Vertical Korean stock-market briefing
A 3-minute vertical stock report for 29 January 2026: dark cards split each ticker into cause, phenomenon, result and future, with Korean captions, a TTS voice and music. An AI agent built it in Remotion from a JSON dataset and an analysis in seven prompts.
2026-01-29.json 2026-01-29-analysis.md 이 주가 리포트 문서를 바탕으로 remotion skills를 활용하여 2026-01-29-analysis.md 문서를 충분히 읽어줄 수 있을 만큼의 길이로 투자 분석 보고서 영상을 만들어줘! 한국어 설명 자막을 함께 넣어서 제작해줘! 가독성이 높은 인포그래픽 형태의 리포트를 잘 만들어서 영상을 생성해줘!
주가 리포트를 앵커가 읽어주는 것처럼 한국어 설명 자막을 넣어줘
구글 TTS한국 여성 음성을 가져와서 자막 싱크와 맞춰서 자막을 읽어주는 음성을 만들고 영상에 추가해줘
TTS 음성을 한국 여성 음성으로 교체하고 TTS 가 다 읽기 전에 영상이 전환되면서 TTS 음성이 겹쳐지고 있어 TTS 음성 속도를 1.5배속으로 올리고 TTS 음성이 겹치지 않게 영상길이를 재조정하고 자막과 TTS음성 싱크를 재조정해줘
주가 리포트 인포그래픽 디자인을 좀 더 눈에 잘 들어오게 글씨랑 이미지를 추가해서 가독성을 높이고 에쁘게 보이게 수정해줘
영상에 TTS 음성 간에 공백이 너무 많은데 공백이 거의 없도록 컷을 좀 편집해줘
주가 리포트 영상에 어울리는 무료 배경음악 BGM을 다운받아서 TTS 음성에 방해되지 않는 볼륨으로 잔잔하게 넣어줘
Make it yours. Swap the two files named at the start of the first prompt ("2026-01-29.json 2026-01-29-analysis.md") for your own data and analysis. To change the language, edit "한국어" (Korean) in the caption prompts and "한국 여성 음성" (Korean female voice) in the voice prompts.
You'll need
- A Remotion project from npx create-video@latest (Blank template, TailwindCSS and agent skills turned on)
- The repo's prompt/2026-01-29.json and prompt/2026-01-29-analysis.md, or your own data and analysis
- Python with edge-tts for the voice script
How it was made
Seven prompts, in Korean, sent in order: build the report video from the JSON file and the analysis doc, add anchor-style captions, add a Korean female TTS voice synced to the captions, replace the voice and speed it to 1.5x without overlaps, redesign the infographic, cut the gaps between lines, and add quiet free background music. The agent wrote scripts/generate_tts.py, which uses edge-tts with the ko-KR-SunHiNeural voice, although the prompt asks for Google TTS. The README suggests Antigravity, Claude Code, Codex, Cursor and others but doesn't say which one made this video. Its step-by-step guide shows edited versions of these prompts; PROMPT.md, used here, keeps the originals. The project was created with npx create-video@latest with agent skills turned on.
Posted Jan 29, 2026 · GitHub
Prompt from: prompt/PROMPT.md in the GitHub repo
Look notesby Reference
One dark card per ticker in a vertical frame: the stock's name and daily change at the top, a quoted headline, a two-by-two grid labeled CAUSE, PHENOMENON, RESULT and FUTURE with one cell lit, and a LIVE · AI ANALYST BRIEFING caption bar at the bottom.
- Color
- Deep navy and near-black, with glowing cell fills in blue, violet, amber and teal, and green, yellow or gray for the percentage change.
- Type
- Bold Korean sans-serif for ticker names and quotes, small bilingual cell labels such as 원인 (CAUSE) with tiny icons, a large percentage figure top right, and the quoted caption line in the bottom bar.
- Framing
- Each 9:16 frame is filled by one card: a header row, a centered quote, two rows of two cells, and the caption bar near the bottom edge.
- Sound
- A Korean female TTS voice reading the captions over quiet free background music, per the prompts; the agent-written script uses edge-tts.
- Structure
- Eight tickers in turn, from Samsung Electronics (+1.8%) and LG Energy Solution (+5.5%) to a KRX gold ETF and smaller names, each with a different cell lit, then a closing market-summary card (시장 요약).
