Image Resizer & Converter: Bulk WebP/AVIF Compressor

Use this image resizer, image compressor and image converter to process pictures in bulk: a long list of image links, every image on a web page, or the results of another Actor, all in one run. It converts to WebP, AVIF or JPG/PNG, resizes to one or several sizes, and can shrink each file to a size limit you set. Every result is a downloadable file plus a row that shows how many bytes it saved.

For each image you get:

• Resized, compressed files as WebP (default), AVIF, JPEG, PNG, or the same format as the source
• Several sizes per image in one run: a width (1280), a height (x600) or a box (800x800, fitted with inside, cover, contain or fill)
• Maximum file size in KB: quality is stepped down until the file fits, and files that cannot fit are flagged with targetMissed
• Metadata cleanup: EXIF and GPS data are removed and the image is rotated upright. The colour profile is kept.
• Duplicate finder: a 64-bit perceptual hash flags look-alike images in duplicateOf, even across different sizes or formats
• A download link, new and original dimensions and bytes, and savedPercent for every file
• An optional ZIP of all files (up to 300 MB)

Inputs can be JPEG, PNG, WebP, AVIF, GIF (first frame only), TIFF or SVG, up to 40 MB each.

It costs $1 per 1,000 images (one output file is one image), with no start fee. Failed, unreadable and skipped images are free.

What people use it for

• Responsive image sets. Web developers list sizes such as 400, 800, 1600 and get one WebP file per width, ready for a srcset.
• Store product photos. Resize to square thumbnails (800x800 with Fit set to cover or contain) and let the duplicate finder flag product images that are the same picture re-saved or resized.
• Site speed and SEO work. Put a page URL in Web pages and leave Sizes empty. Each image's savedPercent shows how much smaller it would be as WebP, and the SUMMARY record has the overall figure.
• Automation. Chain it after a scraper by giving the scraper's dataset ID and the field that holds the image links, or start it from code through the API.

How it works

1. Collect. Image links are used as given (a link listed twice is processed once). For a web page, the HTML is read and every img source is taken (the largest srcset entry, lazy-load data-src), plus picture sources, og:image and twitter:image, with the alt text kept. For a dataset, the field you name is read: dots reach nested fields (images.0.url) and a field holding a list of links works too.
2. Download. Each image is fetched with up to 3 attempts on timeouts, rate limits and server errors. Several images run at once (Parallel downloads, 4 by default, up to 16).
3. Process. The image is rotated upright, resized and encoded. An image smaller than the requested size keeps its size unless Allow enlarging is on.
4. Save. Each file goes to the run's key-value store and one dataset row is added per file. With ZIP on, images.zip is built at the end. The SUMMARY record has the totals and the ZIP link.

Pages are read as plain HTML and no scripts are run, so images a page adds only with JavaScript, or sets as CSS backgrounds, are not found. Use Only images hosted on the same site to skip ads and third-party images, and Skip source images smaller than to skip icons and tracking pixels.

Input example

{
  "pageUrls": ["https://example.com/shop/blue-mug"],
  "sameSiteImagesOnly": true,
  "minSourceKb": 5,
  "format": "webp",
  "sizes": ["1280", "800x800"],
  "fit": "cover",
  "quality": 80,
  "zip": true
}

Output example

{
  "sourceUrl": "https://example.com/media/blue-mug-front.jpg",
  "pageUrl": "https://example.com/shop/blue-mug",
  "altText": "Blue ceramic mug, front view",
  "status": "ok",
  "outputUrl": "https://api.apify.com/v2/key-value-stores/8Wj3xKq1hZy4mTn0b/records/blue-mug-front-3f9a1c2e-800x800.webp",
  "outputKey": "blue-mug-front-3f9a1c2e-800x800.webp",
  "size": "800x800",
  "format": "webp",
  "width": 800,
  "height": 800,
  "bytes": 38900,
  "quality": 80,
  "targetMissed": false,
  "originalFormat": "jpeg",
  "originalWidth": 2400,
  "originalHeight": 1600,
  "originalBytes": 612340,
  "savedPercent": 93.6,
  "hasExif": true,
  "perceptualHash": "c3a1b4f0e2d19687",
  "duplicateOf": null,
  "error": null
}

Download the dataset as CSV, Excel or JSON, or pipe it to Google Sheets, Make, n8n or Zapier. The image files are in the run's key-value store under outputKey.

A source that fails or is skipped gets a row with status set to error or skipped and the reason in error. savedPercent is negative if a file ends up larger than its source (for example when enlarging). The SUMMARY record counts sources, files written, failures, duplicates, skipped images, and total bytes in and out.

Pricing

Pay per event, no start fee:

Event	Price
Output image file written	$0.001

These are not charged: downloads that fail, files that are not readable images, images skipped by Skip source images smaller than, and extra sizes that come out with the same pixel dimensions as one already made. Example: a 600 px wide source asked for 400, 800 and 1600 gives a 400 px file and one 600 px file (it is not enlarged), so 2 files are charged.

• 1,000 product photos, each as a square 800x800 thumbnail and a 1600 wide version: 2,000 files, $2.00.
• One web page with 60 images: 2 fail to download and 8 are icons under your size limit. The other 50 go to one size: 50 files, $0.05.

Duplicates are flagged, not skipped, so they are charged like any other output file. Set a maximum charge per run in the run options and the Actor stops cleanly when it is reached.

Limits and notes

• Animated GIFs: only the first frame is kept, and the output is a still image.
• HEIC/HEIF photos are not supported. Convert them to JPEG first.
• Same as the source keeps JPEG, PNG, WebP and AVIF. GIF, TIFF and SVG come out as PNG (SVG is turned into a bitmap).
• PNG output: a quality below 100 uses palette compression (up to 256 colours), which makes small files but can band gradients. Use 100 for full-colour lossless PNG.
• File size limit works for WebP, AVIF and JPEG, and never goes below quality 10.
• Duplicates: the first look-alike to be processed is reported as the original. With parallel downloads, that is not always the first one in your list.
• Some sites block automated downloads (hotlink protection, logins). Those images come back as error rows. Downloads send a User-Agent that names this Actor.

FAQ

Is my data kept? Images are downloaded and processed inside your run, and the results are written to the run's key-value store and dataset in your Apify account. They stay there under your Apify data retention settings. The Actor does not send images anywhere else.

Which images may I process? Only images you own or have the right to modify and re-host. You are responsible for how you use the results and for following the terms of the sites you take images from.

Why WebP? Every current browser supports it, and it is usually 25 to 35% smaller than JPEG at the same visual quality. AVIF can come out smaller still but is slower to encode. Pick JPEG or PNG when the place you upload to needs them.

Can I call it from code? Yes. Start it with the Apify API, the JavaScript or Python client, a schedule, or an integration, then read the dataset when the run finishes. The Actor ID is open-data-tools/image-resizer-webp-converter.

How much memory do I need? Raise the run's memory for very large photos or when you raise Parallel downloads. The ZIP is built in memory, so a large ZIP needs more memory too.

Questions or an image type that fails? Open an issue on the Actor page.