How to make your own movie using a toilet lid and a neural network. Cheng, Artificial Intelligence

Neural networks allow you to engage in creativity using a minimum of available tools. Director Karen Cheng shows by her own example that anyone can now make a film.

Karen X. Cheng is a director and creator of videos for famous brands. To achieve interesting special effects in her work, she often uses low-budget means (and willingly shares these secrets on social networks).

Lately, Chang has been trying to work with neural networks. She recently showed how to use the GEN-1, a toilet lid, a stack of books and a computer mouse to create a short movie.

Previously, Cheng used the DALL-E 2 neural network to create a music video for Nina Simone’s song “Feeling Good,” and she also tried to generate a cover with a female astronaut for the publication Cosmopolitan.

And although all these experiments look like something simple and not labor-intensive, in fact, working with a neural network is not as easy as it seems. For example, on the sole of the astronaut from the cover it is written that it took 20 seconds to create the picture.

In fact, this is deceit. The director spent at least 100 hours selecting a successful request for the AI. Cosmopolitan employees who tried to repeat the success could not boast of a similar result. Their “girls on Mars” were not suitable for either the cover or publication in the magazine.

Neural networks can make the work of designers and artists easier, but it is unlikely that it will be possible to completely replace people in the creative profession with artificial intelligence.