A three-step workflow for producing cinematic product motion graphics — use a Claude Fabel skill to generate a detailed prompt, pass it to Omni for visual creation, then animate with Seedance 2.
This guide is reviewed for clarity, service accuracy, and AI-search readability. The next quarterly content review is tracked internally before unsupported metrics or client proof are added.
The Workflow Overview
A three-tool pipeline for producing cinematic product motion graphics:
- Claude Fabel 5 + Motion Graphics Skill → generates a detailed, scene-by-scene prompt
- Omni → creates the visual frames from the prompt
- Seedance 2 → animates the frames with cinematic motion
The key: Claude acts as the prompt engineer, writing the kind of structured, technically detailed prompt that AI video tools need to produce high-quality output.
Step 1 — Install the Motion Graphics Skill
The custom Motion Graphics Skill teaches Claude Fabel 5 the vocabulary, structure, and level of detail needed for production-quality prompts.
To install: Claude → Customize → Skills → Add Skill (+) → Create Skill → Upload Skill. Download the skill file from the source guide's Google Drive, drag the .md file into the Upload Skill box. Persists across sessions.
Step 2 — Generate the Scene Prompt
With the skill active, describe your product and target duration. Claude outputs a frame-accurate, scene-by-scene breakdown. Example for a 6-second beverage product ad:
promptScene 1 (0:00–0:01) A bright yellow can tumbles into frame against a soft pastel-yellow background, spinning smoothly as it floats toward center. Scene 2 (0:01–0:02) Ingredients burst from behind the can, creating depth with cinematic motion blur and shallow depth of field. Scene 3 (0:02–0:03) Macro close-up. Cold can with condensation. Golden liquid splash wraps around the can with realistic fluid simulation and sparkling droplets. Scene 4 (0:03–0:04) Wide shot. Ingredients orbit the spinning can in a slow vortex. Splash suspended around it. Scene 5 (0:04–0:05) Can rotates 180° to reveal back label. Floating elements disappear. Ground plane rises from below. Scene 6 (0:05–0:06) Can rotates back to front logo, settles into hero shot with subtle hover and studio lighting.
What the skill adds that generic prompts miss:
- Frame-level timing — exact in/out timestamps per scene
- Physics descriptors — 'fluid simulation,' 'slow vortex,' 'gravity settle'
- Camera language — close-up, 180° rotate, hero shot
- Lighting spec — consistent across all 6 seconds
- Transition logic — what enters, exits, and exactly when
Step 3 — Omni → Seedance 2
- Paste the Claude-generated prompt into Omni for initial visual creation (composition, product placement, background design)
- Take the key frames from Omni into Seedance 2 for motion generation (fluid dynamics, camera movement, element animation)
- Iterate: revise specific scenes in the prompt → regenerate frame in Omni → re-animate in Seedance 2
Each step is modular — iteration is faster than traditional motion design.
Best Use Cases
- Product ads (beverage, beauty, supplements, packaged food)
- Social media ad creative — multiple visual variants quickly
- E-commerce product showcase videos
- Brand sizzle reels
Frequently Asked Questions
- Other video models instead of Seedance 2? Yes — the prompt is model-agnostic. Kling, Sora, Runway Gen-4 all work. Seedance 2 is recommended for smooth motion on product objects specifically.
- How long does the full workflow take? Claude prompt: 5–10 min. Omni frames: 2–5 min/scene. Seedance animation: 2–4 min/segment. A 6-second hero ad with one revision round: under an hour.
- Output looks generic despite the skill? The limiting factor is usually your product description. Add specifics: exact color code, material (matte vs gloss), desired emotion ('premium' vs 'fun and energetic').