GVO had the opportunity to collaborate on developing content with a beta version of Runway ML, a multi-modal AI system that can generate novel videos with text, images, or video clips. It works either by applying the composition and style of an image or text prompt to the structure of a source video (Video to Video) or using nothing but words (Text to Video).
We chose the "nothing but words" route. Leveraging the Text to Video prompts, we challenged ourselves to recreate Friday the 13th Part 3 and The Godfather movie trailers. Watching the trailers then prompting Gen-2 to create the 4 second video clips with no more than 3 attempts. The goal was to allow the AI to be creative and not box it in with our expectations. The results look and feel like a dream or a child's
interpretation of a concept.
We worked quite literally with the first one, Friday the 13th Part 3 trying to match the shots. Getting the hockey mask was super challenging and was the one time we allowed ourselves to break the 3 attempts parameter. Using the text prompt "Jason mask" got us the best result for the last clip. The AI version of this trailer is far more terrifying than the original! Below you can see the final video and a side by side version. We broke this trailer down into 52 shot descriptions. Created in 8 hours with a little over 60 generated shots to get the final result.
PLAY Friday the 13th Part 3 full trailer
PLAY Friday the 13th Part 3 side by side
Then, we had a bit more fun with the next one and got a result that is far more adorable than the original. We recreated The Godfather with dogs. On The Dogfather, we aimed to cast dogs as the characters and animate their mouths for dialogue. Created in 17 hours with a little over 100 generated shots to get the final result. All videos were edited in Adobe Premiere.
PLAY The Dogfather full trailer
PLAY The Dogfather side by side