Website Screenshot Generator — Full Page, Pay Only If Usable

Full-page screenshots of any list of URLs. Lazy content is loaded before the capture, cookie banners are removed, and sticky headers are stopped from repeating down the image.

You are charged for images you can use. When a page comes back blank, blocked, bot-checked, behind a login wall or as a 404, you get the reason in the output row and that URL costs nothing.

What you pay for

One flat price per usable screenshot. No free-tier surcharge, no per-row fee, no charge for anything the run could not capture.

outcome	charged
a real screenshot of a real page	yes
the site returned 4xx or 5xx	no
the page rendered blank or nearly empty	no
a bot check or CAPTCHA came back instead of the page	no
a consent or paywall overlay survived removal and still covers the page	no
a login wall with nothing behind it	no
the page painted almost nothing	no
the URL timed out, failed to load, or the domain does not exist	no
your CSS selector was not on the page	no
the run budget ran out before that URL started	no

Every one of those reasons is written into the results table, the gallery and the run summary. The detector fails open on purpose: a signal we could not measure never withholds payment.

What you get

The run opens on Screenshots — a gallery of every capture in the run, click any image for the full-size version. Anything that was not charged for carries an amber border and says why in plain words. The gallery link is signed and opens with no login, so you can send it to a client as it is.

Also in every run:

• Image files — the PNG/JPEG/WebP records themselves, named 001-example-com.png and so on.
• Results table — one row per URL with a direct image link, page title, pixel size, file size, timing, whether it was charged for, and if not, why.
• Run summary JSON — captured, charged, not charged, failed and skipped counts, plus the settings the run used.

Image links are signed URLs. They open with no Apify account, so they can go straight into a report, a sheet, or another system.

Input

{
  "urls": [
    "https://www.gov.uk",
    "https://vercel.com"
  ],
  "fullPage": true,
  "format": "png"
}

urls is the only field without a working default — a first run never fails because something was left blank.

Options

Page preparation — remove cookie and consent banners (on), load lazy content before capturing (on), stop sticky headers repeating (on), block ad networks (off).

Image — full page or viewport, PNG / JPEG / WebP, viewport width and height, pixel density up to 3x, JPEG/WebP quality, maximum page height, transparent background.

Part of a page — capture one CSS selector such as main or #pricing, or a fixed region in page coordinates such as {"x":0,"y":0,"width":1200,"height":630} for a social preview.

Loading and timeouts — what counts as loaded (load, domcontentloaded, networkidle, commit), wait for a selector, a fixed extra delay, per-URL time limit, ignore certificate errors.

Batch — URLs at a time, a run budget that stops starting new work rather than overrunning, and retries per URL. A domain that does not exist is never retried.

Advanced — colour scheme, locale, time zone, user agent, page preparation budget.

What this cannot do

Read this before you buy. Every limit here is one we measured.

• Sites that block datacenter traffic will not be captured. This Actor runs from Apify's data centre with no residential proxy option. Cloudflare-protected sites — Stack Overflow is the standing example — return a bot check instead of the page. You get http-403 or bot-check as the reason and you are not charged, but you also do not get the screenshot.
• A few consent dialogs cannot be removed without breaking the site. Sourcepoint keeps its state on the page's root element; removing it blanks the whole site, so it is left alone. theguardian.com is the known case: the article renders and the capture is useful, with the dialog still over part of it. Most CMPs (57 selector families) are removed cleanly.
• We remove banners, we never click them. Clicking "Accept all" would consent to tracking on your behalf. A site that renders nothing at all until consent is given comes back with the reason, unbilled.
• Very tall pages are clipped, and say so. The default ceiling is 12,000 pixels and the maximum is 30,000; browsers cannot reliably capture much beyond 16,000. A clipped row is marked truncated. Nothing is ever silently cut.
• No logged-in pages. There is no cookie, session or credential input. Anything behind a login comes back as login-wall and is not charged.
• No interaction before the capture. No clicking, typing, scrolling to a specific element or dismissing modals beyond consent handling. You can wait for a selector and add a fixed delay.
• Images only. No PDF export, no HTML archive, no text extraction.
• One run is one moment. There is no change tracking or visual diff between runs.
• Apify stops any run at 5 minutes. The run budget defaults to 4 minutes and stops starting new URLs before that. For large batches, split them across runs or raise the budget and memory.
• Concurrency 2 is the measured default. It is sized for 2 GB of memory. Raising it without raising memory makes captures slower and more likely to time out.

Pricing

$0.02 per usable screenshot. One price for every buyer on every plan. Apify bills its own compute on top, at cost.

A measured six-URL run — four capturable, two not — costs $0.08 in screenshots plus $0.012 to $0.022 in Apify compute, and finishes in about seventy seconds. The two that failed cost nothing.

Notes

• Set format to jpeg for long pages — a full-page PNG of a tall site can reach several megabytes, and JPEG or WebP cuts that sharply.
• Use width: 390 for a mobile-width layout.
• blockAds is off by default: a cookie banner is an artefact of our visit, and an ad is part of the page as published. Turn it on for a cleaner marketing shot.