AI models we use
Effect Farm makes videos with Seedance 2.0 Mini (paid HD videos) and Kling 2.6 Motion Control (free previews and backup). What each model does and why.
Last updated: October 4, 2026
Effect Farm is an independent product. We do not train our own video model. Every video is generated by one of the two models below, which we call through their API providers. We list them here so you know exactly what makes your video.
| Used for | Model | Made by | Provided through |
|---|---|---|---|
| Paid HD videos | Seedance 2.0 Mini (reference-to-video) | ByteDance | SeeAPI |
| Free previews; backup if the HD model fails | Kling 2.6 Motion Control (standard) | Kuaishou | WaveSpeed |
| Content moderation | NSFW Filter and Video NSFW Filter | SeeAPI | SeeAPI |
Seedance 2.0 Mini: paid HD videos
- Takes your photo and your motion video as references and generates a new video of the person in your photo doing the same motion.
- Output: MP4, 720p, vertical 9:16, 4–10 seconds, no watermark.
- If Seedance fails, we retry the same video with Kling 2.6. You are not charged twice.
Kling 2.6 Motion Control: free previews and backup
- Transfers the motion from your video onto the person in your photo.
- Free previews use this model, have a watermark and are view only.
Content moderation
Every uploaded photo, every uploaded motion video and every generated video is checked by SeeAPI's NSFW classifiers before generation or before you can see the result. Flagged content is blocked and any credit used is refunded. See our Acceptable Use Policy.
What we do not do
- We do not claim to be, or to be endorsed by, ByteDance, Kuaishou, Higgsfield or any model provider. Their names are used only to tell you which model is used.
- We do not use your uploads to train any model.