Runway, the leading artificial intelligence video generator, has just upgraded its Turbo model, adding the ability to create vertical videos for sites like YouTube Shorts, Instagram Reels, and TikTok, as well as video-to-video, so you can add effects using AI to your existing clips.
Video-to-video launched two weeks ago in the standard Runway Gen-3 Alpha model, which is a lot slower to generate a clip but is slightly better in terms of motion and visual effects.
The benefit of Turbo is its speed. In my own testing, I was able to get it to create a clip almost as fast as the duration of the video. So, in just over 10 seconds, I had 10 seconds of video.
We are still waiting for text-to-video to come to Turbo, but the new update also brought with it several new features, as well as features previously exclusive to the slower model.
What’s new in Runway Turbo
Runway is gradually adding features we had in Gen-2, as well as new concepts and approaches to generating video made possible thanks to changes in how Gen-3 works compared to the earlier generation of models.
With the most recent update, we get clip extensions, lip-syncing, and a new "edit time" feature that lets you change the speed of a clip, reverse it or trim it down to just the best bits.
The new ability to create vertical video content, perfect for social media, is exclusive to the Turbo model. For some reason, this hasn’t been enabled for Gen-3 Alpha.
Overall, the new features give you much more control over the video Runway generates, including improving the speed of a clip and allowing you to start with either a video or an image.
Video-to-video is a great way to generate exactly what you are looking for as it adds a layer or style on top of your work—for example, turning yourself into a 3D character.
How well do the new features work?
Today we're excited to share updates to Gen-3 Alpha Turbo that bring more control and the ability to generate vertical videos.Learn more below.(1/4) pic.twitter.com/EucNJmMc3xSeptember 27, 2024
I found video-to-video in Turbo a little bit hit-and-miss. I gave it a short video of me moving a toy card and it turned half my hand into a tiger, but it isn’t terrible and with better prompting you can get some seriously impressive results.
My favorite new feature is the time-related updates. Sometimes you generate a clip and it appears to be going backward, now you can reverse it without loading a video editor.
The other impressive update is extensions. In the normal mode, these can take a while, and as they can vary more than you’d like from the original it can be a pain trying to get exactly what you'd expect.
Having rapid iterations makes a difference as you can do 10 different extensions with Turbo in the time it takes to do one with Alpha.
