Skip to content

docs: add html-to-mp4 and product demo guides, retitle outcome pages - #5940

Open
ukimsanov wants to merge 1 commit into
mainfrom
docs/make-pages
Open

ukimsanov wants to merge 1 commit into
mainfrom
docs/make-pages

Conversation

@ukimsanov

Copy link
Copy Markdown
Collaborator

What

Need-based "Make" pages for the docs:

  • New: docs/guides/html-to-mp4.mdx ("Convert HTML to MP4") and docs/guides/product-demo-video.mdx ("Make a product demo video with AI").
  • Reworked, same URLs: /guides/product-launch-video, /guides/faceless-explainer, /guides/motion-graphics. New title and description, the first two paragraphs now answer the searched question directly, and each page gets a short FAQ. Existing media is unchanged.

/guides/website-to-video already redirects to /guides/product-launch-video, so that page now covers the "website to video" search too (in its intro and keywords).

Not in the nav yet. docs/docs.json is being restructured in a parallel PR, so this PR does not touch it. Please add these to the nav:

  • guides/html-to-mp4
  • guides/product-demo-video

Why

Search Console (Sep 20 – Oct 17) and Ahrefs US volumes:

Query / page Data Gap
/guides/product-launch-video 25,708 impressions at position 1.9, CTR 0.6% (155 clicks), almost all from brand searches. "product launch video" is 300/mo, KD 0 The title ("Create a product or website video") and description don't match the search, so a top-2 result gets few clicks
"html to mp4" 600/mo KD 1, "html to video" 100/mo KD 1 — No page targets it
"product demo video" 350/mo KD 5 — No page
"explainer video maker" 450/mo KD 26 /guides/faceless-explainer: 49 clicks, 503 impressions The title says "faceless explainer", but people search "explainer video"
"ai motion graphics" 350/mo KD 0 /guides/motion-graphics: 35 clicks Title and description don't use the phrase
"website to video" 100/mo KD 5 redirect to product-launch-video Covered by the reworked page

The FAQ entries paraphrase real questions from the docs assistant log (28 days), for example: "can it run on a site on a local server", "can I use my own script", "can I export motion graphics to Premiere", "do I have to put the motion rules in my prompt", "can I make an explainer from my own images". Nothing personal is quoted.

Related work

Refs the docs-as-growth work. A parallel docs.json navigation restructure is in progress, which is why the nav isn't edited here.

How

Titles and descriptions, before → after

Page Before After
product-launch-video Create a product or website video / "Turn a product URL or launch brief into a promo, product tour, or social video." Make a product launch video with AI / "Make a product launch video with AI from your website URL. Your agent captures the real site and builds an editable promo, tour, or social video."
faceless-explainer Create a faceless explainer / "Turn notes, an article, or a difficult idea into a visual explanation without product footage." Make an explainer video with AI / "Make an explainer video with AI from a topic, article, or notes. No footage or presenter needed: your agent invents the diagrams, type, and charts."
motion-graphics Create a motion graphic / "Make a short, motion-led title, statistic, chart, logo sting, or overlay." Make AI motion graphics / "Make AI motion graphics from one sentence: kinetic type, a stat count-up, chart, logo sting, or lower third, rendered as MP4 or a transparent overlay."
html-to-mp4 (new) — Convert HTML to MP4
product-demo-video (new) — Make a product demo video with AI

All descriptions are 130–150 characters. sidebarTitle is unchanged on the existing pages, so the sidebar labels stay the same.

Frontmatter. Uses title, description, and keywords only. Per Mintlify's docs, keywords feeds Mintlify's own search (synonyms like "html to video" or "website to video"). og:title and og:description already default to the title and description, so they aren't repeated.

Commands checked against source and --help (CLI 0.8.181):

  • capture: refuses localhost and private addresses (isPrivateUrl), so the localhost FAQ points people to screenshots or a recording.
  • tts: needs kokoro-onnx and soundfile installed in Python (error message checked); the page says so.
  • transcribe: writes transcript.json beside the audio file (checked by running it).
  • init --video / --resolution portrait, and render --format mov|webm|gif, --video-frame-format png.

Film placeholders. New docs films are made fresh, so no old showcase video goes on the new pages. Each new or reworked page carries a {/* FILM: … */} brief at the top:

  1. html-to-mp4: 10–15 s split screen. An HTML file in a code editor on the left and the same animation scrubbing frame by frame on the right, ending on the finished MP4 playing in a normal video player. No narration; captions name each step.
  2. product-demo-video: 30–45 s demo of a real SaaS task, made by this workflow. The captured screens move through one complete task (open, act, see the result), with a calm voiceover and word-timed captions, ending on the product name and URL. The capture folder shows for one second as the "made from this" proof.
  3. product-launch-video: 20–30 s launch film for one real product, built from its URL. A hook that proves the product in the first two seconds, three real-screen beats and one CTA end card. It opens on the URL being pasted.
  4. faceless-explainer: 30–40 s explainer of one hard technical idea (for example, how a vector database finds similar things), made from one paragraph of notes. A question hook, the mechanism revealed layer by layer on one visual stage, and one sentence to remember. Narrated and captioned.
  5. motion-graphics: 8–12 s reel of three graphics made from one-sentence requests, each request typed on screen first: a kinetic headline, a count-up, and a lower third composited over real footage to prove the transparent export. No narration.

Test plan

  • html-to-mp4 example rendered locally with npx hyperframes@0.8.181. lint gave 0 errors and 0 warnings. render --output hello.mp4 finished in 4.1 s (112.6 KB). ffprobe shows h264, 1920x1080, 30 fps, nb_frames=90, duration=3.000000. A frame pulled at 1.0 s shows the bar mid-grow, so the timeline is really being seeked.
  • mint validate: build validation passed.
  • mint broken-links --check-redirects: no broken links found.
  • check:docs-snippet-motion, check:docs-catalog (448 pages before and after) and check:tracked-artifacts all pass.
  • Still to verify: the tts step end to end. Here it failed on a local espeak-ng data path inside the Python package, which is an environment problem, not the CLI. The flags are taken from --help.
  • Documentation updated

🤖 Generated with Claude Code

New task pages for searches no page targets yet: "html to mp4" and
"product demo video". The html-to-mp4 example was linted and rendered
locally (3.0 s, 90 frames, h264 1920x1080).

Product launch, explainer and motion graphics pages keep their URLs and
media; titles, descriptions and opening paragraphs now answer the
searched question directly, each gains a short FAQ from real assistant
questions, and every page carries a FILM placeholder for a new film.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@mintlify

mintlify Bot commented Oct 20, 2026 •

Copy link
Copy Markdown

Preview deployment for your docs. Learn more about Mintlify Previews.

Project Status Preview Updated
hyperframes 🟢 Ready View Preview Oct 20, 2026, 7:30 AM

💡 Tip: Enable Automations to automatically generate PRs for you.

@ukimsanov
ukimsanov marked this pull request as ready for review October 20, 2026 08:18
Copilot AI balanced review requested due to automatic review settings October 20, 2026 08:18

Copilot AI left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

This branch was successfully deployed

1 active deployment
staging - docs — a58f9ffd Deployed Oct 20, 2026 by mintlify[bot]
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants