AI models & service providers
Last updated: .
Oouuh AI is an independent photo-to-video service. Here is the underlying video model we currently use, how we access it and how your kitchen duet is made.
The video generation model we use
- Model name
- ByteDance Seedance 2 Mini
- Model developer
- ByteDance
- API model identifier
bytedance/seedance-2-mini- API access provider
- KIE API (api.kie.ai)
- Used for
- Both six-second reference edits in every Oouuh kitchen duet
The Oouuh template automatically uses this model. There is currently one video generation model for this service, so you do not need to select a model before generating.
ByteDance develops the Seedance model family. KIE provides the API access and generation processing: our server submits video requests to its jobs API using the model identifier above. We do not train or own the underlying model, and Oouuh AI is not an official ByteDance or KIE product.
How your video and soundtrack are made
Your two uploaded photos supply the appearances of the left ukulele player and the right singer. We send the photos, generation instructions and kitchen reference clips to KIE. ByteDance Seedance 2 Mini generates two silent six-second reference edits, which we assemble into one twelve-second, vertical 720p MP4.
We then add one continuous, prepared twelve-second soundtrack. Choose the original male vocal or a female voice conversion of the same performance. The female track was prepared with local Demucs vocal separation and Seed-VC singing voice conversion, then mixed with the retained ukulele backing. Both versions follow the original song and timing. A synthetic female vocal reference supplied timbre; no particular female artist was selected. Your photos set appearance and are not used to clone voices. Preview both recordings in the generator.
The actual generated example on our homepage also uses ByteDance Seedance 2 Mini, with fictional adult portraits. It shows one result, not a guarantee of how your photos will look.
Content screening and safety services
Waffo Prompt Screening checks generation text before photos are uploaded to KIE or credits are reserved. It is a text safety service; ByteDance Seedance 2 Mini is the model that generates the video. A block, review verdict or unavailable screening check stops the request. KIE’s nsfw_checker is also enabled for generation.
These automated safeguards have limitations and do not mean every image or video frame is inspected or manually approved. Read our AI content policy or report content and appeal a decision.
Results, data and practical limitations
All generated footage is synthetic. Likeness, hands, the ukulele, mouth timing and continuity between the two segments can contain errors. Technical checks verify video length, frame timing and audio presence; they cannot guarantee identity accuracy or perfect lip sync. Review your completed video before sharing it.
Your photos and generation instructions are processed by KIE to fulfill the request. Completed videos are stored privately in Cloudflare R2 and are available to their authorized account. Our privacy policy explains processing and retention. The reference performance and soundtrack do not establish an artist endorsement or grant unrestricted commercial-use rights; see the terms of service.
For questions about the model or a generated result, contact Oouuh AI support. View one-time credit packs or create your kitchen duet.