Claude Code + Remotion (landscape-explainer-video skill)Partly tested🇨🇳 CN

Xian (@Xian0063), a UX designer for 10 years, made 18 animated explainer videos with Claude in 4 days, every frame drawn in code; the Van Gogh episode has one visual per line, and the voiceover was checked word by word with Whisper

Xian (@Xian0063, a self-described 10-year UX designer who doesn't code) wrote a long retrospective: 18 animated videos in 4 days with Claude, ~50 renders, including a 3-minute Van Gogh piece made "without downloading a single image or opening an editor"—the paintings are drawn by code with thousands of brush strokes. The post shares a copyable Remotion prompt template, an 8-step method (topic → emotional-arc script → voiceover checked word by word → visuals grown from the subject → one visual per line → frame-grab QA → cover frame at the start → multi-platform export), 6 design rules, and an open-sourced Claude Code skill, landscape-explainer-video. We inspected the skill repo, syntax-checked its scripts, ran its English-subtitle line-count checker, and generated a line of Xiaoxiao +8% voiceover with edge-tts as described; we did not render a video—the skill depends on the author's own Remotion project and won't run standalone.

Visit site Claude Code + Remotion (landscape-explainer-video skill) ↗
Pricing
Freemium
Region
Global
Where it's from
Chinese community
Added
Oct 27, 2026
Tested on
Oct 27, 2026
Version we used
X article 2026-10-22; xianxie6/landscape-explainer-video main @ 4cc08ce (2026-10-22); edge-tts 7.3.1
Categories
VideoDesign
Original author
弦 Xian @Xian0063 ↗
Our test
Partly tested
Cost / time
Claude Code needs a plan or API; Remotion, edge-tts, Whisper and Source Han fonts are free (Remotion has company licensing terms—check them). Rendering is local, ~10 minutes for 3 minutes of video.
Source
X @Xian0063(长文) ↗GitHub xianxie6/landscape-explainer-video ↗GitHub remotion-dev/skills ↗

The task

Make 2–3 minute animated explainers (art history, humanities, science) with Claude writing code that draws every frame—no stock footage or editing app—in both landscape and vertical for X, Bilibili or WeChat Channels.

Steps

  1. Pick a route: the post splits "animation with Claude" into 4—code-to-video (Remotion / HyperFrames / Manim), web-as-animation (Artifacts, p5.js, Three.js), Claude driving pro tools via MCP (Blender / After Effects / Lottie), and AI assets + code assembly. The Van Gogh piece uses Remotion.
  2. Install the official Remotion skill: `npx skills add remotion-dev/skills` (the author notes generic web-animation code stalls when rendered frame by frame—that's why).
  3. Topic: a little-known fact with a huge contrast ("sold one painting in life, one sells for $100M today"); open with the absurd fact, not "who was Van Gogh".
  4. Script along an emotional arc—hook → struggle → warmth → fall → rhetorical turn → low point → big reversal → closing line; ask Claude for a storyboard script first and only write code after you approve it.
  5. Voiceover: Microsoft Edge cloud voice (Xiaoxiao, +8%), with numbers written as spoken; transcribe it back with Whisper and compare word by word with the script—fix the script and regenerate on any misread.
  6. Visuals: at least one visual change per line (a stamp slams, a red line strikes through, an auction gavel drops, a price counts up); no static text for more than 2 s; hold the climax 2 s longer.
  7. Frame-grab QA: grab one frame per segment into an overview sheet to catch overlaps, occlusion and empty frames; put the richest caption-free frame at 0.5 s as the cover.
  8. Export landscape, vertical, with/without host, bilingual subtitles from the same visuals; check sensitive words before posting on Chinese platforms.
  9. To reuse the author's exact recipe: put landscape-explainer-video in `~/.claude/skills/` and tell Claude Code "make a video about XX with the Van Gogh effect"—but you first need a Remotion project shaped like the author's (see pitfalls).

Prompt

Make a [length]-second animated video in Remotion, [1920×1080 landscape / 1080×1920 vertical]. [Topic] [one sentence on what it's about] [Audience] [who watches, which platform] [Structure] open with a 5-second contrast → [key points] → end with one memorable line + a question to the audience [Visual style] [style name, e.g. painterly brush strokes / explainer cards / product launch]; main colour [colour]; font [serif / sans] [Pacing] at least one visual change per spoken line; no static text longer than 2 s; hold the climax 2 s longer [Constraints] no text errors; write numbers as spoken; keep important content away from the bottom edge (subtitles cover it). Give me the storyboard script first and only write code after I approve it.

What the author says

The author claims 18 animated videos in 4 days, ~50 renders and 35 revisions to build guidance; the 3-minute Van Gogh video was drawn entirely by Claude-written code with no code written by the author; each painting is ~2,600–3,600 strokes; a 3-minute landscape render takes ~10–11 minutes. A cited Chinese case: meng shao's Remotion Skills series reached 80k views and 4k shares on one WeChat Channels video. The article has ~190 likes and 280+ bookmarks.

Our test

We rendered no video and ran no Claude Code session: the skill depends on the author's unpublished Remotion project and companion skill. Shallow clone of landscape-explainer-video (main @ 4cc08ce, 2026-10-22): SKILL.md, four tsx templates (Episode/kit/paint/ui, ~1,000 lines), three reference docs (recipes, script rules, lessons) and six Van Gogh screenshots. `python3 -m py_compile` and `node --check` pass on check_en.py and stills.mjs; zsh isn't installed here, so the .sh scripts weren't run. On dummy data check_en.py correctly failed on a line-count mismatch and passed when aligned. We also generated one Van Gogh opening line with edge-tts 7.3.1 using the article's settings (zh-CN-XiaoxiaoNeural, +8%).

What we found when we tried it:edge-tts needed no key and produced a 5.2 s mp3 in ~5 s, so the voiceover step is free and reproducible. The skill's instructions, templates and lessons are thorough, but you must build the Remotion project scaffold yourself to render; we did not verify the video output.

How to get in

The original is a Chinese X Article (login may be required). The skill repo is on GitHub with Chinese docs; this English page summarises and translates the method and prompt template.

Who it suits

Solo creators of knowledge, art-history and science channels; designers with taste and storytelling but no coding; anyone making bilingual videos for X / YouTube.

Pitfalls

  • The open-source skill isn't plug-and-play: its README says it relies on the author's own Remotion project (`makeTimeline`, an SFX library, `tools/gen_vo.py`), the voiceover scripts of a companion skill (douyin-vertical-video), and cartoon host assets—none of which are published. Treat it as method and code reference, or have Claude build your own project to match.
  • Scripts are zsh, and `new_episode.sh` uses macOS-style `sed -i ''`, which breaks on Linux; the project path defaults to the author's `~/Documents/...`—override with `PROJ`.
  • The first render is rarely right—expect 2–3 rounds; vague prompts give template-looking output, so specify style, colour and pacing.
  • Numbers and polyphonic characters are the usual TTS misreads—always do the Whisper read-back.
  • Anchor subtitles to the bottom; with top positioning a two-line English subtitle gets pushed off-screen (the author hit this).
  • Use only code-drawn, public-domain or CC0 material—"visible online" is not a licence.

More in this category