Website Screenshot & PDF: Full-Page, Mobile, Change Alerts
Pricing
from $2.50 / 1,000 captures
Full-page website screenshots and PDFs in PNG, JPEG, WebP. Lazy-load scrolling, sticky headers fixed, cookie banners dismissed, mobile/tablet/desktop in one run, element and clip capture, visual change alerts with diff %. Same input as apify/screenshot-url. Failed pages are free.
Pages to capture. Full URLs or bare domains ("example.com"). Several URLs pasted into one line separated by spaces or new lines work too.
Capture every page listed in these sitemap.xml files (sitemap indexes and .xml.gz supported), up to "Max URLs".
Upper limit of URLs per run (each URL is captured once per device). 0 = 10,000.
One capture per device for every URL. Presets: desktop (1280×800), laptop (1440×900), desktop-hd (1920×1080), desktop-2k, mobile (iPhone 15), mobile-landscape, android (Pixel 7), tablet (iPad), tablet-landscape. Also any Playwright device name ("iPhone 15 Pro Max", "Galaxy S24") or a size like "1366x768" / "390x844@3 mobile".
One file per format from the same page load: png, jpeg, webp, pdf. Each delivered file is one billable capture.
Capture the whole scrollable page (lazy images are scrolled into view first). Off = only the first screen (the viewport).
Crop full-page captures to this many CSS pixels from the top, e.g. 3000 for "the top of the page". Empty = whole page (up to "Page height limit").
Capture only this element (CSS selector, first match), e.g. "#pricing" or "main article". Pages where it never appears are reported as selector-not-found and not charged.
Capture a fixed rectangle of the page in CSS pixels: {"x": 0, "y": 0, "width": 1200, "height": 630} (e.g. social cards).
1–100. Lower = smaller files.
Override the device pixel ratio: 1 = standard, 2 = retina (4× the pixels). Leave empty to use each device's default.
Shortcut for one custom desktop viewport when "Devices" is empty (apify/screenshot-url compatible).
Height of the custom viewport (also the image height when Full page is off).
Safety cap for endless pages: longer pages are cut here and flagged "truncated". WebP files are additionally limited to 16,383 px by the format.
When the page counts as loaded. "load" suits most sites; "networkidle" waits for background requests to settle (bounded by the idle timeout, never hangs).
Wait until this CSS selector is visible before capturing (max 20 s, then captures anyway with a warning).
Wait this long after the page is ready (animations, charts). Max 60,000 ms.
Scroll down the page before capturing so lazy-loaded images and sections appear. On by default for full-page captures.
Extra wait after the lazy-load scroll.
After scrolling, wait (bounded by the idle timeout) until no requests are in flight.
Longest wait for network idle. The page is captured when it is reached.
Hard limit for one capture (load + waits + all files). A page that exceeds it is retried once, then reported as failed and not charged. The run never hangs on one page.
Click the visible "Accept" button of common consent banners (OneTrust, Cookiebot, Didomi, Quantcast, Sourcepoint, Usercentrics, ...) and hide leftovers. Never touches CAPTCHAs.
unstick = fixed headers appear once at the top, floating chat bubbles/bottom bars are removed; hide = remove all fixed/sticky elements; keep = leave as rendered.
CSS selectors to hide before capturing (ads, popups, chat widgets).
apify/screenshot-url compatible: one CSS selector list, e.g. ".popup, #chat".
Skip requests to major ad/analytics hosts: faster loads and cleaner captures.
Do not load these request types. Default: media (video/audio — a screenshot cannot show it, and it costs CPU). Add font or image for speed; set to an empty list to load everything.
[ "media"]Emulate prefers-color-scheme: dark (sites that support it render dark).
Emulate prefers-reduced-motion (fewer animations; steadier captures for monitoring).
Browser language, e.g. en-US, de-DE, ja-JP. Sites that localise by Accept-Language follow it.
Capture pages with expired or self-signed certificates.
Paper size for PDF output.
screen = the PDF looks like the page on screen; print = use the site's print stylesheet (cleaner articles).
Include background colours and images in the PDF.
For scheduled runs: compare every capture with the last delivered one and deliver it only when it changed by at least the threshold, with the diff % and a highlighted diff image. Unchanged pages cost a cheaper monitor-check. The first run is the baseline.
Separate baselines per schedule/task (e.g. "competitor-pricing").
Minimum share of the page (in %) that must differ to count as a change. Raise it for pages with rotating banners. Decimals allowed, e.g. "0.5".
0–255. How much a region's colour may shift before it counts as changed (ignores anti-aliasing noise).
Add a row with status "unchanged" (no files) for pages that did not change.
For changed pages, also store a JPEG with the changed regions highlighted (included in the price).
Also put each file (up to 5 MB) into the dataset as base64 — handy for AI agents and no-code tools that cannot download URLs.
Save files into this named key-value store (kept until you delete it, timestamped, never overwritten) instead of the run's default store.
Skip pages a site's robots.txt disallows (reported, not charged). Switch off only for sites you own or are allowed to capture.
Retries for timeouts and network errors (with a fresh proxy session when a proxy is used).
Pages rendered at once. 0 = automatic from the run memory (2 GB → 3).