Platform Adapters
Workshop does not compete with AI video generation tools — it orchestrates them. The prompt generation endpoint takes your .ws script and produces structured prompts optimized for specific AI video platforms.
Workshop owns the deterministic foreground (text, data, logos, layout, timing), while AI video tools generate backgrounds, b-roll, and atmospheric footage. The platform adapters bridge the two.
Generating Prompts
curl -X POST https://api.ws.video/api/prompt/generate \
-H "Content-Type: application/json" \
-H "x-api-key: YOUR_API_KEY" \
-H "x-org-id: YOUR_ORG_ID" \
-d '{
"source": "Film \"AI Demo\"\n mood: cinematic\n\nScene \"Open\" intent: reveal\n show \"The future is here\"\n size: hero\n enter: fade",
"platform": "sora"
}'
The response contains per-scene prompts with visual descriptions, negative prompts (what to avoid), camera motion instructions, style parameters, and duration constraints.
Supported Platforms
| Platform | Adapter Behavior |
|---|---|
generic |
Standard prompt output, no platform-specific constraints |
sora |
Duration clamped to 20s max, resolution fields added |
runway |
Camera motion mapped to Runway's control format |
kling |
Motion scaling adjusted for Kling's generation model |
veo |
Formatted for Google Veo's API parameters |
Each adapter reshapes the same creative intent into the format that works best for the target platform.
Themed Prompt Generation
Apply a ThemePack to influence the visual descriptions in generated prompts:
curl -X POST https://api.ws.video/api/prompt/generate \
-H "Content-Type: application/json" \
-H "x-api-key: YOUR_API_KEY" \
-H "x-org-id: YOUR_ORG_ID" \
-d '{
"source": "Film \"Branded\"\n mood: confident\n\nScene \"Brand\"\n show \"On brand, always.\"\n size: hero",
"platform": "generic",
"theme": "corporate-blue"
}'
The theme's palette and styling flow into the generated prompts, so AI-generated backgrounds match your brand's color language.
The Orchestration Model
The typical workflow is:
- Write a
.wsscript defining your video's structure, text, and data - Compile it with Workshop to get the deterministic foreground
- Generate platform-specific prompts for background footage
- Send prompts to Sora, Runway, Kling, or Veo
- Composite the foreground (text, charts, logos) over the AI-generated background
Workshop handles steps 1–3 and provides assembly instructions for step 5. Steps 4 and 5 happen in your pipeline using the AI video platform's own API.
Negative Prompts
Each scene's output includes a negative_prompt field describing visual elements to avoid — text artifacts, wrong color temperatures, distracting elements. These improve output consistency when passed to AI video generation models that support negative prompting.
Next Steps
- Prompt Generation API — endpoint reference and response schema
- Sandwich Architecture — the theory behind combining deterministic compilation with AI generation