Seedance 2.5 capability guide

Use Seedance 2.5 inside Video Loom.

Seedance 2.5 is hosted through FAL with text, image, and multimodal reference routes for 4–30 seconds at 480p and 720p. Older Seedance routes remain available as separately labeled catalog rows.

Open Video Loom See the capability table

Video Loom routing scenes to Seedance 2.5 (via FAL)

Where Seedance 2.5 (via FAL) fits

Seedance 2.5 (via FAL) is one of the configured video providers Video Loom can route a scene to. Longer shots and multimodal reference work where the exact 2.5 route should stay visible in project provenance.

The 2.5 reference route accepts ordered image, video, and audio references.
Seedance 2.5 can generate synchronized audio when that option is enabled.
Text, image, and reference routes are registered as separate selectable models.

Capability facts from the model catalog

These facts are read directly from the Video Loom model catalog across all registered Seedance family routes, so they stay accurate as tiers change.

Registered models: 12.
Max clip duration: 30s.
Supported resolutions: 480p, 720p, 1080p, 4k.
Reference image support: Yes (up to 30 per call on some tiers).
Combined multimodal reference limit: up to 50 per call on some tiers.
Multi-reference input: Yes.
End-frame continuity: Yes.
Source-video input: No.
Lip sync: No.
Temporal extend: No.
Native audio output: Yes.

Per-model and mode facts

Each row below describes one exact catalog route. Family-wide maximums are not promises that every route supports the same inputs or output tiers.

Standard text-to-video (fal:seedance-2-text): text input, up to 15s, 480p, 720p, 1080p, 4k, FAL credential.
Standard image-to-video (fal:seedance-2-image): image input, up to 15s, 480p, 720p, 1080p, 4k, FAL credential.
Standard reference-to-video (fal:seedance-2-ref): multimodal reference input, up to 15s, 480p, 720p, 1080p, 4k, FAL credential.
Fast text-to-video (fal:seedance-2-fast-text): text input, up to 15s, 480p, 720p, FAL credential.
Fast image-to-video (fal:seedance-2-fast-image): image input, up to 15s, 480p, 720p, FAL credential.
Fast reference-to-video (fal:seedance-2-fast-ref): multimodal reference input, up to 15s, 480p, 720p, FAL credential.
Mini text-to-video (fal:seedance-2-mini-text): text input, up to 15s, 480p, 720p, FAL credential.
Mini image-to-video (fal:seedance-2-mini-image): image input, up to 15s, 480p, 720p, FAL credential.
Mini reference-to-video (fal:seedance-2-mini-ref): multimodal reference input, up to 15s, 480p, 720p, FAL credential.
Seedance 2.5 — Text (fal:seedance-2.5-text): text input, up to 30s, 480p, 720p, FAL credential.
Seedance 2.5 — Image (fal:seedance-2.5-image): image input, up to 30s, 480p, 720p, FAL credential.
Seedance 2.5 — Reference (fal:seedance-2.5-ref): multimodal reference input, up to 30s, 480p, 720p, FAL credential.

Plan, route, and finish around it

Seedance 2.5 (via FAL) shots stay inside the same storyboard, continuity library, and timeline as every other configured provider, so switching models never means switching workspaces.

Per-scene provider and model overrides.
Continuity anchors and reference frames travel with the request.
Generated takes finish on the same timeline as source and other-provider clips.

Questions

What teams ask before routing a shot here.

What references does Seedance 2.5 accept?

Its reference route accepts ordered images, videos, and audio files, subject to the per-type and combined limits shown by Video Loom.

Can Seedance and other providers be used in the same project?

Yes. Seedance is one of several configured providers Video Loom can route scenes to per shot.