Claude Code builds a product trailer and marketplace GIFs entirely in code, no After Effects
SinanDev · reddit · 2026-10-01
A developer used Claude Code to produce a product trailer and marketplace GIFs for their VS Code extension—no video editor, no After Effects, everything is code.
Trailer
- Each scene is a web page with a render(t) function drawing the exact frame for time t; Playwright screenshots at 60 fps, ffmpeg assembles the MP4—every frame reproducible
- Sound effects generated with Web Audio, timed to the same timestamps; narration via ElevenLabs with a phone filter, placed frame-exact
- VS Code, the Claude Code chat and the product panel are HTML replicas so the camera can zoom into any detail crisply
- Six scenes, each starting on the previous scene's last frame; before every render Claude screenshots stills and self-checks for clipping, overlaps and readability
- 20 iterations went mostly into story: promise in the first 5 seconds, then a relatable dev situation; calm moments matter as much as effects
README GIFs took the opposite approach
- The UI is the real extension, not a replica: Claude found the extension's dev server renders the actual webview in a browser and fed it demo data through the same message channel VS Code uses
- Playwright clicks real buttons and types into real inputs frame by frame; all nine GIFs total under 700 KB
- One idea per clip, under 10 seconds, no text—the README above does the explaining
The author says a year ago they'd have happily paid hundreds of dollars for this quality, and offers to share the skill they built for the format.
More from coding & agent
- Translating an entire book with DeepSeek: pennies and under an hour, decent quality — teortaxesTex · 2026-10-02
- OmniSeek turns Omni-LLMs into agents that actively seek audio-visual evidence — Haibo Wang · 2026-10-02
- Microsoft's ActiveSaddler Uses Automated Curriculum Learning to Boost Agent Harnesses by 7.5 Points — microsoft · 2026-10-02
- Alibaba's PoS Maintains Explicit Belief States to Fix Long-Horizon Agent 'Belief Trapping' — alibabagroup · 2026-10-02
- IntentFlux Benchmarks 'Intent Drift' in LLM Agents: Scores Fall from 0.476 to 0.384 as Users Change Their Minds — Yanjie Zhang · 2026-10-02
- Founder says 36 hours with OpenAI dots may replace his months of monorepo agent setup — hugobowne · 2026-10-02