Historical Story Video · publiczny skill Open Skill
Historical Story Video
This skill is designed to produce short, documentary-style historical videos (approximately 30 seconds) in a vertical format (9:16), making it perfect for YouTube Shorts, TikTok, or Instagram Reels. Key Features of this Skill: Flexible Story Sources: Simply provide a historical topic or figure for the AI to automatically gather accurate facts from the web, or input your own custom story script. Voiceover / Narration: Generates a calm, clear, and cinematic documentary narrator voice in any language you choose. Cinematic Visuals: Automatically produces 3 realistic, documentary-style scene images, which are then animated into video clips with smooth camera movements (such as slow pans or zooms). Complete Audio: Adds atmospheric background music that perfectly suits the historical theme. Smart Revision: If a scene or audio track isn't quite right, you can request revisions for specific parts (e.g., only replacing the image for scene 2, or just changing the narration language) without having to recreate the entire video from scratch.
Series
7 5 8
https://d7qlwi7tl2u27.cloudfront.net/skill-covers/prod/1728783/2026-09-13/hks3nvcn.jpg
# Historical Story Video (30s) ## When to use - The user wants a ~30-second vertical (9:16) documentary-style video telling a historical story, with spoken narration (voiceover) in a language they choose. - The user wants the story either from their own input OR from factual information gathered via web search (so the info is valid). - The user wants to revise one stage of an existing video from this skill (change the story/script, regenerate a scene image or clip, change the narration language, change the music, generate a thumbnail, or re-mix the final video). ## Steps ### 1. Gather input (ask only what's missing) Ask the user for (if not already given): - **Topic / era / person** of the historical story. - **Narration language** (e.g. Indonesian, English, Malay, etc.). - **Source**: (a) they provide their own story text, or (b) use web search to gather valid facts. - Optional: preferred visual mood (e.g. ancient, war, royal, industrial). ### 2. Gather the story content - If the user chose **web search** (or gave no story): call `search_web` with the topic to collect accurate, factual historical details (dates, places, key figures, events). Use the returned facts as the basis. - If the user provided **their own story**: use their text directly. - Keep the content focused enough to fit a ~30-second narration. ### 3. Write the narration script Write a short documentary-style narration script (~70–80 words, roughly 30 seconds at a calm narrator pace) **in the chosen language**. Tone: factual, cinematic, engaging. Structure: hook → key facts → closing line. Save it with `create_document` (name="Narration Script", entity_type="scripts") so the user can review/edit it. If the user edits it, use `update_document` on the same asset. ### 4. Generate the narration (voiceover) Call `generate_tts` with `model='tts.design'`, `text` = the narration script, and a `voice_instruction` describing a calm, clear documentary narrator voice (e.g. "deep, calm, clear documentary narrator, neutral accent"). The speech will be produced in the language of the script. Name="Narration Audio", entity_type="voiceover". ### 5. Generate scene images (documentary style) Generate **3 vertical (9:16) documentary-style images** that illustrate the story (e.g. period-appropriate scenes, key moments, figures). Use `generate_image` with `model='t2i'`, `orientation='portrait'`, `resolution='720P'`, and a detailed prompt describing each scene in a realistic, cinematic documentary style. Name them "Scene 1", "Scene 2", "Scene 3" (entity_type="storyboard_images"). ### 6. Generate video clips from the images For each scene image, call `generate_video` with `model='i2v'`, `input_urls=[scene_image]`, `orientation='portrait'`, `duration=10` (3 clips × 10s ≈ 30s). Prompt each clip with a subtle cinematic motion (slow pan, zoom, drifting smoke, waving flag, etc.) matching the scene. Name them "Clip 1", "Clip 2", "Clip 3" (entity_type="storyboard_videos"). ### 7. Generate background music Call `generate_music` with a prompt for a subtle, atmospheric documentary score (e.g. "soft, cinematic, historical documentary background music, low strings and piano, calm and evocative"). Name="Background Music", entity_type="music". ### 8. Mix the final video Call `mix_clips` to assemble the final ~30s vertical video: - `clips`: one clip config that concatenates Clip 1 → Clip 2 → Clip 3 (use `videos` list in order). - Add the narration audio track (`audio` = Narration Audio) and the music track (`audios` = Background Music, low volume ~0.3). - `orientation='portrait'` (9:16). Name="Final Video", entity_type="final_videos". ### 9. Generate a thumbnail (cover image) Generate a striking vertical (9:16) documentary-style thumbnail/cover image that represents the story (e.g. the most iconic scene or figure, cinematic lighting, title-worthy composition). Use `generate_image` with `model='t2i'`, `orientation='portrait'`, `resolution='720P'`. Name="Thumbnail", entity_type="storyboard_