Shipped batch convert web pages to JPG with viewport, JPEG quality, and full-page vs first-screen capture so it is not a PDF-batch doorway.

Date: 2026-09-07 08:45 Summary: Shipped batch convert web pages to JPG with viewport, JPEG quality, and full-page vs first-screen capture so it is not a PDF-batch doorway. Visibility: people

[question] 立项：batch convert web pages to jpg，并实现（参考batch convert web page to pdf的实现），过程中info Gain一定要分析出和实施好方案和差异性方案

Tool links (English)

https://onlinefreetools.org/tools/batch-convert-web-pages-to-jpg Chinese: https://onlinefreetools.org/zh/tools/batch-convert-web-pages-to-jpg Neighbor for contrast: https://onlinefreetools.org/tools/batch-convert-web-pages-to-pdf

[try to solve]

Context

The site already had a URL-list batch that builds A4 PDFs. The request was a sibling that turns the same kind of list into JPEGs, with Information Gain that is real in the product, not a title/suffix swap (doorway). Keyword Planner also has a “webpage to jpg / url to jpg” cluster; those single-URL jobs had to be absorbed into one page rather than a second URL.

Process

Wrote work-tasks/batch-convert-web-pages-to-jpg/ (00–03), filled coverage + intent + Ads tables, ran coverage:gate --phase=0b.
Compared SERP/neighbors: batch web-to-PDF (A4 pagination), convert-html-to-pdf (paste HTML / one URL), pdf-to-jpg (upload a PDF), bulk image-to-jpg (local photos).
Implemented a new page cloned from the PDF batch for fetch, sanitize, progress HUD, and ZIP — then replaced html2pdf/jsPDF with html2canvas → canvas.toBlob('image/jpeg').
Added first-screen controls that change the raster: viewport 1280 / 768 / 390, JPEG quality 0.7 / 0.85 / 0.95, full page vs first screen, plus the existing canvas-edge cap and noscript/refresh strip.
Ten-locale rewrite from locale briefs (not English calque). verify:tool --slug=batch-convert-web-pages-to-jpg went green.

Root cause / analysis

A JPEG sibling that only changed H1 and .pdf → .jpg would be scaled thin content: same A4 print pipeline, same artifact family. Searchers who want screenshots (tickets, chat, mobile vs desktop QA) need a different layout and a different file. The IG that actually moves the pixels is:

Neighbor	Artifact	This page
Batch web → PDF	A4 paginated documents	One JPEG per URL at a chosen viewport
PDF → JPG	Uploaded PDF pages	Webpage URLs, not a PDF file
Bulk images → JPG	Local photos	HTML capture, not an image upload

Planner “webpage to jpg” (100–1,000) is the same job with one line in the list, so it is absorbed (no extra doorway URL). free / online were not stuffed into the H1.

Solution

Slug / H1: batch-convert-web-pages-to-jpg / Batch convert web pages to JPG.
Fetch: existing GET /api/tools/convert-html-to-pdf?url= (once, not stored). localProcessing: false.
Capture: inlined html2canvas JPEG pipeline (isolation: no new shared helper). Same sanitizer as PDF batch (noscript / meta refresh stripped; phone kept).
Controls that change output: viewport width, JPEG quality, full page vs first screen; canvas edge capped at 8192.
UX: serial, max 10 URLs, skip failures, ZIP via fflate. Progress HUD. loadSample() exists for the button and lint, but does not auto-run on entry (html2canvas would freeze the tab).
Related: batch web-to-PDF, pdf-to-jpg. FAQ/Rules state the three-way split in visible copy.
Ship: npm run verify:tool -- --slug=batch-convert-web-pages-to-jpg OK. Home cards and prerendered HTML include the new slug.

Notes / boundaries

Not a site crawler; not an HTML paste editor; not A4 print; not PDF-file rasterization.
Inbound related from the PDF sibling was left unchanged to keep tool-isolation on this slug (SEO link validator warned of zero inbound related).
Live click-through in a running browser was not available in this session; verification used verify:tool, prerendered HTML (H1, viewport/quality/capture selects, html2canvas, click-only loadSample), and home-card presence on / and /zh/.

[actions]