Video Generation

Latest: AB-Attention Training-Free Method Matches Dense Attention PSNR at High Sparsity on Wan2.1-T2V-14B · huggingface.co, Oct 30, 2026

Thème Mots-clés suivis : video generation, text-to-video, video ai

5 AI Weekly issues and 88 live stories from the last 30 days mention Video Generation.

5 mentions au total 1/20 numéros récents -100% vs référence pic 1/numéro suivi depuis 2026-01-20

Tendance des mentions — 20 derniers numéros

Actus en direct (88)

AB-Attention Training-Free Method Matches Dense Attention PSNR at High Sparsity on Wan2.1-T2V-14B

huggingface.co · 2026-10-30 · Score : 19

Pika Opens 33B-Parameter H3-Flex Video Model With Six-Step Audio-Video Generation Across 11 Aspect Ratios

huggingface.co · 2026-10-29 · Score : 20

SenseTime Open-Sources 27B SenseNova-6.8-Code2Video That Programs Each Shot Instead of Generating Pixels

huggingface.co · 2026-10-27 · Score : 24

VibeVideoBench Caps Top Code-Driven Video Agents at 76.3% Core-Requirement Pass Across 239 Tasks

huggingface.co · 2026-10-27 · Score : 21

ALERTE Hy-Nomas Opens 8B Interactive World Model With Free-Form Text and Continuous Camera Control

huggingface.co · 2026-10-28 · Score : 28

LociArt Introduces 100K Training Pairs and Polar Style-Aware Masks for Mask-Guided Local Video Stylization

huggingface.co · 2026-10-28 · Score : 16

HKUST and KlingAI VMem-Bench Finds More Visual Memory in Long-Video Gen Hurts Precision as Recall Lifts

huggingface.co · 2026-10-27 · Score : 18

AI Film Studio Utopai Raises $60M at $1B With 20+ Movies and Series Due in 2027

theinformation.com · 2026-10-23 · Score : 21

AnyGameRenderer Turns Programmer-Art Game Blockouts Into Final Video With Entity Consistency

huggingface.co · 2026-10-22 · Score : 18

Nvidia's JCA Co-Generates Replayable 3D Motion and Video From Arbitrary Character Rigs in One Pass

huggingface.co · 2026-10-20 · Score : 19

'Execution Before Memory' Paper: Video World Models Fail at Execution, Not Memory

huggingface.co · 2026-10-19 · Score : 22

ComfyUI v0.40.0 Ships Assets System by Default, Qwen 3.5 Low-VRAM Boost and Custom-Node Governance

github.com · 2026-10-13 · Score : 18

WorldGuide Video World Model Clears 47.69% of VideoCraft-Bench via Closed-Loop Planner-Executor Loop

huggingface.co · 2026-10-09 · Score : 22

OuroWorld Mask-Free Framework Turns Static 3D Gaussian Scenes Into Looping 3D Cinemagraphs

huggingface.co · 2026-10-08 · Score : 17

SGF+ Decouples Context-Writing and Denoising Gradients, Enables 24-Hour Continuous Autoregressive Video From 5s Clips

huggingface.co · 2026-10-08 · Score : 19

GRACE Latent-Compressed Video Diffusion Runs 11.1x Faster at 480x832 With No VBench Quality Drop

huggingface.co · 2026-10-07 · Score : 21

CtrlCache Paper Cuts Interactive Video World Model Compute 1.21–1.41x Using Control-Aware Caching

huggingface.co · 2026-10-07 · Score : 22

HF Paper Ego2Act Benchmarks Video Models on 110 Goal-Directed Manipulation Tasks, Finds Dropped Steps and Physics Breaks

huggingface.co · 2026-10-02 · Score : 17

HF Paper VTR-Bench Finds Best Video Gen Model Hits 0.25 Word Error Rate Rendering On-Screen Text

arxiv.org · 2026-10-01 · Score : 16

Stability AI's 4Director Lets Video World Models Take Rigid 3D Scene Control

huggingface.co · 2026-10-01 · Score : 19

HF paper: SoL-Refiner claims one-step 'speed-of-light' refinement for high-resolution video diffusion

huggingface.co · 2026-09-30 · Score : 15

Kuaishou's Kling Ships Kling 4.0 Flash to Annual Subscribers, 30-Second 4K/HDR AI Video Generation

bloomberg.com · 2026-09-28 · Score : 22

Google Ships Gemini 3.8 Live With Avatar for Enterprise, Lip-Syncs in 97 Languages

blog.google · 2026-09-24 · Score : 27

Multiples Leads $150M Round in Brahma AI at $2B Valuation, DNEG Founder Malhotra Joins Cap Table

cnbc.com · 2026-09-24 · Score : 22

Nvidia/Bar-Ilan Paper Adds KL Regularizer That Cuts Audio-Video Desynchronization From 0.804 to 0.752 in Joint Diffusion

huggingface.co · 2026-09-24 · Score : 17

Zing-0.5 Ships 5B Playable-World Model at 24 FPS With Joint Keyboard and Text Control for ~$0.009 per Stream-Minute

arxiv.org · 2026-09-17 · Score : 24

PhysStream Ships Autoregressive Physics-Grounded Video Model With Interactive Velocity Control

huggingface.co · 2026-09-15 · Score : 21

LynnReal-Omni Paper Unifies Video Generation With 32B Multimodal DiT, Ships 27B Flash at 377ms per 540p Clip

arxiv.org · 2026-09-14 · Score : 18

Westlake AGI Lab's 'World in World' Reframes Video World Models as Ongoing Exploration Within a Video

huggingface.co · 2026-09-10 · Score : 16

Programmable World Model Paper Decouples State From Rendering, Hits 98% State Accuracy on CombatStateBench

arxiv.org · 2026-09-09 · Score : 18

Katzenberg and Ex-Sora Head Bill Peebles Launch AI Video Startup for Filmmakers

theinformation.com · 2026-09-09 · Score : 20

Adobe Puts Firefly, Google Veo, Runway, Kling and Luma in Premiere's Timeline for IBC 2026

blog.adobe.com · 2026-09-08 · Score : 22

Mask Forcing Paper Cuts Over-Saturation in AR Video Diffusion Distillation, Wins 79-83% Human Vote

huggingface.co · 2026-09-08 · Score : 19

Runway ARR Hits $200M as Revenue Doubles in Five Months, Targets $350M by Year-End

bloomberg.com · 2026-09-08 · Score : 21

ByteDance's Zhang Yiming Personally Leads Real-Time Spatial-Video World Model, Launch Possible Next Month

bloomberg.com · 2026-09-07 · Score : 23

OpenVDN Drops VDN-MiniMax-H3, Hybrid-Attention Video Model Renders 14.4s at 768p in 11 Seconds

huggingface.co · 2026-09-06 · Score : 24

WorldReward Paper Unifies Action-Consistency and Visual-Quality Reward for Camera-Conditioned World Models

huggingface.co · 2026-09-04 · Score : 18

SolarWM Ships Fully Open Long-Horizon Video World Model Trained on 1.43M Clips

huggingface.co · 2026-09-02 · Score : 24

DramaChain Bench Pipelines 60 AI Short Dramas, Finds Upstream Defects Cost Up to 0.83 Points

huggingface.co · 2026-09-01 · Score : 19

Tencent-Backed H3-World Turns MiniMax-H3's 33B Video Model Into a Language-Controlled Interactive World Model

huggingface.co · 2026-09-02 · Score : 21

DreamX-Creator Paper Pitches Native Audio-Video Generation at 2K Resolution

huggingface.co · 2026-09-01 · Score : 18

Alibaba Launches Wan3.0 Video Model, Generates 30-Second Clips From PDFs, Slides and Spreadsheets

reuters.com · 2026-08-24 · Score : 22

Guardian: Out-of-Work Hollywood Creatives Are Now Training the AI Systems Taking Their Jobs

theguardian.com · 2026-08-22 · Score : 22

Wired: Rogue Studio Ships Rogue 1.0, Positioning Uncensored AI Video as the 'HBO of AI'

wired.com · 2026-08-17 · Score : 18

Guardian: AI Film Startups Are Setting Up Studios in Hollywood, Blending US and Chinese Models

theguardian.com · 2026-08-16 · Score : 16

Redis Creator antirez Ships h3.c, a Native Metal Runtime for MiniMax H3 Video Generation on Apple Silicon

github.com · 2026-08-11 · Score : 22

Zhejiang and Yale's Sci-VBench Finds Video Generators Cluster on Realism but Split on Scientific Correctness

huggingface.co · 2026-08-10 · Score : 15

JD's JoyAI-Video-Edit runs open-ended 720p video editing at 30 FPS on a single Nvidia B200

huggingface.co · 2026-08-04 · Score : 19

Kling and Peking U's 'Motion Beyond Morphology' Transfers Video Motion Across Object Categories

huggingface.co · 2026-08-03 · Score : 20

ByteDance Ships Seedance 2.5 With 30-Second 4K Video, Doubling Prior Length Cap

bloomberg.com · 2026-07-31 · Score : 18

ShadowDancer Trains Video World Models on 'Shadow Pairs' of Same Dynamics, 86% Average Win Rate vs Olaf-World, Yume-1.5 and LingBot-World

huggingface.co · 2026-07-30 · Score : 20

NVIDIA and Weizmann Ship Parallel Decoding Distillation, Predicts Multiple Denoising Steps in One Pass

huggingface.co · 2026-07-28 · Score : 24

Adobe Research and JHU Ship Wonder, a Camera-Controllable Video World Model Running 16 FPS Over Minute-Long Rollouts

huggingface.co · 2026-07-29 · Score : 20

Alibaba and Beijing Film Academy Ship FilmBench, Cinematic Video Benchmark With 1,169 Award-Film Prompts

huggingface.co · 2026-07-27 · Score : 17

NVLabs Ships Sol-Attn, Training-Free Sparse Attention That Speeds Wan 2.1 and HunyuanVideo Inference 2-5x

huggingface.co · 2026-07-27 · Score : 19

Nvidia's SANA-Video 2.0 Mixes Linear and Softmax Attention, Claims 120x Speedup Over Wan 2.2 on a Single H100

arxiv.org · 2026-07-23 · Score : 21

Runway Ships Media Router That Auto-Picks Image, Video and Audio Models by Quality, Speed or Cost

techcrunch.com · 2026-07-23 · Score : 22

AlayaRenderer-Flash Turns Diffusion World Renderer Into Real-Time Game Engine, Hits 31.5 FPS From 0.56 FPS

huggingface.co · 2026-07-21 · Score : 25

ABot-World-0 Ships Real-Time Interactive World Model Running 720p at 16 FPS on a Single RTX 5090

huggingface.co · 2026-07-21 · Score : 22

ByteDance Ships FlowMimic, Mask-Free Video-Editing Model That Treats Images as Special-Case Videos

huggingface.co · 2026-07-20 · Score : 17

Alibaba-Backed Video Generation Startup PixVerse Closes $439M Series C Extension at $2B+ Valuation — 150M Registered Users, 15M MAU, Positioned to Fill Void Left by Sora Shutdown

techcrunch.com · 2026-07-13 · Score : 25

NYCU Paper 'LongE2V': Video-Diffusion Model Unifies Long-Horizon Event-Camera Reconstruction, Prediction, and Frame Interpolation — Hits Trending on Hugging Face at 19 Upvotes

huggingface.co · 2026-07-09 · Score : 15

EPFL 'NEvo' Paper Hits HN: Neural-Guided Evolutionary Video Synthesis Generates AI Videos That Drive Higher Brain-Region Activation Than Handcrafted Localizer Clips

arxiv.org · 2026-07-09 · Score : 16

LingBot-World-Infinity Paper: Open-Source Causal World Model Sustains 720p/60fps Interactive Worlds for Hour-Long Sessions With Combat, Spellcasting, Weather Control and a Pilot/Director VLM Harness

huggingface.co · 2026-07-08 · Score : 22

AlayaWorld: Alibaba-Backed Team Releases Open-Source Full-Stack Framework for Long-Horizon, Playable Video World Generation at 720p/24fps

huggingface.co · 2026-07-07 · Score : 18

HF Paper 'MV-Forcing': 4D-Grounded Spatio-Temporal Self-Forcing Extends Long Multi-View Video Generation Beyond Prior State of the Art

huggingface.co · 2026-07-06 · Score : 16

MIRA: First 5B Multiplayer Interactive World Model Runs Four-Player Rocket League at 20 FPS on a Single B200, Stable for 5+ Minutes

arxiv.org · 2026-07-06 · Score : 18

LA Times: ByteDance Quietly Courting Hollywood Filmmakers and Executives With Seedance Video AI — Low Pricing, Striking Realism, and Timeline-Based Prompting Win Beachhead After 2.0 Cease-and-Desist Wave

latimes.com · 2026-07-03 · Score : 22

HF Paper: 'Representation Distribution Matching' Cuts One-Step Visual Generation Gap With Multi-Step Diffusion

huggingface.co · 2026-07-02 · Score : 15

Kuaishou's Kling AI Raises $2B at $15B Pre-Money, Round Could Extend to $3B — General Atlantic Leads First External Financing Ahead of Hong Kong IPO

bloomberg.com · 2026-07-02 · Score : 23

fal Releases LTX-2.3-3DREAL-LoRA — Photoreal 3D-Consistent Video LoRA for LTX 2.3 Base Model, Ships With Camera-Path Controls

huggingface.co · 2026-06-30 · Score : 15

Goku: USTC + Tencent Hunyuan Release 2M-Pair Universal Video-Editing Dataset and Goku-Edit Model, Bench Beats LucyEdit and Rivals Runway Gen-4 on Complex Multi-Task Edits

huggingface.co · 2026-06-30 · Score : 17

MemLearner (HKU, Fudan, Kuaishou): Learned Context Memory Queries Fix Scene Consistency in Video World Models Under Occlusion

huggingface.co · 2026-06-30 · Score : 17

ALERTE Google Ships Gemini Omni Flash and Nano Banana 2 Lite — $0.10/sec Conversational Video Editing and $0.034 per 1K Images in 4 Seconds

blog.google · 2026-06-30 · Score : 29

DreamForge AI Lab (Kazakhstan) Releases DF-World 0.1 Preview: Single-GPU Streaming World Model With Mid-Stream Reprompting and Dual First/Third-Person Control on Just 64 Hours of Adaptation Data

huggingface.co · 2026-06-29 · Score : 16

AI Video Startup Higgsfield in Talks to Raise $300M–$500M at $5B Pre-Money — Nearly 4× Its January Round Valuation as Run-Rate Hits ~$500M

theinformation.com · 2026-06-29 · Score : 22

Animation's AI Reckoning: Filmmakers Say They Can Make Movies for 90% Less

bloomberg.com · 2026-06-28 · Score : 22

LISA: Likelihood Score Alignment Regularization Accelerates Training and Improves Quality Across Controllable Image and Video Generation Tasks

huggingface.co · 2026-06-25 · Score : 15

Adobe Acquires AI Video and Image Enhancement Startup Topaz Labs to Bolster Firefly Against Canva and Blackmagic

techcrunch.com · 2026-06-25 · Score : 24

TryOnCrafter: First Camera-Controllable Video Virtual Try-On via Renderable 4D Proxy From Chinese Academy of Sciences and Alibaba

huggingface.co · 2026-06-25 · Score : 15

MVTrack4Gen: Multi-View Point Tracking as Geometric Supervision Achieves SOTA Consistency on 4D Video Generation Without 3D Reconstruction at Inference Time

huggingface.co · 2026-06-25 · Score : 16

DomainShuttle: HKUST Achieves High Subject Fidelity and Cross-Domain Flexibility for Open-Domain Subject-Driven Text-to-Video Generation via Domain-MoT and DualRoPE

huggingface.co · 2026-06-24 · Score : 16

Vera: Caltech and Netflix Release Layered Diffusion Framework for Content-Preserving Video Editing via Composite Rather Than Regeneration

huggingface.co · 2026-06-22 · Score : 16

Snap Spins Off AI Video Team Into New Company Dotmo, CTO Bobby Murphy Leads as Personal Investor

techcrunch.com · 2026-06-18 · Score : 20

Kuaishou's Kling AI Seeks $2 Billion at $18 Billion Valuation From General Atlantic in Pre-IPO Round

bloomberg.com · 2026-06-17 · Score : 27

Asian Streaming Services Go Big on AI-Generated Microdramas as ReelShort CEO Says Technology 'Completely Changes' Competitive Landscape

asia.nikkei.com · 2026-06-17 · Score : 15

Decart Launches Oasis 3 World Model for AV Simulation — Hours of Photorealistic Driving at $0.02/Second With Physics Caveats

techcrunch.com · 2026-06-10 · Score : 20

Reactor Emerges From Stealth With $59M Lightspeed-Led Series A to Build Real-Time AI World Model Infrastructure — Katzenberg's WndrCo Joins Board

variety.com · 2026-05-28 · Score : 18

Numéros de la newsletter (5)

#524 What AI models are actually coming in the next six months?

2026-08-20 · 1 mentions

Generative AI: DeepSeek's 75% cut rewrites the cost math

2026-05-26 · 1 mentions

Machine Learning: a reasoning model just cracked Erdős

2026-05-26 · 1 mentions

World Models: Google's Genie now simulates real streets, not just dreams

2026-05-21 · 1 mentions

#460 Agentic AI takes over as Gartner sees $2.5T spending

2026-01-20 · 1 mentions