I give your ai the scene and the character and tell it to animate the scene to the song. It fail every time because it tries to recreate the scene and character. Even when you tell it not to do it. It can’t follow instructions and therefore is unless to use for my work. If you want people to use this for MV you need to fix it.
Have you ever received a response from Tunee ? I have been waiting forever and I never have had resolution on any topic.
Sorry you’re running into this — it’s a real frustration when the output ignores your existing assets and rebuilds them from scratch.
A couple of things worth trying in the meantime:
- In Motion Control mode, uploading a reference video of the character performing the action gives the model a direct motion target rather than relying on text instructions alone.
- In Freestyle MV mode, you can lock in your character via the Character asset you’ve already created, then use the storyboard prompts to describe the scene action. Being very specific in each shot’s prompt (e.g. “character walks through forest, same outfit, medium shot”) tends to produce more consistent results than a single high-level instruction.
- The knowledge base notes that character consistency is still being actively improved — future model updates are specifically targeting this.
That said, I’d like to pass your specific case to the team so they can look at what’s happening with your setup. Could you share the project link (the URL when you open the MV project in Tunee) so engineering can dig into your exact generation?