In a few short months, the world of generative AI has evolved from chatbots like ChatGPT to photorealistic images created by Midjourney. In case you thought things were slowing down any time soon, AI-generated videos might be about to have their big moment in the spotlight.
As highlighted by a tweet from Philipp Tsipman, the founder of a marketing platform for such videos, there have been five new AI video generators launched in just the last seven days. The quality varies, but it’s clear that the technology is moving at a relentless pace. So, let’s break down the five AI video generators, which range from janky to seriously impressive.
If you’re having trouble keeping up with Video AI😅, there have been 5 state-of-the-art generative video models released *in last 7 days*: 🤯😎🧵 pic.twitter.com/cki9SMfozr
— Philipp Tsipman (@ptsi) March 26, 2023
Runway
The wait is over.
Gen-1 is now available at https://t.co/ekldoIshdw pic.twitter.com/Wm2YVOvm26
— Runway (@runwayml) March 27, 2023
Runway has been leading the pack in publicly available AI editing. Masking, motion-tracking, super slow-motion, and color grading all benefit from AI assistance. This is a subscription-based service, so it has extra incentive to push the limits.
In February, Runway started testing generative video with its Gen-1 update, which lets you restyle a video with words. A video of a walk down the street becomes an animated short that appears to be made with hand-sculpted and posed Claymation characters.
Runway released Gen-1 publicly today and is already testing Gen-2, eliminating the need for a reference video. Soon, Runway will let you create a video with a simple text prompt.
Generate videos with nothing but words. If you can say it, now you can see it.
Introducing, Text to Video. With Gen-2.
Learn more at https://t.co/PsJh664G0Q pic.twitter.com/6qEgcZ9QV4
— Runway (@runwayml) March 20, 2023
Picsart
Text2Video-Zero: Text-to-Image Diffusion Models are Zero-Shot Video Generators
abs: https://t.co/5xCsj4PNRj
github: https://t.co/BdSzlepGQG pic.twitter.com/XY4piH6j4v— AK (@_akhaliq) March 24, 2023
You might recognize Picsart as a free photo editor, but it also offers video editing. Text2Video-Zero is a Picsart research tool that uses the Stable Diffusion AI image generator to create multiple frames with enough in common to look cohesive.
This is significant progress. as early experiments at assembling AI images as video frames produced chaotic scenes where every detail constantly shifted. The effect was eye-catching, but distracted from the content.
Video-P2P
Video-P2P is an open-source project similar to Runway Gen-1. It takes video input and uses Stable Diffusion to style it with text prompts. The examples shown look good, with steady backgrounds and consistency in the altered elements.
Video-P2P: Video Editing with Cross-attention Control@Gradio demo is out on @huggingface
demo: https://t.co/dHHoiLh18v
github: https://t.co/vbq1MqwjXO pic.twitter.com/EYBCr9TkbK— AK (@_akhaliq) March 20, 2023
TemporalNet
TemporalNet is based on ControlNet, an advanced AI image generator that allows more control of the poses of figures that appear in pictures.
#TemporalNet has been published!
You can access the model to give it a try yourself here:https://t.co/0tdfmGxsVH
Curious to see how far people can push it ^^#ControlNet #AIart #stablediffusion pic.twitter.com/hd0e3MEFhG— CiaraRowles (@CiaraRowles1) March 24, 2023
The videos have some flicker, but the results are promising. ControlNet took a big step toward producing predictable scenes, and TemporalNet might provide a nice alternative method of creating videos.
Text-to-video
Text-to-video is an AI model created by Alibaba. It’s clear that this model needs a bit more work, but having more options to explore is always a good thing.
fast & longer text-to-video with 🧨 diffusers
you maybe saw fun junky text-to-video from the ModelScope's research model lately
with diffusers you can control how long the video is – and fit it on smol VRAM GPUs, including free colab. Try out here:https://t.co/uPVd9hEOtr pic.twitter.com/mIFkGnGT5V
— apolinario 🌐 (@multimodalart) March 22, 2023
Fast-paced innovation
The rapid pace of AI innovation is phenomenal. Just months ago, we were teased with glimpses of the potential of similar videos created in AI labs. In 2022, one of the first public AI video generators simply superimposed tiny AI images over stock footage.
Those “old school” methods, from a few months ago, are laughable when you see how far AI has progressed in such a short amount of time.
