Scale an Image to 1.5×, 2×, or 3×

1. 1. Loads the local image through an object URL.
2. 2. Reads its decoded width and height.
3. 3. Multiplies both dimensions by 1.5, 2, or 3 and rounds them.
4. 4. Creates a destination canvas with the enlarged dimensions.
5. 5. Applies the selected method.
6. 6. Encodes the destination as PNG or JPEG and downloads a filename such as scaled-2x.png.

Nearest neighbor disables canvas image smoothing. Bilinear enables smoothing and requests high smoothing quality. Lanczos-style mode reads source pixels into an array, evaluates nearby samples within a three-pixel radius, writes a new destination pixel array, and yields periodically while processing rows.

Pixel count grows with area, not with one dimension alone: 1.5× means 2.25× as many output pixels, 2× means 4×, and 3× means 9×.

An uncompressed RGBA pixel buffer commonly uses four bytes per pixel. A 4000 × 3000 source contains 12 million pixels, or about 48 MB for one basic RGBA buffer before overhead; a 3× output contains 108 million pixels, requiring about 432 MB for one such buffer. Lanczos-style processing may hold the source image, source pixel data, destination data, canvases, and encoded output at the same time, so actual memory use can be much higher. Browser canvas dimension limits also vary, and a valid source may fail when the enlarged destination exceeds the current limit.

Text and screenshots. Scaling text does not equal rerendering the original font at a larger size. Nearest neighbor keeps hard pixel boundaries but looks blocky; bilinear smooths but can blur letters; Lanczos-style may stay sharp but can add halos. When possible, recreate the screenshot at a higher native display scale or export the original vector/text design again.

Pixel art. Pixel art relies on a deliberate grid, so use nearest-neighbor scaling with integer factors such as 2× or 3×. A 1.5× scale does not map each source pixel to an equal integer-sized block, so some pixels occupy different output widths or heights, creating uneven shapes. Use PNG input and output, and inspect edges at 100% zoom.

Canvas re-encoding focuses on decoded pixel values. Do not assume the output retains EXIF metadata, GPS coordinates, camera settings, IPTC or XMP fields, ICC color profiles, copyright fields, original timestamps, or embedded thumbnails. This can improve privacy when location metadata is removed, but it can disrupt professional color and asset-management workflows — keep the source and verify output requirements.

Increasing pixel dimensions changes the file's pixel grid; it does not create new optical information. Print quality depends on original detail, final physical dimensions, effective pixels per inch, viewing distance, printer and paper, and sharpening. Scaling may satisfy a software minimum or reduce visible pixelation at a layout size, but it cannot guarantee a sharp print from a poor source.

Image contents are decoded, scaled, and encoded in browser memory and are not sent to WeConvertFiles for scaling. This tool uses browser APIs and local JavaScript processing rather than a remote AI service.

If a visitor consents to analytics, separate usage information such as visits, clicks, device details, or scaling events may be collected. Those analytics do not receive the image contents.

Browser extensions, malware, shared devices, downloaded files, backups, and cloud-synchronized folders can still expose images independently of scaling.