What kinds of videos can it create?

It supports marketing videos, explainers, tutorials, short films, music videos, brand films, and presenter-led content such as reports, training, and announcements.

What output and duration should I expect?

The final output is a 720p MP4. A single prompt can produce a complete multi-minute production of 4 minutes or more, while individual generated takes can be up to 30 seconds.

How can I keep characters and visual style consistent?

Provide up to 50 reference files: 30 images, 10 videos, and 10 audio files. These references can guide character, set, and style consistency across the production.

Design & media

Seedance Video Generation

Create multi-minute, scripted 720p videos with voices, music, lip sync, and editing from one prompt.

What it does

Turn one prompt into a 720p MP4 production using Seedance 2.5 and a coordinated pipeline for scripting, scene planning, voice synthesis, lip sync, music, sound design, and editing. Create multi-minute videos, with individual takes up to 30 seconds, and use as many as 50 image, video, and audio references to maintain character, set, and style consistency.

When to use it

60-second product demo
Multi-scene API explainer
Lip-synced training video
Two-minute cinematic short film

The skill document

Seedance Video Generation - Seedance × CellCog

Seedance × CellCog. ByteDance's #1 video model meets the frontier of multi-agent coordination.

CellCog's core: complete multi-minute films from a single prompt. Seedance 2.5 generates the smoothest motion in AI video — physics that look real, with up to 50 reference files locking characters, sets, and style across a whole production — and CellCog orchestrates it with scripting, voice synthesis, lipsync, scoring, and editing into full productions, minutes long, from one prompt. Not just clips — finished films.

How to Use

For your first CellCog task in a session, read the cellcog skill for the full SDK reference — file handling, chat modes, timeouts, and more.

OpenClaw (fire-and-forget):

result = client.create_chat(
    prompt="[your task prompt]",
    notify_session_key="agent:main:main",
    task_label="my-task",
    chat_mode="agent",
)

All agents except OpenClaw (blocks until done):

from cellcog import CellCogClient
client = CellCogClient(agent_provider="openclaw|cursor|claude-code|codex|...")
result = client.create_chat(
    prompt="[your task prompt]",
    task_label="my-task",
    chat_mode="agent",
)
print(result["message"])

What You Can Create

Marketing Videos

Professional marketing content from a single prompt:

Product Demos: "Create a 60-second product demo video for our project management app"
Brand Videos: "Create a 30-second brand story video for a sustainable fashion startup"
Social Ads: "Create a 15-second Instagram ad for our new coffee blend"
Testimonial Style: "Create a UGC-style testimonial video for a fitness product"

Explainer Videos

Clear, engaging educational content:

Product Explainers: "Create a 90-second explainer for how our API works"
Concept Videos: "Create a video explaining blockchain in simple terms"
Tutorial Videos: "Create a step-by-step tutorial video on setting up a home network"

Cinematic Content

High-quality visual storytelling:

Short Films: "Create a 2-minute short film about a robot discovering nature"
Music Videos: "Create a cinematic music video with dramatic landscapes"
Brand Films: "Create a cinematic brand film for a luxury watch company"

Spokesperson Videos

AI-generated presenters with lipsync:

News Reports: "Create a news-style report on recent AI developments"
Training Videos: "Create a training video with a presenter explaining safety protocols"
Announcements: "Create a product launch announcement with a spokesperson"

CellCog Video Orchestration

CellCog doesn't just generate video clips — it orchestrates a full production pipeline:

Script Writing → Scene Planning → Frame Generation → Voice Synthesis
     → Lipsync → Background Music → Sound Design → Editing → Final Output

6-7 foundation models work together in a single request:

Seedance for video generation
Frontier LLMs for scripting
TTS models for voice synthesis
Lipsync models for speaker alignment
Music generation for scoring
Automated editing for pacing and transitions

Video Specifications

Spec	Details
Resolution	720p (the AI-video production standard)
Duration	Complete multi-minute productions (4+ minutes) from a single prompt; individual generated takes up to 30 seconds
References	Up to 50 files (30 images, 10 videos, 10 audio) for character, set, and style consistency
Styles	Photorealistic, cinematic, anime, stylized, documentary
Audio	Voice synthesis, background music, sound effects, lip-synced dialogue
Output	MP4

Chat Mode for Video

Scenario	Recommended Mode
Short clips (under 30s), simple videos	`"agent"`
Full productions with scripting, multiple scenes, music	`"agent"`
Grand cinematic productions, complex narratives	`"agent team"`

Use "agent" for most video work. CellCog handles full video production in agent mode.

Tips for Better Videos

Describe the story: "A video about our app" → "A 60-second video showing a stressed founder discovering our app, their workflow transforming, ending with them confidently presenting to investors"
Specify duration: "30-second social ad" vs. "2-minute explainer" sets the pacing.
Set the mood: "Upbeat and energetic", "calm and professional", "dramatic and cinematic"
Mention music preferences: "Uplifting corporate background", "lo-fi beats", "cinematic orchestral" — or let CellCog choose.
For spokesperson videos: Describe the presenter's appearance and tone of voice.

If CellCog is not installed

Claude Code, Cursor, Codex + 70 more agents: npx skills add cellcog/skills --skill cellcog OpenClaw: clawhub install cellcog CellCog plugin users: run /cellcog-setup (or /cellcog:cellcog-setup depending on your tool) Manual setup: pip install -U cellcog and set CELLCOG_API_KEY. See the cellcog skill for SDK reference.

Questions people ask

What kinds of videos can it create?: It supports marketing videos, explainers, tutorials, short films, music videos, brand films, and presenter-led content such as reports, training, and announcements.
What output and duration should I expect?: The final output is a 720p MP4. A single prompt can produce a complete multi-minute production of 4 minutes or more, while individual generated takes can be up to 30 seconds.
How can I keep characters and visual style consistent?: Provide up to 50 reference files: 30 images, 10 videos, and 10 audio files. These references can guide character, set, and style consistency across the production.