This video was made with just a single image using Runway Gen-4. A short generated musical experiment.

The starting point was this frame from the 1996 movie "Microcosmos". From there, I generated multiple variations of the water drop

For the video I wanted to use I then use the "Use current frame" option to use the last frame of the generation as the starting point of the next using very simple prompts. Gen-4 was trained to understand really well camera movements and intention so it's very easy. "hyper fast zoom out to reveal a cinematic jungle scene"

I did that extension for 12 clips. Generated around 100 videos in total. The hit ratio is so good with Gen-4. I felt you could branch to infinite version of each scene everytime.

My son's debut single is now on Spotify (and everywhere else)! This first song anchors in country / Americana with some sweet bluegrass notes. Sound on and enjoy!

While this is very impressive, I don't understand why we go to point zero with every release. Why isn't a basic function like extend you are clearly using here not in the service?

What do you mean? Everything here was done with Gen-4 using the "Use current frame" option to extend your generation using an existing frame. All available to everyone in Runway.

🔥🔥🔥

@runwayml Cool 😎

Wow! This is great. Super slick. And nice music. Crank up the volume everybody. @grok @TheWonderFrame

Incredible how much emotion and motion can come from just one frame. Gen-4 is pure magic. Quality data is the fuel driving this transformation and with @navigate_ai, you can contribute and earn rewards with every click, scroll, and label. Built on @Base.

This is going to be a game changer 🔥

it's wild