00%
=
alt
  • Stable Diffusion
  • Disco Diffusion Warp
  • Deforum Stable Diffusion
  • Deforum Warpfusion
  • Topaz AI + Chronos
  • Dreambooth

This first-of-its-kind reel was created with state of the art, month-young technologies. These methods demonstrate incredible power and potential, and they will continue to rapidly improve.

Fresh after a month long deep dive into Stable Diffusion and Midjourney, I started seeing the appearance of experimental colab notebooks that could string diffused images into videos. After painstaking attempts to understand the formulas and dials, I slowly learned how to steer the beast. With proper whispering in the ear of the AI, results I was starting to get were unlike anything I had ever seen. Visual ideas and transitions occur in a ways that are seamless, counter-intuitive, hypnotic, literally unimaginable. Because of this, and the idea that any combination of any style can be synthesized to produce AI motion graphics, 'god effect' came to mind.

Eagle Eye Capabilities

01

Leverage of input videos: Simple, even low budget videos can be leveraged to create insane graphic output. For instance, a video of a male skateboarder gliding down a street can be synthesized into of a female astronaut gliding through a highly detailed alien landscape.

02

Character transformations: After accurately training the model, a character from a video can be transformed into any other character. A video of a dancing man can be transformed into a video of a dancing Tom Cruise, or any specific character, with convincing accuracy.

03

Limitless access any style or combination of styles: Using stable diffusion with the guidance of Lexica, virtually any style on earth can be achieved, down to very specific results. "An anime hulk with emo makeup" is a result that can be achieved through the magic of AI.

04

Entire worlds within: As shown in the reel, richly endowed worlds can be created in motion with stable diffusion. These worlds are starkly detailed, self consistent and incredibly rich in content. Interestingly they also have self similarity, so if one were to zoom into a world of "street artwork", the AI would produce infinite non-repeating graphics.

05

!Dream transformationsAs shown in the reel, endless worlds can be created out of !dream prompts

Expect massive improvements to the emerging field of AI motion graphics as the models rapidly improve. Much more to come in a short amount of time.