Pixels or percent · Contain or cover · Optional no-upscale

Image Resizer by Pixels or Percent

Resize a static JPEG, PNG, or WebP with explicit geometry. Set dimensions in pixels or percent, preserve the aspect ratio, fit inside a box, fill and crop a box, or prevent accidental enlargement.

Resize an imageLocal preview · No sign-up
contain fit
source  2400 × 1600
box     1200 × 1200
result  1200 × 800

Aspect ratio preservedNo crop in contain mode

Resize one image or a local batch of up to 20

Define the output canvas before you download

Select validated static images, choose pixel or percentage sizing and a fit mode, then compare source and output dimensions for each result. Downloads are created only after the browser has decoded and re-encoded them.

Loading the image resizer…

Enable JavaScript to select, validate, resize, preview, and download an image. Geometry rules and format limitations remain readable below.

Choose the rule that matches the destination

Pixels, percentages, contain, and cover solve different sizing problems

A width and height can describe an exact output, a bounding box, or a crop frame. The selected mode determines which interpretation applies.

ControlResultBest used for
PixelsTargets explicit width and height values.CMS slots, social assets, thumbnails, and interface specifications.
PercentScales from the decoded source dimensions, such as 50% on each axis.Making a proportional half-size or quarter-size copy without calculating pixels.
ContainFits the entire image within the box; one axis may be smaller than requested.Product photos, documents, and any image that must not be cropped.
CoverFills the box and crops centered overflow on one axis.Avatars, cards, banners, and fixed-ratio frames that cannot have gaps.

Protect proportions and real detail

Aspect lock and no-upscale prevent common resize mistakes

Aspect lock couples width and height to the decoded source ratio. Changing one dimension recalculates the other, preventing an accidental stretch that makes circles oval and faces unnaturally narrow. Contain and cover also preserve the ratio, but differ in whether the entire source or the entire destination box wins.

No-upscale caps the scale at 100% of the source. For example, a 640 × 480 image contained in a 1200 × 1200 box remains 640 × 480 when no-upscale is enabled. Without the cap, interpolation can produce a larger file and more pixels, but it cannot recreate texture, focus, or lettering that the source never captured.

Cover mode necessarily discards pixels outside the destination crop. Check the preview for subjects near the edges; a general-purpose centered crop cannot understand which face, label, or object is important.

Aspect lock
Preserves width ÷ height while one dimension changes.
No-upscale
Stops the computed scale from exceeding the original dimensions.
Interpolation
Estimates new pixel values; it does not recover missing source detail.
Centered crop
Removes equal overflow where possible; inspect edge content before saving.

Geometry first, encoding second

Choose Original format, JPEG, PNG, or WebP deliberately

Original format

The output MIME type follows the validated JPEG, PNG, or WebP source. Resizing still requires decoding and re-encoding, so the result is not byte-identical and source metadata is not promised.

JPEG

Useful for opaque photographs. JPEG cannot store an alpha channel, so transparent source pixels are composited over the background color selected in the tool.

PNG

Keeps decoded alpha and sharp pixel boundaries through lossless encoding, but photographs can produce much larger files than lossy JPEG or WebP.

WebP

Can combine browser-controlled lossy encoding with transparency. Browser encoder support is checked: a fallback blob with the wrong MIME type is rejected.

Validate before allocating a canvas

Supported input and browser processing limits

CheckContractReason
SignatureMagic bytes must identify a static JPEG, PNG, or WebP.Extensions and declared MIME types can be renamed or incorrect.
File sizeThe encoded input may be no larger than 25 MiB.Decoded pixels need substantially more browser memory than compressed bytes.
Pixel countDecoded width × height may be no larger than 40 megapixels.This bounds the source canvas before transformation.
UnsupportedHEIC, AVIF, animated PNG/WebP, and other signatures are rejected.The tool does not silently export one frame from an animation.
Output MIMEThe browser encoder must return the requested JPEG, PNG, or WebP type.Silent encoder fallback is reported as an error, not mislabeled for download.

Start from the destination

Practical resizing examples

  1. 1

    Uncropped catalog image

    Choose a 1200 × 1200 pixel box with contain. A 3:2 landscape source becomes 1200 × 800; the whole photograph remains visible.

  2. 2

    Square profile image

    Choose 512 × 512 with cover. Inspect the centered crop, especially faces or logos near an edge, before exporting.

  3. 3

    Smaller review copy

    Choose 50% with aspect lock and no-upscale. Each dimension is halved, so the output has one quarter of the source pixel count.

Local pixels, bounded claim

Resizing happens in the current browser tab

The selected file is read with browser file APIs, decoded, drawn to a canvas at the computed geometry, encoded, previewed, and downloaded without intentionally sending its image bytes to an image-processing service. Large dimensions can still consume significant local memory; a decoded RGBA canvas commonly requires about four bytes per pixel before browser overhead.

Ordinary page requests are separate from image processing. Browser extensions, device security tools, cloud-synced folders, and any destination where you later share the download have their own data handling. Re-encoding also should not be treated as a guaranteed metadata-sanitization process.

Need a byte ceiling instead?

When delivery size matters more than exact geometry, use measured JPEG/WebP quality search and proportional fallback resizing.

Open the Image Compressor →

Geometry and format answers

Image resizer FAQ

Should I resize an image by pixels or percent?

Use pixels when a layout, platform, or asset specification requires exact dimensions. Use percent when you want a relative scale such as half-size while preserving the relationship to the source dimensions.

What is the difference between contain and cover?

Contain fits the whole image inside the requested box and can leave unused space on one axis. Cover fills the entire box and crops overflow from the other axis. Both preserve the image's aspect ratio.

What does no-upscale do?

No-upscale prevents the image from being enlarged beyond its source dimensions. It is useful when a requested box is larger than the original and you prefer a smaller output over invented extra pixels.

Will enlarging an image improve its quality?

No. Upscaling creates more pixels by interpolation but cannot restore detail that was not present in the source. The larger result may look softer or expose compression artifacts.

Can the resized image keep transparency?

PNG and WebP output can retain decoded transparency when the browser encoder supports it. JPEG has no alpha channel, so transparent pixels are flattened onto the selected background color.

Which images can the resizer process locally?

It accepts validated static JPEG, PNG, and WebP files up to 25 MiB and 40 megapixels. HEIC, AVIF, animated images, unsupported signatures, and undecodable files are rejected; processing occurs in the current browser tab.