It’s been a while since OpenAI launched Sora to the public (available to ChatGPT Plus users, by the way), and I finally got a chance to experiment with it. Spoiler alert: it’s incredible!
You can check it out here: Sora Library.
I spent some time playing around with it, and here’s one of the results I came up with:
This is a scene from a puzzle game we’ve been working on. The game has a unique twist — it combines classic puzzle-solving mechanics with a meta-layer where players can fix up different parts of a room and earn stars as rewards. Think of it as a mix of puzzles and home improvement, similar to some of the trending games out there right now.
So, here’s how I experimented with Sora: I kept it simple — just straightforward prompts, nothing too fancy or overly detailed.
Prompt: animate the girl and the cat. move the camera and keep the rest of the scene static (attached the image as the source)
Result:
Prompt: animate this scene (attached the image as the source)
As you can see, it’s pretty unbelievable how Sora has managed to take the entire image, turn it into a three-dimensional world, and move the character while keeping everything spot-on with the static image. The attention to detail, from the shadows to the textures, is absolutely flawless.
Merging the two videos and adding some sound effects turned it into a solid video that’s perfect for use in marketing materials or as an engaging landing video before the game kicks off.
End Result: 🥁🎶
I’m honestly fascinated by the potential of Sora. The ability to use 2D artwork as the source opens up a whole new world of possibilities!