Over the past few weeks, I’ve been experimenting with OpenAI's video generation model, Sora, to explore how deeply emotional stories can be told in just 30 seconds, with surprising results.
I created two short films using Sora, each rooted in emotional ideas. The process was creatively rich, but it also came with a learning curve—and a deeper appreciation for how less is more in this medium.
Film 1: The paper plane
In the first film, a young boy makes a paper plane and throws it out a window. We follow it as it glides through a park, where an elderly woman gently catches it on a walk. Her face lights up — not with surprise, but recognition.
The message is simple:
The spirit of wonder never ages.
Content generated in SORA–Music from Pixabay.
Film 2: The pancake memory
The second film follows a more reflective tone. An elderly man sits silently in a nursing home, lost in thought. We enter his memory: a warm morning, his young daughter, making pancakes. That memory returns to him just as, in real life, his now-grown daughter visits him at the nursing home.
The final message:
The ordinary moments we carry forever.
Content generated in SORA–Music from Pixabay.
What I learned through this process
It takes time.
We distilled each of these 30-second videos from over 40 individual video generations. You can’t rush it, but the emotional payoff is real.
Less detail, more clarity.
Sora thrives when you keep each prompt focused and clean. Overloading the scene with motion or too many objects can make it muddy.
For example, I spent a long time trying to generate a clean shot of a boy throwing a paper plane out the window, but it was too complex: arm motion, timing, camera angle, and the plane’s path. It kept falling apart visually. What worked instead?
A shadow on the wall — the boy’s silhouette, arm raised, and in slow motion.
Look for the moment inside the shot.
Each 5-second clip might only have 1–2 seconds of usable gold. That’s okay. Find the fragment that says something and cut the rest.
Every frame must serve the message.
Ask yourself: What am I trying to say? Then make every scene a quiet step toward that feeling.
Stillness is powerful.
Don’t be afraid of moments with little to no motion. The most emotional beats in these films were simple still frames: a hand, a plate, and a glance.
To wrap things up, testing Sora's video output has been an exciting glimpse into the future of generative AI and storytelling. The creative possibilities it opens up are unlike anything we've seen before.
If you’re as fascinated by this space as I am, I’ve got more experiments, insights, and results waiting for you. Check out some of my Gen AI explorations articles below:
The euphoria of discovering Bolt.new
Exploring visual development with ChatGPT’s image generator
Behind the scenes: How I created this AI-generated video
How AI-Generated images transformed my creative workflow: A year in review
