🖼️ Bulk Image Downloader: Website Crawl, ZIP and Image Dataset

Download images from webpages or direct URLs with bounded crawling, ZIP output, source provenance, dimensions, hashes, and honest per-image errors. Pay-per-event billing is based on successfully stored images; failed downloads are reported and are not represented as successful image records.

This Actor is a generic image downloader for public URLs. Choose a preset: Quick ZIP, Website Inventory, or ML Dataset, then override only the controls you need. It discovers <img>, <picture>, srcset, Open Graph, Twitter Card, and lazy attributes, and can follow a bounded set of page links. Each attempted image has source/input correlation, status, provenance, retry information, and an error code when the source cannot be downloaded.

---

💡 What can you do with it?

• You are building an AI training dataset. Pull thousands of product photos, real estate shots, or stock images for CLIP, DINOv2, or SigLIP. Auto-hash for dedup means you never train on the same image twice.
• You are a scraper developer. Hand the Actor a list of image URLs returned by your catalog scraper (REI, IndiaMART, eBay, Poshmark) and get back a ZIP of the binaries plus a clean metadata dataset. One Actor replaces three.
• You are an e-commerce operator. Mirror product image catalogs. Detect when a competitor swaps an image. Track pricing-page visual changes over time.
• You are an archivist or newsroom tool. Grab every image from a story page in one call. Use the per-URL ZIP mode to keep sources separated.
• You are a research analyst. Pull the full visual corpus of any public site for content analysis, brand tracking, or visual trend reports.
• You are a builder integrating via webhook. The Actor POSTs a JSON summary on completion. Pipe the dataset URL into your BigQuery, Sheets, or n8n pipeline.

Store search terms

bulk image downloader, website image scraper, download images from webpage, image URL extractor, image dataset builder.

---

🚀 How to use it

1. Open the Actor in the Apify Store and click "Try for free".
2. Paste your URLs. Mix webpages (the Actor parses the HTML) and direct image links (it downloads straight) in a single list.
3. Pick your options. Turn on SHA-256 dedup, EXIF strip, format conversion, or ZIP output as needed.
4. Click Start. The Actor fetches each URL, discovers or downloads the images, and pushes metadata to the dataset and binaries to the key-value store.
5. Download your results. Pull the dataset as JSON, CSV, or Excel. Grab the image binaries from the key-value store (links in the dataset's kv_url column). Or use the single-click ZIP download.

---

📥 Input

Field	Type	Required	Description
urls	array	Yes	List of URLs. Each can be a webpage (HTML is parsed for images) or a direct image link. Mix freely.
preset	enum	No	quickZip, websiteInventory, or mlDataset. Explicit fields override the preset.
mode	enum	No	auto (recommended, detects by extension), page (force HTML parse), or direct (force image URL).
includeSrcset	boolean	No	Discover images from srcset, picture>source, and lazy data-src. Default true.
includeOgTags	boolean	No	Discover Open Graph and Twitter Card images. Default true.
minWidth	integer	No	Skip images narrower than this. Default 0.
minHeight	integer	No	Skip images shorter than this. Default 0.
minSizeBytes	integer	No	Skip images smaller than this. Filters tracking pixels. Default 0.
maxImagesPerUrl	integer	No	Cap images per source URL. Default 1000.
maxUrls	integer	No	Cap total URLs processed. Default 10000.
crawlDepth	integer	No	Link depth for bounded page crawling. 0 stays on supplied pages.
maxPages	integer	No	Hard page cap per supplied URL.
urlGlobs	array	No	Optional URL patterns allowed during page crawling.
includeExtensions	array	No	Optional image extension allow-list such as jpg, png, webp.
includeMimeTypes	array	No	Optional response MIME allow-list such as image/jpeg.
maxSizeBytes	integer	No	Optional upper bound for downloaded image bytes.
dedupByHash	boolean	No	Compute SHA-256 of each image body and skip duplicates. Default true.
stripExif	boolean	No	Re-encode JPEGs without EXIF metadata. Default false.
convertFormat	enum	No	none, webp-to-png, or png-to-jpg. Default none.
filenamePattern	string	No	Templated filename using {slug}, {hash}, {ext}, {idx}, {source}. Default {slug}-{hash}.{ext}.
outputFormat	array	No	dataset (always), kv-store (binaries), zip (single archive), zipPerUrl (one ZIP per source), s3 (upload to bucket), webhook (POST summary on completion).
s3Bucket	string	No	Required when outputFormat includes s3. Uses standard AWS_* env vars for credentials.
webhookUrl	string	No	URL to receive a JSON run summary on completion.
maxConcurrency	integer	No	Max parallel image downloads. Default 10.
downloadTimeoutMs	integer	No	Per-image fetch timeout. Default 15000.
imageCheckMaxRetries	integer	No	Retries per failed image. Default 3.
proxyConfiguration	object	No	Optional proxy. Default off. Use residential if source sites are hotlink-protected.
failFast	boolean	No	Stop on first error. Default false.
debugLogging	boolean	No	Verbose per-image tracing. Default false.

---

📤 Output

The Actor pushes one row to the dataset per downloaded image. Binaries are written to the default key-value store under IMAGES/{filename}. Use the dataset's kv_url column to download each binary.

{
  "filename": "picsum-photos-800x600-a1b2c3d4e5f67890.jpg",
  "source_url": "https://example.com/gallery",
  "image_url": "https://picsum.photos/800/600.jpg",
  "kv_store_key": "IMG-picsum-photos-800x600-a1b2c3d4e5f67890.jpg",
  "kv_url": "https://api.apify.com/v2/key-value-stores/abc/records/IMG-picsum-photos-800x600-a1b2c3d4e5f67890.jpg",
  "content_type": "image/jpeg",
  "size_bytes": 54321,
  "width": 800,
  "height": 600,
  "format": "jpeg",
  "sha256": "a1b2c3d4e5f6789012345678901234567890abcdef1234567890abcdef123456",
  "is_duplicate": false,
  "exif_stripped": false,
  "from_srcset": true,
  "from_picture_source": false,
  "from_og_tag": false,
  "from_twitter_tag": false,
  "from_data_attr": false,
  "from_direct_url": false,
  "downloaded_at": "2026-06-20T12:34:56.000Z",
  "duration_ms": 423,
  "http_status": 200,
  "status": "SUCCEEDED",
  "ok": true,
  "inputId": "catalog-home",
  "sourceUrl": "https://example.com/gallery",
  "retrievedAt": "2026-08-01T12:34:56.000Z",
  "crawl_depth": 0,
  "discovery_method": "srcset"
}

You can download the dataset in various formats such as JSON, HTML, CSV, or Excel.

🧪 Store examples

Quick ZIP

{
  "preset": "quickZip",
  "urls": [{ "url": "https://example.com/gallery", "inputId": "gallery" }]
}

Website inventory

{
  "preset": "websiteInventory",
  "urls": [{ "url": "https://example.com/catalog", "inputId": "catalog" }],
  "urlGlobs": ["https://example.com/catalog/*"],
  "includeExtensions": ["jpg", "png", "webp"]
}

Recurring ML dataset refresh

{
  "preset": "mlDataset",
  "urls": [{ "url": "https://example.com/products", "inputId": "products" }],
  "crawlDepth": 1,
  "maxPages": 25,
  "dedupByHash": true
}

---

📋 Output data fields

Field	Description
filename	Final filename (per filenamePattern).
source_url	The page URL the image was discovered on (or its direct URL).
image_url	Final resolved image URL (after srcset expansion, redirects).
kv_store_key	Key in the run's key-value store (IMG-...).
kv_url	Signed download URL for the binary (24-hour default).
content_type	MIME type (e.g. image/jpeg, image/webp).
size_bytes	Downloaded size.
width	Image width in pixels (from sharp metadata).
height	Image height in pixels (from sharp metadata).
format	Normalized format: jpeg, png, webp, gif, svg, avif, bmp, ico, other.
sha256	Content hash (when dedupByHash=true).
is_duplicate	True if hash matched a previously-seen image in this run.
exif_stripped	True if JPEG was re-encoded to remove EXIF.
from_srcset	True if discovered via srcset / picture / data-srcset.
from_picture_source	True if discovered via <picture><source>.
from_og_tag	True if discovered via <meta og:image>.
from_twitter_tag	True if discovered via <meta twitter:image>.
from_data_attr	True if discovered via lazy data-src / data-srcset.
from_direct_url	True if the URL was treated as a direct image (mode=direct/auto).
downloaded_at	ISO timestamp of the download.
duration_ms	Time to fetch + process.
http_status	HTTP response code when the image server returned one.
error	Per-image error string when the attempt failed; omitted on a successful row.
status / ok	Honest success state for the image attempt. Failed attempts include an errorCode and omit unavailable image fields.
inputId / sourceUrl / retrievedAt	Stable input correlation, canonical source page, and retrieval timestamp.
finalUrl / retryCount / errorCode	Redirect destination and retry/error evidence when available.
alt / title / page_title	Source HTML metadata when published by the page.
crawl_depth / discovery_method	Bounded crawl provenance.

---

💰 Pricing

The local PPE configuration bills the Stored image event only after an image is successfully stored. Apify applies the account tier to the event price; the current Store tiers are:

Volume	What you pay
Free tier	$0.89 per 1,000 stored images
Bronze	$0.82 per 1,000 stored images
Silver	$0.74 per 1,000 stored images
Gold and above	$0.67 per 1,000 stored images

Billing is tied to the stored-image event, not to input pages or attempted links. There is no standby mode or separate scheduled-run price.

---

⭐ Enjoying Bulk Image Downloader?

⭐ ⭐ ⭐ ⭐ ⭐ Save hours of manually right-clicking and saving images one by one from any site. A 5-star rating takes 10 seconds and helps other AI dataset builders and e-commerce operators find it. Your feedback also tells us what to build next.

★  Rate this Actor on Apify

---

🛠️ Tips and advanced options

• Set includeSrcset to false if you only want the page's primary images. This skips lazy data-src and responsive variants, which is faster on heavy pages.
• Use minSizeBytes to filter tracking pixels. A typical tracking pixel is under 1KB. Set minSizeBytes: 2000 to skip them.
• Use minWidth and minHeight to focus on useful images. Set minWidth: 400 to skip thumbnails and avatars.
• Pick the right output mode. zip for a single archive, zipPerUrl to keep source pages separated, s3 to push directly to your training bucket.
• Pair with a catalog scraper. Run one of our catalog scrapers (REI, IndiaMART, eBay) first, then feed the image URLs to this Actor for a complete e-commerce dataset.
• Schedule weekly runs to refresh your image corpus. Most product catalogs update slowly; daily is overkill.
• Use SHA-256 dedup within a run. Hashes are stable for downstream reconciliation; cross-run history belongs in the dataset or an external state store.

---

❓ FAQ

Is this Actor legal to use? The Actor downloads images that are publicly accessible. You are responsible for ensuring your use case complies with the source site's Terms of Service and applicable copyright laws. Do not use the Actor to bypass access controls, scrape private content, or violate copyright.

Why does it work on any site? The Actor is generic. It fetches the URL you give it, parses the HTML for image tags, and downloads the images it finds. There is no per-site configuration.

Does it execute JavaScript? No. Single-page apps that render images via React/Vue hydration will return an empty image list. If your target site is a SPA, use a Playwright-based scraper first to get the image URLs, then pass them to this Actor with mode: 'direct'.

Do I need a proxy? No. Most public sites serve images to any client. Default useApifyProxy: false works perfectly. If your source site is hotlink-protected, set residential proxy as an opt-in via the proxyConfiguration field.

What is the largest image it can handle? Sharp auto-streams, so peak memory is around 5x the size of the largest single image. A 50MB image is fine. A 500MB image may cause memory pressure on smaller container sizes.

Does the EXIF strip work on PNG or WebP? No, EXIF strip is JPEG-only. PNG metadata stripping is a v2 feature.

How is billing calculated? The Stored image PPE event is emitted only after the binary is successfully stored. Failed image attempts stay in the dataset with an error code and are not billed as successful image events.

Can I get a single ZIP of all images? Yes. Set outputFormat: ['dataset', 'kv-store', 'zip']. The ZIP is written to OUT-images.zip and is also linked in the dataset summary.

Can I push directly to S3? Yes. Set outputFormat: ['dataset', 's3'], fill in s3Bucket, and set AWS_ACCESS_KEY_ID, AWS_SECRET_ACCESS_KEY, and AWS_REGION as Apify Secrets. Each image uploads to s3://{bucket}/images/{filename}.

Can I get a webhook on completion? Yes. Set outputFormat: ['dataset', 'webhook'] and fill in webhookUrl. The Actor POSTs a JSON summary with run stats (counts, errors, total size) to the URL when the run finishes.

---

🛡️ Disclaimers and support

• Disclaimer: This Actor retrieves publicly accessible images. Make sure your usage complies with the source site's terms of service and applicable copyright laws. The Actor is a generic utility and does not bypass authentication, paywalls, or access controls.
• Support: Open an issue from the Issues tab for bug reports or feature requests. Custom scrapers and integration help are available on request.

🔗 Other actors

• Google Lens OCR API: Sub-second Image to Text ↗ - extracts text from any image via Google Lens OCR.
• Google Lens Search: Reverse Image Finder & OCR ↗ - runs reverse image search and visual matches through Google Lens.
• Loom Video Downloader: MP4, transcript and metadata in one run ↗ - downloads Loom videos as MP4 with transcript and metadata.
• Streamable Video Downloader: Direct MP4 links in seconds ↗ - resolves direct MP4 download links from Streamable.
• Loom Transcript Scraper ↗ - pulls transcripts, captions, and metadata from public Loom videos.