Dream Oddly

You don't need to know the jargon to make a decent video What usually trips people up isn't a lack of ideas — it's not knowing the lingo. You've got a picture in your head, but no idea how to turn it into a script, a shot list, all that. Let alone hand it to someone and have them shoot what you meant. That's the gap this project fills. You describe what you want in plain words — even one sentence — and it writes the script first, then breaks the script into a storyboard. The professional vocabulary is the part AI handles for you. Say you tell it, "I want a 30-second clip of pour-over coffee, slow, kind of soothing." It takes it from there: how many shots, what each one shows, how to pace the framing and rhythm. Generating it is one thing; keeping it steady is the harder part. Actors, props, and locations are managed as shared resources, so the same person, the same cup, the same kitchen still look right when the shot changes. That cuts down a lot of wasted output — and reshoots are where the money goes. Underneath, the models are sorted by what they do: text, image, video. At each step you can fire off several at once and get a few versions back to pick from. Let one idea take a few different paths and your odds of landing a good result go up, without dragging the whole thing down. This site is where all of that comes together.

Built With

• ffmpeg
• go
• openai
• postgresql
• react
• seedance
• typescript
• vite
• whisper