wavespeed-seedance-15-pro_skill

Overview

This skill generates short high-quality videos using ByteDance's Seedance V1.5 Pro model via WaveSpeed AI. It supports text-to-video and image-to-video workflows, optional audio, camera control, smart duration, and configurable seeds for reproducible outputs. Videos run 4–12 seconds at up to 1080p and are suitable for rapid visual prototyping and content creation.

How this skill works

Provide a text prompt or a source image URL and the skill calls WaveSpeed’s Seedance V1.5 Pro endpoints to synthesize a 4–12 second video. Parameters let you control aspect ratio, resolution (480p/720p/1080p), duration (or smart duration), camera motion, audio generation, and deterministic seeds. The skill returns a hosted video URL once the generation task completes; error and retry handling options are available for robustness.

When to use it

• Create short cinematic or social-format videos directly from a text description
• Animate a single photo into a short clip (image-to-video) for promos or social posts
• Quickly prototype visual concepts, story beats, or camera moves without filming
• Generate background visuals with optional audio for ads, trailers, or title cards
• Produce repeatable outputs by setting a seed for A/B testing or iteration

Best practices

• Write prompts that describe scene, subject actions, mood, style, and camera motion for predictable results
• Use camera_fixed:true for tripod/stable shots; set to false and include motion phrases for dynamic camera behavior
• Prefer 1080p for final assets and 720p for faster/cheaper iterations
• Use generate_audio:false when you plan to score or replace audio later
• Validate and host images on trusted URLs before passing them to the skill; never send untrusted or user-submitted URLs directly

Example use cases

• Text-to-video: produce a 10s cinematic timelapse for a trailer with aspect_ratio:21:9 and generate_audio:true
• Image-to-video: animate a portrait to add a slow turn and smile for a social post, camera_fixed:true to preserve framing
• Smart duration: use duration:-1 to let the model pick the best length for short narrative beats
• Batch creative testing: vary seed values to generate multiple stylistic options for the same prompt
• Silent background loop: generate a 5s ambient clip with generate_audio:false for seamless editing