MIT AI Filmmaking Hackathon
I recently joined the MIT AI Filmmaking Hackathon — a wild, immersive event that celebrated the art of movement and storytelling under the theme “Dance in Life.” The experience was intense, collaborative, and full of unexpected turns.
I kicked things off by diving into a film idea I’d been toying with for a while: a story about a 55-year-old dancer, shrouded in mystery, whose life held as many secrets as graceful movements. I spent an entire day piecing together a 45-second sequence by combining multiple 5-second clips. I created these snippets using the OpenArt tool, and to add a bit of extra visual punch, I generated some frames with Midjourney for image-to-video transitions.
Connecting those 5-second clips was a tedious process. I used ffmpeg to extract the last frame from each clip as a starting point for generating the continuation. Each clip took 3–4 minutes to generate, and most of the time the result was weird or off—about 9 times out of 10. Even though I thought I’d continue and finish my original idea, the long, laborious process left me completely drained. I’m definitely taking a break from AI filmmaking tools for a few weeks.
The hackathon’s final day brought a spontaneous twist. I met my good friend Nemo, who introduced me to a newly forming team. I pitched my film idea, and they were instantly interested. In the excitement of collaboration, my original plot was significantly reworked—a change I wasn’t entirely comfortable with. Still, the fun of working with a creative group outweighed my reservations, and I decided to roll with it.
We were a five-person team. I had completed the first 30% of the film, and for the remainder, we each took on distinct roles. I generated a sequence featuring an old man watching ballet on TV with his cat, who becomes upset by the loud noise from the dancers upstairs. I even animated dust falling from the ceiling—amusingly, the TV characters reacted by running away in fear, while the man emerged both furious and comical. I also created all the sound effects, from door openings and footsteps to birds chirping, using ElevenLabs, and produced a few voice-overs with the same tool.
Todd handled the final cut, expertly piecing together clips and sounds. He even helped with another idea by quickly replacing a picture frame in one scene with a character I wanted, thanks to his Photoshop skills. Brayan’s final script idea won the vote, and he used his AI tool Storyfave to generate vivid, high-resolution images for the memory flashbacks. Kamsai, a co-founder with Brayan, designed a very cute mouse character for the final scene, contributed to the memory sequences, and actively participated in every discussion. Lastly, Bhuvan created the remaining dancing scene shots and polished the final script.
One thing I realized during this journey is that making films is more accessible than ever. In the past, you needed an expensive camera and a full crew. Now, even an iPhone can get you started—and with AI, a director might not even need actors, a cameraman, lighting, or sound. It’s essentially a one-person job. If you have a good idea and the taste for a challenge, you can make it happen!
We wrapped up our project and submitted the film just before midnight. Here is your final submission:
While we didn’t win—something that didn’t surprise me given the caliber of other projects—the entire journey was a blast. I was completely in the flow for the entire two days, and even with the unexpected changes, the hackathon was an unforgettable creative sprint.