AI Image Upscaler: Enlarge Photos 4x with Real-ESRGAN

Make small or low-resolution photos 2x or 4x larger with AI. Edges, eyes, text and fine structure come out crisp instead of blurry, unlike normal resizing in Photoshop, Lightroom or Canva.

• Up to 64 megapixels of output (16 MP by default, about 4900×3270)
• Batch any number of images from URLs in one run
• JPG, PNG or WebP output, transparency kept for PNG/WebP
• Pay per image, from $0.01, no subscription, no GPU, no software to install
• Powered by Real-ESRGAN, the widely used open-source upscaling model

What can I use it for?

• Photographers: enlarge older or cropped photos for prints and portfolios.
• Print-on-demand and Etsy sellers: bring artwork and photos up to print resolution (for example 300 DPI posters).
• E-commerce: sharpen small supplier product images for your store.
• Designers and marketers: upscale assets for banners, slides and social posts.
• Restoring old images: give scanned or web-sized pictures a cleaner, sharper look.

How it compares to normal resizing

Normal resizing (bicubic) only stretches the pixels you already have, so a 4x enlargement looks soft. The AI model has learned what real edges and textures look like and redraws them sharply. It works best on photos and illustrations. Very fine, irregular textures (grass, gravel, fabric) can look slightly smoothed or "painted".

How to use it

1. Paste direct links to your images in Image URLs, one per line. To upscale files on your computer, upload them to Dropbox, Google Drive (public link) or any image host first and paste the direct links.
2. Choose 4x or 2x, and the output format.
3. Click Start.
4. Open the Output tab: each row has a download link (outputUrl) for the upscaled image. Files are also in the run's Key-value store tab.

Only upload images you own or have the right to edit.

Big images

To keep runs fast and predictable, outputs are capped at Max output size (16 MP by default, up to 64 MP). If an image would end up bigger, it's shrunk first so the result fits exactly, and inputWasReduced is set to true. For example, a 2000×1500 photo at 4x would be 48 MP, so with the default cap it comes out at about 4620×3460 (16 MP).

Input example

{
    "imageUrls": [
        "https://upload.wikimedia.org/wikipedia/commons/thumb/b/bc/Parrot.red.macaw.1.arp.750pix.jpg/500px-Parrot.red.macaw.1.arp.750pix.jpg"
    ],
    "scale": "4",
    "outputFormat": "jpg",
    "quality": 92,
    "maxOutputMegapixels": 16
}

Output example

{
    "sourceUrl": "https://upload.wikimedia.org/.../500px-Parrot.red.macaw.1.arp.750pix.jpg",
    "outputUrl": "https://api.apify.com/v2/key-value-stores/ibmJh92iGZ4wOYO2o/records/0001-500px-Parrot-red-macaw-1-arp-750pix.jpg",
    "outputKey": "0001-500px-Parrot-red-macaw-1-arp-750pix.jpg",
    "scale": 4,
    "format": "jpg",
    "inputWidth": 500,
    "inputHeight": 668,
    "outputWidth": 2000,
    "outputHeight": 2672,
    "outputMegapixels": 5.34,
    "inputWasReduced": false,
    "fileSizeBytes": 1193119,
    "chargedUnits": 2,
    "processingSeconds": 7.41,
    "error": null
}

Images that fail (broken link, not an image, larger than 40 MB) get a row with an error message and are not charged.

How much does it cost?

You pay $0.01 per started 4 megapixels of output, plus a tiny start fee per run ($0.00025 per GB of memory, so $0.001 at the default 4 GB). No platform usage costs.

Example	Output size	Price
480×640 photo, 4x	1920×2560 (4.9 MP)	$0.02
400×300 image, 4x	1600×1200 (1.9 MP)	$0.01
1000×750 photo, 2x	2000×1500 (3 MP)	$0.01
750×1000 photo, 4x	3000×4000 (12 MP)	$0.03
Any image at the default 16 MP cap	up to 16 MP	$0.04

100 typical product photos (about 2–5 MP each) cost $1–2. You can set a maximum cost per run in the run options, and the Actor stops before going over it.

How long does it take?

On the default 4 GB memory, roughly 1.5–3 seconds per output megapixel: about 7 seconds for a 5 MP result and 45 seconds for 16 MP. Larger memory gives more CPU cores and shorter runs.

Use it as an API

curl -X POST "https://api.apify.com/v2/acts/piracema_labs~ai-image-upscaler/run-sync-get-dataset-items?token=YOUR_APIFY_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"imageUrls": ["https://example.com/photo.jpg"], "scale": "4"}'

The response contains the outputUrl of each upscaled image. Official Apify clients for JavaScript and Python work too, and you can connect runs to Zapier, Make, Google Drive or Dropbox with Apify integrations.

FAQ

Is this the same as Topaz Gigapixel or Photoshop Super Resolution? It's a similar AI technique, delivered pay-per-use instead of as desktop software or a subscription. Results differ by image; try a few of yours first.

Does it add fake detail? It sharpens real structure and can invent plausible fine texture, as all AI upscalers do. For legal, medical or forensic images, use the original.

Is my image data private? Your images and results are stored only in your own run's storage on Apify. The download links are long, unguessable URLs. Delete the run's storage when you no longer need it.

Which model is used? Real-ESRGAN's compact general model (realesr-general-x4v3), BSD 3-Clause licensed, by Xintao Wang et al.

Something broke or looks wrong? Open an issue on the Issues tab with a link to your run.

---

Made by Piracema Labs. Also see our Greenhouse, Lever & Ashby Jobs Scraper and Workday Jobs Scraper.