This video generator powered by artificial intelligence takes text and turns it into a 200px x 200px video that visualizes your description.

This online version of the generator is roughly based off of the Colab Notebook written by Hillel Wayne.

The original technique was first surfaced in this notebook by Katherine Crowson and involves creating images using VQGAN, a generative adversarial network, and CLIP, a neural network centered around visual learning recently released by Open AI.

Because the GAN uses a unique seed with each run, no two videos will ever be the same even for the same text input. Having trouble thinking of something to generate? Try a simple, natural scene first and then choose whatever you like after.

Have an awesome video to add to the gallery? Please contact us or tweet at @kapwingapp to let us know!