skills.rest
calesthio
calesthioCOMMUNITY
Explore all skills by this author →
𝕏@calesthioailabsGitHub

seedance-2-5

Generate 4-30 second cinematic videos with ByteDance Seedance 2.5 via hosted provider APIs.

55.1k|6.9k|Updated Mar 29, 2026
Search Tags:#video generation#Prompt Engineering#cinematic#runway#text-to-video#Seedance#Fal.ai
One-click install
npx skills add https://github.com/calesthio/OpenMontage --skill seedance-2-5
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: seedance-2-5
Source: https://github.com/calesthio/OpenMontage/tree/main/.agents/skills/seedance-2-5
Command: npx skills add https://github.com/calesthio/OpenMontage --skill seedance-2-5

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

Writing prompts for long-form AI video generation is error-prone: multi-shot sequences drift between cuts, characters lose identity, and providers expose different parameters for the same model. This Skill encodes the complete Seedance 2.5 prompt contract and routing rules so generations stay coherent across cuts and providers.

Core Features & Use Cases

  • Provider Routing: Calls Seedance 2.5 through fal.ai, Volcengine Ark, Runway, or ComfyUI Partner Nodes with the correct tool names, model identifiers, and reference-field formats for each route.
  • Structured Prompt Contract: Enforces a labeled section order (GLOBAL STYLE, SCENE, CHARACTERS, shot-by-shot breakdown, OPTICS, PHYSICS, LIGHTING, AUDIO) plus named continuity locks such as LENS LOCK, COUNT LOCK, and SCREEN DIRECTION.
  • Asset and Voice Method: Defines the three-plate character sheet, one-asset-per-state rule, rule-of-10 validation, and verbatim voice conditioning sentences for synchronized audio.
  • Use Case: A creator needs a 25-second noir sequence with two consistent characters across five hard cuts. The Skill supplies the shot breakdown template, screen-direction and lens locks, reference limits (30 images, 10 videos, 10 audio clips), and iteration discipline to produce it without identity drift.

Quick Start

Use the seedance-2-5 skill to generate a 15-second two-shot cinematic video of a character walking through a rainy neon street via fal.ai, keeping the character consistent across both shots.

Frequently Asked Questions about seedance-2-5

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I generate video with Seedance 2.5?▼

Call one of the supported OpenMontage routes: fal.ai with model_version 2.5, Volcengine Ark with model 2.5, Runway with model seedance2_5, or a ComfyUI Partner Node with model_family seedance_2.5. All routes are hosted and paid; there are no local weights.

How do I keep characters consistent across multiple shots in AI video?▼

Supply reference images and use named locks: FIRST FRAME AND SPATIAL BLOCKING, SCREEN DIRECTION, LENS LOCK, and COUNT LOCK. Build a three-plate character sheet with the face only on a sharp close-up, and write state monotonicity so details never reset between cuts.

What reference limits does Seedance 2.5 support?▼

Seedance 2.5 accepts up to 30 reference images, 10 reference videos, and 10 reference audio clips. Runway caps combined reference video and audio duration at 30 seconds, and a video-to-video source consumes one video slot.

Does Seedance 2.5 support 1080p or 4K output?▼

The supported API routes expose only 480p and 720p. The model generates 1080p natively on the vendor's own web platform, which is not an OpenMontage route, so check which surface a resolution claim refers to.

Why does my multi-shot video sequence fall apart between cuts?▼

Vague transitions are the most common cause. Give each shot an explicit time range, camera distance, and framing, separate them with Hard cut markers, and state genre, era, and lighting explicitly in GLOBAL STYLE since the API routes expose no UI selectors.

What can Seedance 2.5 not generate reliably?▼

Singing, sustained physical contact such as fights, and multi-character scenes are the hardest categories. Workarounds include pre-recorded labeled audio tracks, phone-shot motion references, and more reference material per face.

Related Skills

Explore More Design & Creative Skills→
kaiganikaigani
30

character-reference-generator

Generate visual reference assets for character consistency in media production.

Community
Advanced
video-production-buddyvideo-production-buddy
602

seedance-2-0

Generate cinematic video clips with synchronized audio and multi-shot cuts.

Official
Advanced
1veso1veso

higgsfield-soul-id

Train a personalized face-faithful identity model from photos using Bash scripts.

Community
Intermediate
framerslabframerslab

video-generation

Generates videos from text or images and analyzes scenes with object detection.

Official
Advanced