Sycago
All toolsReport an issue ↗

Image tools / FREE ONLINE TOOL

Image Resizer

Resize images to custom pixel dimensions with proportional fitting, centre cropping or padding.

No signupNo API keysLocal processing

YOUR WORKSPACE

Image Resizer

How to use ↓

Add your files

Drop files here, or choose from your device.

Still JPG, PNG or WebP · 20 MB per image · 100 MB total
Up to 20 files · Device memory may impose lower limits.

Ready. Your inputs stay in this tool's browser session.

Resizing an image means changing its pixel dimensions. You might need a smaller photograph for a page layout, a square product thumbnail, or a copy that fits a form’s stated width and height. These goals sound similar, but they require different choices. Keeping a complete photograph inside a box is different from filling that box edge to edge, and both are different from stretching the original.

The Image Resizer creates a separate copy using proportional fitting, centre cropping, or padding. Start with the destination’s requirements, then choose the treatment that preserves the information you care about. The tool can make a larger pixel grid when you allow upscaling, but it does not use AI to invent sharper detail. A careful choice of source, dimensions, and fit mode matters more than choosing the largest available output.

Understand dimensions, shape, and detail

Pixel dimensions describe how many samples an image contains across and down. A 2,400 × 1,600 photograph has a 3:2 aspect ratio. Reducing it to 1,200 × 800 preserves that ratio and uses one quarter as many pixels overall. The subject’s proportions remain natural because both dimensions change by the same factor. A requested box of 1,200 × 1,200 has a different shape, so keeping proportions requires a choice about the surrounding space or the image edges.

Fit inside keeps the full picture within the chosen bounds. The resulting picture can be smaller than one bound because its original proportions are retained. Centre crop fills the requested shape and discards material around the edges. Padding keeps the complete picture and adds background space to achieve the requested canvas. These choices solve layout problems; none can restore a face, word, or object that is already missing or indistinct in the source.

Dimensions and file size are separate. A 1,000-pixel photograph can still be a large file if its format and compression preserve a lot of information. Conversely, an image with many pixels can be strongly compressed and look poor. Use dimensions to solve the geometry requirement, then evaluate the encoded file size and appearance independently.

A step-by-step workflow

  1. Read the destination requirement

    Write down the requested width, height, accepted formats, and any byte-size limit. Separate a maximum boundary from an exact canvas requirement. “At most 1,200 pixels wide” allows a proportional result, while “1,200 × 1,200 pixels” may require cropping or padding. Do not assume that a platform’s displayed slot is its upload specification.

  2. Choose a useful original

    Load the clearest available source rather than a screenshot of a smaller copy. Check that the subject has room around it if you intend to crop. A source with unreadable text will still have unreadable text after resizing. For a batch, notice whether the images share one shape or include both portrait and landscape pictures.

  3. Enter the target pixels

    Set width and height according to the destination. Use ordinary pixel dimensions here; physical centimetres require a PPI conversion in the physical-size tool. Compare the requested output with the source dimensions before enabling upscaling. If the output is smaller, the browser resamples the original into a new, more compact pixel grid.

  4. Pick a fit method

    Use fit inside when every edge matters, centre crop when the frame must be completely filled, or padding when both complete content and an exact canvas are required. Inspect the preview after changing modes. A centred crop is not a subject-aware crop: the tool does not identify which face, label, or object you intended to protect.

  5. Select an output format

    Keep transparency with a suitable transparent format when needed, or choose JPG for a photographic copy that can have a solid background. Adjust quality where the selected encoder uses it. PNG ignores that quality control. If the destination imposes a strict maximum size, inspect the downloaded size rather than guessing it from a quality number.

  6. Check and save the result

    Examine the result at its actual intended display size and at a close zoom. Confirm its dimensions, edge treatment, readable details, and background. Download a separate copy and test that copy in the receiving application. For several sources, inspect each result because identical settings can produce very different crops on differently shaped pictures.

Choose settings for the destination

Choose the mode by deciding what is allowed to change: the frame, the visible edges, or the surrounding canvas. If more than one choice is acceptable, make a small test output and compare it in the layout where it will actually appear. The following examples use made-up destination dimensions, not requirements for any particular application.

Settings and checks for common situations
SituationPractical choiceWhat to inspect
Complete landscape photo in a flexible layoutFit a 3:2 image inside 1,200 × 1,200 pixels.Expect a proportional picture, such as 1,200 × 800, instead of assuming the height will fill the box.
Uniform square product cardsUse a 900 × 900 centre crop if the item has safe space around it.Check the top, bottom, and sides of tall packaging or wide products before accepting the square.
Full artwork on an exact square canvasUse padding to retain the whole image within a 1,000 × 1,000 canvas.Choose an appropriate background and check whether the added borders suit the destination.
Small source for a larger slotEnable upscaling only when a larger pixel count is actually needed.Inspect softness and visible compression artifacts; more pixels do not mean newly recovered detail.
Transparent icon or logoUse a transparency-capable output and inspect it over the intended background.Look for pale edge halos, small lettering, and unintended opaque background areas.

A good default for content-sensitive documents is to preserve the complete image. A good default for tightly controlled visual grids may be a consistent canvas. Neither is universal. A product thumbnail that loses part of the product is less useful than one with margins; a decorative cover image may tolerate an edge crop easily.

Worked examples

A landscape photo for a square card

Suppose the source is 3,000 × 2,000 pixels and the card needs an 800 × 800 image. Fit inside preserves all of the photograph and produces an 800 × roughly 533 picture, subject to whole-pixel rounding. Centre crop selects a square portion of the source before producing the 800-pixel square. Padding places the full proportional picture on a square canvas with space above and below.

If two people stand near opposite edges, the centre crop may remove part of one person. Padding is a safer choice when retaining both matters. If the picture is a textured wall with no important edge content, a crop may be visually cleaner. Make this judgment from the actual subject, rather than treating square output as an automatic improvement.

A page illustration that only needs a width limit

A 1,600 × 900 diagram is destined for an article that accepts images up to 1,000 pixels wide. A proportional result of 1,000 × about 563 keeps the original 16:9 shape. There is no reason to fill a 1,000 × 1,000 canvas unless the layout specifically requires it. Extra padding would increase the displayed space without making the diagram’s text larger.

Check the smallest labels after reduction. If the labels become difficult to read, use a less aggressive reduction or replace the diagram with a version designed for a narrower layout. Resizing faithfully reduces the image; it does not reflow the labels as editable text.

Preparing a mixed batch

Imagine six landscape photographs and four portrait photographs for a gallery. A shared fit-inside boundary of 1,200 × 1,200 can keep every full picture while limiting the longest side. A shared centre-crop setting would instead give ten square frames and remove different parts of each source. These are two legitimate gallery designs with different tradeoffs.

Process a representative portrait and landscape first. Check the results in the gallery’s actual CSS or page template before preparing the rest. If the destination also crops images at display time, a carefully padded image can acquire another border or an unexpected crop, so the final layout remains the decisive check.

Build a set of sizes from one master

If a project needs a small card, a medium article image, and a large feature image, make each from the source master. Decide the composition for each role before generating the sizes. The card may need a square frame, while the article image can retain its original proportions. These are separate layout decisions, so a single crop should not be stretched into every role merely to keep the filenames consistent.

Compare the variants together in the actual page. Check that the subject has a similar visual emphasis and that any text remains readable in the smallest version. If the small card cannot carry the same amount of detail, a simpler source or a closer composition may be more useful. Save a brief note about each role’s dimensions and fit method so the next image in the series can follow the same intentional treatment.

Inspect edges, text, and real display size

Start with the geometry. Confirm that the output has the dimensions you intended and that the chosen fit mode explains any visible border or missing edge. Examine faces, hands, product corners, labels, and any information close to the frame. Centre cropping deserves particular attention when the subject is off centre. For a precise off-centre selection, crop the desired region first and then resize that copy.

Next, inspect detail at a sensible scale. Viewing a small image at several hundred percent makes normal resampling look rough, while a tiny preview can conceal unreadable text. Judge the image at its intended use size, then zoom in to diagnose anything suspicious. Fine grids, narrow stripes, and small lettering can change appearance when reduced because the original detail no longer fits within the available pixels.

Finally, open the downloaded file and check its properties. A browser preview is useful but the file is what another application receives. Confirm the output format and final byte size separately from the dimensions. If a receiving form rejects it, compare its stated requirement with the actual file properties instead of repeatedly enlarging or shrinking at random.

Files, limits, and browser support

The image tools work within explicit input and output limits. Each input file may be up to 20 MB, with no more than 20 files and 100 MB combined in one selection. A decoded image may contain no more than 24 million pixels, and neither input side may exceed 12,000 pixels. These checks concern different properties: a highly compressed photograph can have a modest file size while expanding into a very large pixel grid. A file below the megabyte limit can therefore still exceed the decoded-image limit.

An exported image is limited to 8,000 pixels on either side and 24 million pixels in total. Multiplying width by height gives the pixel count. For example, 6,000 × 4,000 is exactly 24 million pixels, while 8,000 × 4,000 is 32 million and exceeds the area limit even though each side fits the side limit. Choose dimensions that satisfy both limits. These are processing guardrails, not a statement that every allowed file will be equally fast on every device.

Browser decoding and encoding support can vary. A filename ending in an expected extension does not prove that the underlying file has that format or that the browser can decode it. Use a real still image in the format the particular tool accepts. If loading fails, confirm the source opens correctly in an image viewer and obtain a fresh export from the application that created it. Renaming a file extension does not convert its contents.

Processing happens locally in the browser. The tool does not use an image-processing API or upload your selected image files for the operation. The surrounding website may have analytics or other site features governed separately; local processing describes this tool’s treatment of the image, rather than every possible network request made by the page. Keep the tab available until processing and downloads finish. A slow phone or a busy browser may need fewer files per batch.

Keep results useful after downloading

For a set of layout assets, name results consistently, for example product-a-square-900 or landscape-wide-1200. Keep a record of whether the set uses cropping or padding. This makes a later replacement easier because another editor can reproduce the visual treatment without guessing from one finished image.

If several sizes are required, return to the original for each size rather than repeatedly resizing the smallest result. Each reduction discards image information, and enlarging that reduced copy later cannot recover it. Maintain an original master and generate the needed delivery copies from that master. This also lets you revise crop decisions when the surrounding layout changes.

Keep the source image and treat the downloaded result as a separate working copy. The workflow does not destroy your original file. A useful filename describes the change and purpose, such as a size, format, or version label, so a later handoff does not depend on remembering which download came first. Save important copies somewhere you can find again; a preview remaining on screen is not the same as a saved file.

Where multiple image results are offered, processing runs sequentially and you can inspect and download the results individually or collect them in a ZIP created locally. Check the count before closing the page, especially if one source failed while others succeeded. A ZIP groups the result files for convenience; it does not improve image quality or change the encoding inside each file. Open a representative output after extracting it, then review any unusually small or visually different result separately.

For JPG, PNG, and WebP exports, choose the format for the recipient’s actual needs. JPG has no transparent pixels. PNG can retain transparency but its encoder ignores a photographic quality slider; changing that control does not provide a PNG file-size target. WebP availability depends on the browser and destination. A quality value for a lossy format is an encoder setting, not a promise of a particular number of kilobytes. The image dimensions, texture, noise, and chosen format all influence the final size.

Frequently asked questions

Does resizing change my original image?

No. The tool creates a processed copy for downloading and leaves the selected original intact. Keep the original as your master, especially if you may need another size later. Downloading a new image does not replace the source unless you deliberately overwrite a file when saving it.

What is the difference between fit inside and crop?

Fit inside keeps the complete image within the requested boundaries, so one output dimension may be smaller than its boundary. Centre crop fills the requested shape and removes some edge content when the aspect ratios differ. Choose based on whether the complete subject or the exact filled frame matters more.

When should I use padding?

Use padding when you need an exact canvas size and must retain the whole source. The image is fitted proportionally and the unused canvas becomes background space. Check the background colour and how the result looks in its destination, where another automatic crop or background may change its appearance.

Can I enlarge a small image?

Yes, when you allow upscaling and the requested output remains within the output limits. Enlargement adds pixel samples, but it does not recover genuine detail that the source never contained. Expect softness in tiny text, faces, or compressed edges, and choose a better source when detail is important.

Will this use AI to sharpen my picture?

No. This resizer performs ordinary browser image processing, not AI reconstruction or generative upscaling. The result follows the content of the source. If a source is blurry, heavily compressed, or too small, resizing can adjust its dimensions but cannot reliably recreate the missing information.

Why is my result not the exact height I entered?

Fit-inside mode preserves proportions within both bounds. A wide source fitted into a square boundary naturally leaves unused vertical space rather than stretching. Use padding for a full exact canvas or centre crop for a filled exact frame, then inspect whether that treatment suits the picture.

Does a smaller width guarantee a smaller file?

It often reduces the amount of image data, but it does not guarantee a specific byte size. The output format, compression setting, transparency, and picture content matter too. Compare the actual downloaded file size, particularly when a destination enforces a maximum measured in KB or MB.

Why does PNG quality not change the file much?

PNG uses a different encoding approach from a photographic JPG quality control. The PNG encoder ignores the quality slider, so moving it is not a reliable way to reduce a PNG. If transparency is unnecessary, compare a suitable JPG or supported WebP copy and inspect the visual tradeoff.

Can I resize a transparent image?

A transparent source can remain transparent when you export through a supported transparency-capable format. JPG cannot store transparency and needs a solid background instead. Always view the resulting edges against the actual background where the image will be displayed, especially for logos and icons.

Can I choose a crop away from the centre?

The resizer’s centre-crop mode uses the centre of the image. If your important content is off centre, use the Image Cropper to select the region first, then resize that result. Review the cropped copy before resizing so you do not enlarge or polish an accidental composition.

Is changing pixels the same as setting centimetres?

No. Pixels describe the image grid. A physical size additionally needs a pixels-per-inch assumption, and the receiving print software controls placement. Use the CM & MM Image Resizer for that calculation, then verify the actual print settings rather than assuming pixels alone define paper dimensions.

Can I process several images together?

The input limits allow up to 20 files with a combined maximum of 100 MB, while each file must be no more than 20 MB. Images are processed sequentially. Review every result because a shared crop or canvas setting can affect portrait and landscape sources differently.

Why can a small file exceed the image limit?

Compressed bytes and decoded pixel dimensions measure different things. A highly compressed picture can contain more than 24 million pixels despite a small file size. Check width multiplied by height as well as megabytes. The input side limit is 12,000 pixels, and both checks must pass.

Will resizing remove an unwanted background?

No. Resizing changes dimensions and may crop or add surrounding space, but it does not identify and remove a subject’s background. Padding adds a canvas around the image; it does not make the existing background transparent. Use a source that already has the background treatment you need.

Why does text look softer after reducing an image?

Small lettering is made from a limited number of pixels, and reduction leaves fewer samples to represent each stroke. A slight blur or uneven line can therefore appear. Keep a larger output, simplify the original artwork, or use editable text in the destination when readability is essential.

What should I check before uploading the result?

Check dimensions, format, final file size, complete subject coverage, readable text, and the intended background. Open the downloaded copy rather than relying only on its preview. Then test the destination, because an application may apply its own crop, compression, or display rules after receiving a valid image.