diffusion priors to improve realism

Permutation-equivariant flow matching generates neural network weights without alignment

Persistent memory is essential for coherent autoregressive video generation

Synthetic GUI trajectories train agents without running target software

Simulator-aligned training makes world models more responsive to actions

Staged AI agent improves generation of positive and negative space compositions

Motion-space analysis exposes object movement flaws in generated videos

Tracking world states improves coherence in generated video stories

Stateful code revision improves programmable image and video generation

Reciprocal attention regularization improves alignment in joint video generation

Dynamics-aware distillation preserves motion in four-step video world models

Dual-expert diffusion improves long-horizon predictions of collisions and smooth dynamics

Randomized layer fusion improves image reconstruction and diffusion generation

Video-grounded prompt planning improves long-form text-to-video generation quality

Executable code gives generative worlds persistent rules and evolving state

Text-to-video model that learns composite physics and adapts to new dynamics

Diffusion models generate high-resolution elevation maps from low-resolution inputs guided by optical imagery

Identity preservation remains a distinct challenge for generative image models

Image-only pretraining helps build a strong open image generator

Language enables precise character and camera control in video worlds

Explicit identity-layout planning improves group images with specified people

World state registers enable consistent multi-agent video generation across views

Hybrid attention enables high-quality video generation on a single GPU

Multi-agent system creates editable 4D scenes from text prompts

Skill evolution improves agent performance in image generation workflows

An agentic framework that fills in missing context for real-world image generation

Agent framework auto-tunes video diffusion for 2x speedup

Multi-agent orchestration generates 3D scenes from single images

Multi-agent framework coordinates narrative and visual consistency for long-form video

Code-driven agents generate images via sketching and texturing stages

A multi-agent framework for coherent video storytelling via global optimization

Unified visual-generation agentic model outperforms larger closed-source models