Content rules

Rules for Step 2 (tool_synthesia_prepare_request) — how to distribute content across slides, write voiceover scripts, and format every variable type correctly.

slide_0_script_intro

Every video begins with slide_0_script_intro — a spoken intro of 1–3 seconds. Generated from the video title and/or the first key idea. No generic "Welcome!" unless stated in the source text.

Choose one pattern from the library and fill it with topic keywords:

Content distribution

• Distribute content so each slide covers one aspect — no repetition between slides.
• Slide scripts must reflect only what is on that slide. Never mix slide content.
• If video_length = short: combined word count across all script fields = 100–150 words. Keep text variables brief; rely on narration.

Variable type rules

No HTML or Markdown anywhere. Exceptions, inside script variables only:

• <break time="1s" /> — pause between ideas
• <sub alias="Term">Term</sub> — special pronunciation of key terms

Variable completeness rules

• Script variables must never be empty. A slide with a missing script will fail validation.
• Text variables should not be empty — use a single space " " if truly nothing fits.
• Variables named slide_3_* belong to slide 3. Never mix content between slide numbers.
• Numbered list slides (codes 2–8) are already numbered in the template. All item slots must be filled with consistent, parallel content. Do not pad a 4-item list with 3 bullets and one example.
• _max_N in a variable name = strict character limit. _title = strict max 30 chars.

Address forms by language

Use consistently throughout all scripts and text variables in a video:

Output format (Step 2)

templateData is an array of strings — each entry is "variable_name: value". The validator converts this to a dict before the API call.

"""
{
  "test": false,
  "templateData": [
    "course_title: ...",
    "video_title: ...",
    "avatar_id: ...",
    "background_image: ...",
    "slide_0_script_intro: ...",
    "slide_1_title: ...",
    "slide_1_script: ...",
    "slide_2_item_1: ...",
    "slide_2_item_2: ..."
  ],
  "visibility": "public",
  "templateId": "...",
  "title": "1.1.2 Video Title (Template Name)"
}
"""

Field rules:

• title = "<sections_position> <video_title> (<template_title>)"
• test = value of test_mode input
• visibility = always "public"
• templateId = UUID from Step 1 output
• All template variables including avatar_id and background_image go inside templateData

Image generation

Vertical / horizontal (stock photo)

Used for slide_#_image_vertical and slide_#_image_horizontal variables.

• Professional stock-photo style, soft daylight, blue tones in clothing and objects
• No text or labels in the image
• No dramatic facial expressions
• For screens: one specific, cinematically plausible camera angle — focus on the human interaction, not the screen content

Conceptual diagram (flat vector)

Used for the conceptual_diagram variable in I-slide templates. Always horizontal.

• Flat vector illustration, no strokes or borders, soft pastel and blue tones
• Must be concept-specific — never generic icons or abstract shapes
• Brand colors: #223F52 #0A5BEA #EDEBF2 #FFFFFF #D6EAFE
• Brand font: Aeonik, Helvetica, sans-serif
• Neutral daylight color temperature, no sepia filter
• Arranged for horizontal (3:2) layout with generous margins