Does Seedance 2.5 Support Text-to-Video?

Overview

Yes. Seedance 2.5 supports text-to-video on WaveSpeedAI through a dedicated text-to-video route. Use it when your starting point is a written scene rather than a source image: a product moment, a cinematic shot, a social creative, or a concept preview that can be described clearly in natural language.

The route is strongest when the prompt reads like a shot brief. Include the subject, setting, action, camera movement, lighting, style, duration target, and any constraints that should stay stable through the clip. If audio matters, check the current route settings and write what kind of sound should or should not be generated.

Text-to-video is not the best default when product shape, character identity, or a specific composition must be preserved. In those cases, start from image-to-video or a reference-driven route so the model has visual anchors instead of relying only on descriptive text.

On WaveSpeedAI, text-to-video is useful for fast ideation and API workflows because the same prediction pattern can later be extended to other Seedance 2.5 routes. Test text-only prompts first, then promote the ones that work into reference-based variants if the output needs tighter brand or subject control.

Source: ByteDance Seed official Seedance 2.5 page; WaveSpeedAI Seedance 2.5 API collection. Source check: 2026-08-29.