We like to get a little experimental with our 48 Hour Film Project shorts. There’s nothing quite like testing out new techniques and tricks in the heat of a timed filmmaking project, and for 2023’s Washington DC 48 Hour Film Project I decided to put the generative AI app Midjourney to the task of helping us create several scenes for our short, INFORMATIONAL HUMAN CENTRIC VACATION FILM.
As an image generator, the way Midjourney works is that when prompted with a few lines of text, it uses machine learning to create a new image, drawn from patterns learned by sorting through a dataset of many, many other images, and Vox made a great video explaining how this all works much better than I can, available here:
https://www.youtube.com/watch?v=SVcsDDABEkM
The first time I used it, it felt almost like magic (and much like magic, it has a bright side and a dark side, something I’ll be talking about in another post, soon). In the interest of transparency in the face of a new technology, I wanted to share the prompts we used for each of the Midjourney generated images included in our short:
Side note, the first time I did this prompt, I accidentally misspelled it as ‘Buffett’, which led to this hilarious image of Warren Buffett at a Buffet:
If you’re interested in a more in depth look at the process of generating each of these images, and the practical aspects of how we integrated them into the short, drop a comment below and I may make a behind-the-scenes video of how we used Midjourney in the edit in more detail!
And if you haven’t seen the short, here it is!