Crop Image

Crop an image by keeping one exact pixel rectangle

Cropping removes everything outside a selected rectangular region. SnakTool asks for the rectangle's X and Y starting coordinates plus its width and height, then creates an output whose pixel dimensions are exactly that crop width and height.

The retained region is copied at the same dimensions: a 600 × 400 crop becomes a 600 × 400 output. The crop operation does not squeeze the whole source into that rectangle and does not scale the selected region to another size.

Use cropping when the composition needs to change: removing empty edges, isolating a subject, creating a square from a landscape photo or excluding an unwanted area. Use Image Resizer instead when the whole composition should remain but its pixel dimensions need to change.

X and Y start at the decoded image's top-left corner

The coordinate system begins at the top-left of the decoded image. X increases as you move right and Y increases as you move down. A crop at X 0 and Y 0 therefore begins in the source image's top-left corner.

Width and height describe how far the rectangle extends from that starting point. On a 1200 × 800 image, X 200, Y 0, width 800 and height 800 selects the centered 800 × 800 square.

SnakTool decodes the image with its orientation applied before cropping, so work from the image as it is decoded for display rather than trying to calculate against an unrotated camera sensor layout.

Input

1200 × 800 landscape image

Method

(1200 - 800) / 2 = 200 pixels from the left

Result

x 200 · y 0 · width 800 · height 800

The entire crop rectangle must remain inside the source

A valid crop needs a non-negative starting position, a width and height of at least one pixel, and a right and bottom edge that do not extend beyond the decoded source. SnakTool rejects the request when X plus width exceeds source width or Y plus height exceeds source height.

For a 1000 × 700 source, X 800 with width 300 is invalid even though both individual numbers look reasonable, because the right edge would land at pixel position 1100. The same rule applies vertically.

When a crop is rejected as outside the image, check the sums rather than only the crop dimensions. Moving the starting coordinate closer to the origin or reducing the rectangle size will bring the selection back inside the available pixels.

Crop dimensions determine the output aspect ratio

Aspect ratio is the relationship between width and height. A 600 × 600 crop is 1:1, an 800 × 600 crop is 4:3, and a 1600 × 900 crop is 16:9. SnakTool does not need a separate aspect-ratio setting because the width and height you enter define that ratio directly.

The current interface does not provide named presets such as square, 4:3 or 16:9 and does not provide a draggable crop overlay. If a destination specifies a ratio, calculate or choose rectangle dimensions that match it while keeping the rectangle inside the source.

A ratio alone does not specify output size. Both 400 × 400 and 1200 × 1200 are square crops, but they retain very different numbers of source pixels. Choose a rectangle large enough for the destination before doing any later resize.

Centering a crop is a coordinate calculation, not an automatic mode

To center a crop horizontally, subtract crop width from source width and divide the remainder by two. To center it vertically, subtract crop height from source height and divide by two. The resulting X and Y place equal unused space on opposite sides when the differences divide evenly.

For a 1200 × 800 image and an 800 × 800 square, horizontal X is (1200 - 800) / 2 = 200 and vertical Y is 0. For a 1600 × 1200 image and a 1000 × 800 crop, X is 300 and Y is 200.

SnakTool does not automatically detect a face, subject or focal point. Centering mathematically can still cut off an off-center subject, so composition should determine the coordinates rather than assuming the geometric center is always the visual center.

Cropping and resizing solve different geometry problems

Cropping changes which pixels remain. Resizing changes how many pixels represent the whole retained image. A landscape photo can be cropped to square without stretching because side pixels are discarded; resizing that same landscape directly to equal width and height with aspect-ratio locking disabled would distort the subjects.

If a profile image must be 512 × 512, one practical workflow is to choose a square crop around the subject first and then resize that square to 512 × 512 if its crop dimensions are larger. This separates composition from final delivery dimensions.

Do not crop merely because a file is too large in kilobytes. Removing a large part of an image can reduce pixel count, but Image Compressor is the operation designed around encoded byte reduction when composition and dimensions are already correct.

Cropping does not resample the retained rectangle

The engine uses the selected source rectangle and draws it into an output canvas with the same width and height. Because source and destination crop dimensions match, the crop itself is not an enlargement or downscale of the selected pixels.

This distinction is useful for small text, pixel art and fine edges: cropping does not intentionally blur them through geometric scaling. The output is still newly encoded, however, so 'not resampled' should not be confused with 'the compressed file bytes are unchanged.'

If you crop first and resize later, the resize step is where geometric resampling occurs. Keeping those operations conceptually separate makes it easier to identify whether a visible change came from selecting a different region, scaling pixels or re-encoding the format.

The source format stays the same, but the cropped file is re-encoded

A JPEG crop is exported as JPEG, a PNG crop as PNG and a WebP crop as WebP. SnakTool does not silently change formats during this operation, so format conversion remains a separate choice.

The output is created through the browser canvas rather than by copying a rectangular byte range from the compressed source file. JPEG and WebP therefore go through the engine's default encoding quality path for this operation, while PNG follows its browser PNG encoding path.

For JPEG in particular, this means the crop is not a guaranteed lossless JPEG transform even though the retained region is not geometrically resized. Keep the original if avoiding any additional lossy re-encoding is important.

Transparency is preserved where the source format supports it

PNG and WebP use a transparency-capable output canvas, so transparent and semi-transparent pixels inside the selected rectangle can remain transparent in supported browser encoding. Cropping simply removes pixels outside the rectangle; it does not require filling the retained region with a background.

JPEG cannot store transparency. A JPEG crop remains JPEG and is drawn over the engine's white JPEG canvas, which is consistent with an opaque output format.

If the real goal is to change a transparent PNG or WebP into JPEG, use the corresponding format converter where background color is an explicit option rather than relying on crop behavior to perform an unrelated format change.

Cropping can change file size without being a file-size targeter

A smaller crop often contains fewer pixels than the full source, so its encoded file can be smaller. The exact byte size still depends on format, image complexity and re-encoding, so a 50% reduction in pixel area does not promise a 50% reduction in kilobytes.

SnakTool's crop interface does not offer a target such as 500 KB or a quality slider. If the composition is correct after cropping but the result must meet a strict upload-size limit, evaluate the output bytes and use Image Compressor when additional reduction is needed.

Likewise, cropping should not be used as a substitute for resizing when the destination specifies exact pixel dimensions. A crop changes the visible field; resizing changes the dimensions without intentionally discarding the outer composition.

Cropping is destructive to the exported copy, so keep the original

Pixels outside the selected rectangle are not present in the cropped output. Expanding the canvas later cannot restore those discarded surroundings from the cropped file itself; it can only add new empty or generated content.

Keep the original image when you may need another composition, a wider framing or a different aspect ratio later. This is especially important when preparing several versions of the same photograph for avatars, banners, cards and full-size viewing.

Also remember that removing visible surroundings is not the same as a dedicated metadata-privacy workflow. If location or camera metadata is the concern, use Remove Image Metadata and verify that output separately.

Browser-local cropping is bounded by decoded image resources

SnakTool validates and decodes one JPEG, PNG or WebP locally in a browser worker, crops it with an offscreen canvas and encodes the result without sending the image to a SnakTool image-processing backend.

The normal policy allows images up to 8192 pixels per side and 32 million pixels, with 192 MiB of estimated memory, a 25 MiB output ceiling and 60-second processing limit. Known low-memory devices use tighter limits, including 4096 pixels per side and 12 million pixels.

The crop output is also checked against those resource limits. File size on disk is not the whole workload because the browser must decode the source pixels in memory before it can select and encode the requested rectangle.

Verify the rectangle before treating the crop as final

Check the output dimensions first: they should equal the crop width and height you entered. Then inspect all four edges for accidentally clipped text, limbs, logos or other content that sat close to the selection boundary.

For ratio-sensitive uses such as avatars or thumbnails, confirm the actual width-to-height relationship instead of judging by appearance alone. For JPEG, also inspect fine detail because the file has been re-encoded even though the retained region was not geometrically scaled.

A good crop solves the composition requirement while leaving enough source resolution for its final use. When another size is required afterward, keep the cropped composition proportional and perform the resize as a deliberate second step.