A Visual History of AI Film & Video
2015-2021
BeforetheDawn
Before text-to-video models existed, a few pioneers hacked together ways to make AI generate moving images. Frame by frame, zoom by zoom, prompt by prompt. These experiments laid the aesthetic and technical foundation for everything that followed.
ThisPersonDoesNotExist.com
Phillip Wang / NVIDIA StyleGAN
Video Enhance AI v1
Topaz Labs
DALL-E Announced
OpenAI
VQGAN+CLIP Goes Viral
Katherine Crowson & Ryan Murdock
Disco Diffusion v1
Somnai & Katherine Crowson
2022
TheDawn
Research labs crack text-to-video. The outputs are blurry, short, and uncanny. Nobody takes it seriously yet, but early creators had a lot of fun experimenting.
Midjourney V1
Midjourney
CogVideo
Tsinghua University / BAAI
Make-A-Video
Meta AI
Deforum Stable Diffusion
Deforum Community
Imagen Video
Google Brain
Topaz Video AI v3
Topaz Labs
Phenaki
Google Research
Midjourney V4
Midjourney
2023
TheCambrianExplosion
Consumer tools arrive. Runway ships Gen-1 and Gen-2. ModelScope goes viral. The first AI film contests launch. The community forms.
Runway Gen-1
Runway
Runway Gen-2
Runway
Midjourney V5
Midjourney
Stable Video Diffusion
Stability AI
Magnific AI
Javi López & Emilio Nicolás Riquelme.
Pika 1.0
Pika Labs
Midjourney V6
Midjourney
2024
TheGreatConvergence
Quality explodes. Big tech companies enter the race. Filmmakers get serious. The model wars begin.
Sora Preview
OpenAI
Late Night with the Devil Sparks a Backlash Over Three AI Stills
Cameron & Colin Cairnes / IFC Films
Veo 1
Google DeepMind
Luma Dream Machine
Luma Labs
Kling 1.0
Kuaishou
Runway Gen-3 Alpha
Runway
Hailuo Video-01
MiniMax
Kling 1.5
Kuaishou
Here Uses Metaphysic for Generative De-Aging
Robert Zemeckis / Metaphysic
Coca-Cola AI Holiday Ad
Coca-Cola / Silverside AI / Secret Level / Wild Card
Veo 2
Google DeepMind
Sora Public Launch
OpenAI
Kling 1.6
Kuaishou
2025
TheStudioEra
AI films go mainstream. IMAX screenings at Lincoln Center. Feature films get greenlit. Generative animation becomes its own art form.
Luma Ray2
Luma Labs
Wan 2.1
Alibaba / Tongyi Lab
Runway Gen-4
Runway
Midjourney V7
Midjourney
Kling 2.0
Kuaishou
Wan 2.1-VACE
Alibaba / Tongyi Lab
Veo 3
Google DeepMind
Kling 2.1
Kuaishou
ANCESTRA Premieres at Tribeca
Eliza McNitt / Primordial Soup
Seedance 1.0
ByteDance
Midjourney V1 Video Model
Midjourney
Hailuo 02
MiniMax
Wan 2.2
Alibaba / Tongyi Lab
Grok Imagine
xAI
Nano Banana (Gemini 2.5 Flash Image)
Google DeepMind
Sora 2
OpenAI
Luma Ray3
Luma Labs
Kling 2.5 Turbo
Kuaishou
Veo 3.1
Google DeepMind
Hailuo 2.3
MiniMax
LTX-2
Lightricks
Nano Banana Pro (Gemini 3 Pro Image)
Google DeepMind
Kling 2.6
Kuaishou
Seedance 1.5 Pro
ByteDance
Runway Gen-4.5
Runway
Wan 2.6
Alibaba / Tongyi Lab
Luma Ray3 Modify
Luma Labs
2026
TheShakeout
The field consolidates. Sora is shut down, Wan goes closed, Adobe buys Topaz, and a government switches off a frontier model overnight. The majors move in — Netflix, Disney, Amazon, Lionsgate, George Lucas — as the first AI feature reaches US theaters. Meanwhile creators stop making clips and start making shows: 30-second single-pass models with native audio turn viral shorts into serialized, AI-native series.
Luma Ray3.14
Luma Labs
Seedance 2.0
ByteDance
Kling 3.0
Kuaishou
Magnific Video Upscaler
Magnific AI
Nano Banana 2 (Gemini 3.1 Flash Image)
Google DeepMind
Sora Shutdown Announced
OpenAI
HappyHorse 1.0 Limited Beta
Alibaba Cloud
Krea 2
Krea
PINOC
ViggleAI
Seedance 2.5
ByteDance
Seedance 2.0 Gets Native 4K
ByteDance
Seed Audio 1.0
ByteDance / BytePlus
FLUX 3
Black Forest Labs
MiniMax-H3
MiniMax
LTX-2.5
LTX (spun out of Lightricks)
MiniMax-Music3
MiniMax
Wan 3.0
Alibaba / Tongyi Lab
