Introducing image-to-video — and 3-shot coverage

Until last week SceneWeaver produced still images. This week, every shot can become a 5-second motion clip — and you can generate three different motion takes from the same still, giving you coverage choices when you cut.

What shipped

• Image-to-video via fal.ai (Seedance 2.0 by default, plus Seedance Fast, Kling v2.1, and two Alibaba Happy Horse models).
• 3-shot coverage — POST /api/projects/{id}/shots/{shotId}/video with variants: 3 runs three parallel generations: a slow push-in, a lateral pan reveal, and a static frame with subject micro-motion.
• Admin model switcher — pick the active video model on /admin/ai. No redeploy needed.
• Render mixer — the project render endpoint now auto-uses the video clip for shots that have one, still images for the rest.

Pricing

Stills are 10 credits each. Video clips are 100 credits each. A 3-shot coverage burns 300 credits.

If you want to try it, head to a shot you already generated, hit "Animate," and pick "3 variants." We'll show you the three takes side by side and you pick which one renders into the final cut.