Website Screenshot: Full-Page PNG/JPEG with Chinese, Japanese, Korean and Emoji Fonts

Capture screenshots of a list of web pages for website monitoring, QA and visual testing, archiving or internal reference. You can capture the full page or just the viewport, at the width and height you choose (within the limits below), as PNG or JPEG. Every image is saved in the run's key-value store, and the dataset gets one row per URL with a direct link to the image, the final URL, the page title and the HTTP status.

Chinese (Traditional and Simplified), Japanese and Korean pages render with Noto CJK fonts, not empty boxes ("tofu"), and emoji render in colour. The Actor respects each site's robots.txt by default.

Features

• Full page or viewport. With fullPage: true, the Actor scrolls first so lazy-loaded images appear, then captures the whole page up to maxHeightPx (default 16,000 px). Longer pages are cut at that height and flagged with truncatedToMaxHeight. The image stays viewportWidth wide, including on pages with horizontal overflow.

• Custom viewport. viewportWidth (320–3840) and viewportHeight (240–2160), plus deviceScaleFactor 1–3 for retina images.

• PNG or JPEG, with adjustable jpegQuality.

• Wait options:

  • waitUntil: load (default), domcontentloaded or networkidle;
  • an extra delayMs after loading;
  • waitForSelector waits until a specific element appears.

  The Actor also waits for web fonts to finish loading.

• Clean shots. Hide cookie banners and chat widgets with hideSelectors (CSS selectors). Use it for cookie banners and chat widgets only. Do not use it to hide paywalls, registration walls, or copyright and attribution notices. Scrollbars are hidden.

• Locale and time zone. For example zh-TW and Asia/Taipei, so the site serves the language and dates you need.

• Built for real-world sites.

  • Each URL gets a fresh, isolated browser context.
  • Navigation is fault-tolerant: if the load event does not fire because of a slow tracker or a stuck video, the Actor still captures what has rendered and adds a note.
  • Failed pages are retried with a fresh context (maxRetries), and every URL has an overall time budget.
  • A crashed browser is restarted automatically.

• Polite by design.

  • robots.txt is checked before each page and for every redirect hop, using the rules for RondelBot (or for * if there is no RondelBot group). If a site's robots.txt cannot be fetched because of a server error or timeout, the site is treated as disallowed.
  • respectRobotsTxt is on by default. Turning it off is your decision and your responsibility; do it only for sites you own or have written permission to capture.
  • At least 1 second between page loads on the same site; a longer robots.txt Crawl-delay is honoured while respectRobotsTxt is on.
  • The browser identifies itself with a RondelBot token in its user agent.
  • Anti-bot challenges are reported, not bypassed.
  • Site owners can opt out with User-agent: RondelBot / Disallow: / in robots.txt.

• Network safety. Page URLs that point to private or local network addresses are refused.

Output

One dataset row per URL:

{
  "url": "https://www.python.org/",
  "finalUrl": "https://www.python.org/",
  "status": "ok",
  "httpStatus": 200,
  "title": "Welcome to Python.org",
  "screenshotKey": "0012-www-python-org-1ab11d99.png",
  "screenshotUrl": "https://api.apify.com/v2/key-value-stores/<store>/records/0012-www-python-org-1ab11d99.png",
  "format": "png", "fullPage": true,
  "viewportWidth": 1280, "viewportHeight": 800,
  "imageWidth": 1280, "imageHeight": 2525,
  "truncatedToMaxHeight": false,
  "bytes": 477812,
  "notes": [],
  "error": null,
  "charged": true,
  "seconds": 17.4,
  "takenAt": "2026-10-17T07:51:24+00:00"
}

status is one of:

• ok: the screenshot was taken;
• ok-http-error: the page answered 4xx/5xx; the error page is still captured so you can see it, and it is free;
• blocked-by-robots-txt;
• blocked-by-anti-bot;
• invalid-url;
• unreachable: the domain does not exist or could not be resolved;
• error;
• skipped-spending-limit;
• duplicate: the same page appeared earlier in your list (see below); the row has duplicateOf and is free.

Duplicate URLs. Before anything is loaded or charged, URLs are normalised for duplicate detection: surrounding spaces trimmed, scheme and host lower-cased, the #fragment, utm_* and common click-tracking parameters (such as fbclid and gclid) removed, default ports dropped, and /path and /path/ treated as the same page. A URL that matches an earlier one gets status duplicate with duplicateOf naming the first occurrence, is not loaded and is not charged. The first occurrence is captured from the URL as you entered it. http:// and https:// versions of a page are not treated as duplicates.

The Overview dataset view shows the screenshots as images. Files can also be downloaded from the run's key-value store (Storage tab) or through the API.

How to use

1. Add your pages to URLs, one per line.
2. Choose full page or viewport, the size and the format.
3. Run it. Each row's screenshotUrl opens the image.

Examples:

• Mobile-width full page: {"urls": ["https://www.ikea.com.tw/zh"], "viewportWidth": 390, "viewportHeight": 844, "deviceScaleFactor": 2}
• Small JPEG thumbnails of the first screen only: {"urls": [...], "fullPage": false, "format": "jpeg", "jpegQuality": 70}
• Taiwanese locale: {"urls": [...], "locale": "zh-TW", "timezoneId": "Asia/Taipei"}

Some websites block cloud servers: what happens then

Be aware of this before you start:

• Some websites refuse visitors coming from cloud data centres, which is where Apify runs. Large corporate, government and news sites are the usual examples.

• This Actor uses no proxies and no disguise. It visits from Apify's own servers, tells the site who it is through its user agent (RondelBot), and does not try to get past anti-bot checks.

• When a site refuses, the URL is marked as failed with a clear reason and is NOT charged:

  • blocked-by-anti-bot for a challenge page;
  • ok-http-error for a 403 or other error page, which is still captured so you can see what the site returned;
  • error with the browser's message for a dropped connection;
  • blocked-by-robots-txt when the site's robots.txt disallows the page.

  Every row has charged: true or charged: false, so you can see exactly what you paid for.

In an internal test (2026-10-17, 30 popular Taiwanese and international sites), 4 sites refused cloud visitors. All 4 were reported this way and none were charged.

Pricing

Pay per event. You pay only for screenshots of pages that loaded with an HTTP status below 400:

Event	Price
Screenshot (screenshot)	$0.004 per screenshot stored ($4 per 1,000)
Actor start (apify-actor-start)	$0.0005 per GB of run memory, charged once per run ($0.002 at the default 4 GB). Put many URLs in one run to spread it.

Free (not charged):

• pages blocked by the site (anti-bot, cloud-IP blocks, dropped connections) or by robots.txt;
• pages that answer with an HTTP error (the error page is still captured);
• invalid URLs, unreachable domains, failed loads and duplicate URLs (duplicate);
• URLs skipped because of your spending limit.

What the price includes

• Full-page capture built for heavy pages. The Actor scrolls so lazy images load, captures up to 16,000 px by default (30,000 px maximum), and keeps the image at the width you asked for. In an internal test (2026-10-17, 30 sites), all 26 sites that served their page produced a correct screenshot. These included Stripe, Apple, GitHub, BBC and a Wikipedia article 75,481 px tall (captured up to the height limit), with no crashes or timeouts in that test. Results on other sites may differ.
• Your viewport, your format. Width 320–3840 and height 240–2160, 1–3× retina, and PNG or JPEG with adjustable quality.
• Chinese, Japanese, Korean and emoji fonts. Noto CJK and Noto Color Emoji fonts are built in to avoid empty "tofu" boxes on Asian-language pages.
• Stability features. Each URL gets a fresh, isolated browser, with a retry, a time budget per URL and automatic browser restart. If a page's load event does not fire, the Actor captures what has rendered instead of failing.
• Failures are free, and every URL gets a transparent status: HTTP status, final URL, title, notes such as "page cut at maxHeightPx", and whether it was charged.
• Spending limit. Set Maximum cost per run and the Actor stops before going over it. If a run is restarted, URLs already in the dataset are not captured or charged again.

Memory: the default 4 GB processes about 6 pages in parallel (the Actor chooses about one page per 680 MB unless you set maxConcurrency). 2 GB works for a few URLs, and more memory finishes large batches faster.

Limitations

• Logins, paywalls and anti-bot walls. The Actor has no login, cookie, header or proxy options and does not try to get past logins, paywalls or anti-bot walls. Such pages usually fail (reported and free). If the site instead serves a normal page, such as its login form or paywall notice, that page is captured as an anonymous visitor sees it and is charged like a normal page.
• Sites that block cloud IPs cannot be captured (reported and free).
• Very tall pages are cut at maxHeightPx (maximum 30,000 px).
• Cookie and consent banners are shown unless you list them in hideSelectors.
• Videos and animations are captured as a single still frame.
• Heavy pages can take a long time each. In an internal stress test (2026-10-17, 200 URLs at the default 4 GB), the whole run took 1,244 seconds. You pay per screenshot, not per second.

Responsible use and legal notice

This Actor only visits the URLs you give it, from Apify's servers, with an honest user agent (RondelBot). By default it obeys robots.txt (including redirects) and waits at least 1 second between page loads on the same site. It has no login, cookie, header or proxy options and does not try to get past logins, paywalls, CAPTCHAs or other anti-bot protections.

You are responsible for how you use it and for what you do with the results. Before running it, make sure you have the right to access and copy the pages you submit, and that your use complies with the websites' terms, copyright law and data-protection law (for example GDPR, CCPA, or Taiwan's Personal Data Protection Act). Turning off respectRobotsTxt is your decision and your responsibility; do it only for sites you own or have written permission to access.

Results can contain personal data (for example names or photos shown on a page). Do not collect personal data unless you have a lawful basis to do so. Your inputs and results are stored only in your own Apify account (the run's dataset and key-value store); the Actor does not send them to the developer or to third parties.

Screenshots are copies. A screenshot reproduces the page's text, images and design, which may be protected by copyright. Keep screenshots for purposes such as monitoring, testing, archiving or internal reference, and do not republish them without permission. Use hideSelectors only for cookie banners and chat widgets, not to hide paywalls, registration walls or copyright and attribution notices.

Site owners can opt out by adding User-agent: RondelBot and Disallow: / to their robots.txt; the change applies from the Actor's next visit to the site.

This notice is not legal advice. If you are unsure whether your use is lawful, consult a lawyer.

負責任使用與法律聲明

本 Actor 只會從 Apify 伺服器造訪您提供的網址，並以誠實的使用者代理名稱（RondelBot）表明身分。預設遵守 robots.txt（含轉址），同一網站的頁面載入之間至少間隔 1 秒。本 Actor 不提供登入、cookie、自訂標頭或 proxy 功能，也不會嘗試繞過登入、付費牆、驗證碼或其他防爬機制。

使用方式與結果的用途由您自行負責。 執行前，請確認您有權存取及複製所提交的網頁，且您的使用符合各網站的使用條款、著作權法與個人資料保護法規（例如 GDPR、CCPA 或我國《個人資料保護法》）。關閉 respectRobotsTxt 是您的決定與責任，請僅用於您擁有或已取得書面許可的網站。

結果可能含有個人資料（例如網頁上的姓名或照片）。除非您有合法依據，請勿蒐集個人資料。您的輸入與結果只存放在您自己的 Apify 帳號中（該次執行的 dataset 與 key-value store），本 Actor 不會將其傳送給開發者或第三方。

截圖即重製。 截圖會重製網頁上的文字、圖片與設計，這些可能受著作權保護。請將截圖用於監控、測試、存檔或內部參考等用途，未經許可請勿公開轉載。hideSelectors 僅用於隱藏 cookie 橫幅或客服視窗，不得用來隱藏付費牆、註冊牆或著作權與出處標示。

網站管理者可在 robots.txt 加入 User-agent: RondelBot 與 Disallow: / 拒絕本 Actor 造訪；自本 Actor 下次造訪該網站起生效。

本聲明並非法律意見；如不確定您的用途是否合法，請諮詢律師。