Account

What new tool do you want?

Image Tools

Image Upscaler (Enlarge Without Pixelation)

Enlarge images 1.5× to 4× with high-quality Lanczos resampling and adjustable sharpening — smooth, crisp edges instead of blocky pixels. Compare before and after, download PNG, JPG or WebP.

How to use

  1. 1Upload an image.
  2. 2Choose how much to enlarge and how much to sharpen.
  3. 3Compare with the slider and download.

Frequently asked questions

Is this AI upscaling?

No — it uses high-quality mathematical resampling, which keeps edges smooth and natural but can't invent detail that isn't in the original.

What's the largest output?

Up to about 40 megapixels (for example 4× a 1600×1000 image).

Is my image uploaded?

No — it's processed on your device.

What Lanczos resampling actually buys you over a basic resize

A browser's default image scaling (the kind you get from just setting a larger width/height in CSS or a basic canvas draw) uses simple interpolation that tends to blur edges or leave visible blockiness when enlarging significantly. Lanczos-3 resampling, which runs here in a Web Worker so the UI doesn't freeze during the computation, uses a windowed sinc function that weighs a wider neighborhood of source pixels per output pixel, producing noticeably sharper, more natural edges on enlargement. The adjustable sharpening pass afterward is a separate step that increases local contrast at edges to counteract the slight softening that's inherent to any resampling method.

The honest limit: it can't invent detail

This is mathematical resampling, not an AI model trained to hallucinate plausible texture and detail that was never captured — so a blurry or low-resolution source stays fundamentally blurry or low-resolution after upscaling, just rendered at larger dimensions with smoother edges instead of jagged ones. That's a meaningfully different (and more honest) result than AI upscalers that fabricate convincing-looking but invented detail; it's the right tool when you need a clean, predictable enlargement of something already reasonably sharp, and the wrong tool if you're hoping to recover detail from a genuinely low-quality source.

Example: enlarging a 1600×1000 web image for print-sized use

Say you have a 1600×1000 marketing graphic that looks sharp on a website but needs to go into a printed flyer at roughly double the linear dimensions. Set scale to 2×, which targets 3200×2000 (6.4 megapixels, well under the 40 MP ceiling), bump sharpening up somewhat to counteract softening at that enlargement factor, and run it. The before/after slider then lets you drag across the result to directly compare the browser's naive pixelated stretch of the original against the Lanczos-resampled, sharpened output at the same size — the difference in edge quality is usually immediately visible.

Why there's a 40-megapixel ceiling and a scaling timeout

Processing happens entirely in-browser with no server doing the heavy lifting, which keeps your photo private but means the computation is bounded by your own device's memory and CPU. The tool caps output at roughly 40 megapixels and blocks the Upscale button beyond that limit, and separately times out a running job (proportionally longer for bigger jobs, but never unlimited) rather than letting a huge 4× enlargement lock up the tab indefinitely.

Two times is the start, and forty million pixels is the ceiling

Scale starts at 2, sharpening at 40, and the download format at PNG, with JPG and WebP also available. The compare slider starts in the middle, original pixels on the left and the enlarged image on the right. Changing the scale clears the previous result so you do not download the old size by mistake. If the enlarged width times height would pass about 40 million pixels, the run is blocked as too big. If it runs too long, the message says it took too long and asks you to try a smaller scale. A failed enlarge says upscaling failed. The output size is your original width and height times the scale, shown as whole pixels.

Related tools

More: Image Tools