Sora Text to Video: Playing with AI Like It’s 2049
Oh, dear readers, I am excited!
I have been experimenting with a new tool called Sora. I’ve been using Stable Diffusion since the moment it leaked, and it’s only been getting better and better. But now, with the addition of text-to-video… it’s just… insane. And fun!
I finally had some downtime to try the new tools on the block that have been making waves, at least with text.
What really gets me about this isn’t just that it’s AI generating video. We’ve seen generative AI for images, we’ve seen AI-assisted video editing. But this? This feels different.
It’s not just about typing words to generate a clip, it’s about describing a moment in time. The way we craft prompts for Stable Diffusion to generate an image is evolving. Just like how a well-structured prompt can yield a masterpiece in MidJourney or Stable Diffusion, the same principle applies here. But instead of an image, we’re manifesting a moment in time itself.
Think about it, when you describe a video, you’re not just describing a single static frame. You’re capturing motion, atmosphere, intent. The slight hesitation before someone speaks. The way light shifts as the sun sets. The tension of an approaching storm, not just in how the clouds look, but how the wind feels.
It’s like prompting a dream.
I ran a few tests, and wow. Some were rough, some were uncanny, and some were downright magical. AI is still learning, still refining. But we’re at the edge of something wild here. The kind of game-changing moment that hits you out of nowhere—like spit from a llama. You knew it was possible, but when it happens, it’s still surprising.
Here’s an example of what I generated:
This is just the beginning. The same way we adapted our thinking to prompt-based image generation, we’ll soon be thinking in cinematic moments. We’re not just describing a scene anymore, we’re directing it.
And that? That’s cool.
-Rob