Putar & Balik Imej

Flip and rotate any image, right in your browser

This flip image tool mirrors and rotates pictures in a single click — no app to install, no account, and nothing to upload. Drop in a PNG, JPG, WebP, or GIF, then flip it horizontally or vertically, rotate it 90 degrees left or right, turn it a full 180°, or dial in a custom straighten angle with the slider. A live preview shows the result as you go, and you download a clean PNG or JPG when you're happy.

Because every transform runs locally on an HTML canvas, your image never leaves your device. That makes it a private, instant alternative to upload-based tools like iLoveIMG and Rotateimages.com — both of which send your file to a remote server before returning the result. On a slow connection, those tools can take 30+ seconds per image; this one is instant.

Mirror an image (horizontal flip) or flip it vertically

To mirror an image, use the horizontal flip: it reflects the picture across its vertical center line so the left and right sides swap. It's the classic fix for a selfie that came out reversed from a front-facing camera, and the easy way to make two arrows or two faces point toward each other. A vertical flip reflects top-to-bottom instead — like a reflection on water or a floor — useful for creating symmetrical compositions or creative double-exposure effects.

Flips are completely lossless: the pixels are rearranged in memory and no recompression happens. You can flip, un-flip, and combine a flip with a rotation as many times as you like without ever degrading the image. Photoshop and GIMP can do this too, but they require the software to be installed. This tool works in any browser tab, on any device, with no account and no install.

Rotate 90°, 180°, or by any custom angle

The quick buttons rotate the image 90 degrees clockwise or counter-clockwise, and 180° turns it upside down. A 90° turn swaps width and height — a 4000x3000 landscape photo becomes 3000x4000 portrait — with no quality loss and no transparent corners, because an exact quarter-turn just rearranges pixel rows and columns without any resampling.

For finer control, the Straighten slider rotates by any angle from -45° to +45° on top of those quarter-turns. This lets you fix a crooked horizon by a few degrees without doing a full 90° turn. The canvas auto-expands to fit the tilted image, and you decide whether the new triangular corners are transparent (best as a PNG) or filled with a background color (best for JPG output).

Auto-fix sideways phone photos (EXIF orientation)

Photos from smartphones often appear sideways because the camera stores them in landscape pixel layout with a hidden EXIF 'orientation' tag telling apps to rotate the display. Many programs ignore this tag, showing the raw layout — a portrait photo appears sideways. This tool reads the EXIF tag and auto-corrects the orientation on load, so a portrait shot displays upright without any clicks from you.

You can switch the auto-fix off to work with raw pixel data, which matters if you're specifically trying to edit EXIF orientation metadata. For most users though, the auto-fix means you open a sideways photo and it's already upright — then you're just deciding whether to flip or rotate further. This is faster than opening the same photo in Photoshop or Windows Photos and manually rotating.

Competitor comparison — flip and rotate tools

iLoveIMG offers a flip and rotate tool but uploads your image to their servers for processing. Their free plan limits file size and has a daily cap on operations. Rotateimages.com is similar — server-side, with an interface that hasn't changed in years and no batch support. Canva can flip images inside a design but requires an account and doesn't let you work with a standalone image without creating a project first. Photoshop is the most powerful option, but it costs around $22/month and only runs on desktop.

UtiloKit Flip & Rotate does all of this client-side: no upload, no account, no file size cap, no watermark. It also adds features most flip/rotate tools skip — custom angle straightening, EXIF auto-correction, batch processing, and ZIP download for multiple images. If you've been using iLoveIMG for quick flips and want something that doesn't upload your photos, this is the direct replacement.

Batch mode, output formats, and device support

Load several images at once and your chosen flip, rotation, angle, and background color apply to all of them simultaneously. Grab them all as a ZIP archive or download the currently previewed image on its own. For output, choose PNG to preserve transparency and maximum quality, or JPG with adjustable quality and background color for a smaller file. The tool always shows exact output dimensions and estimated file size before you download.

The tool is fully mobile-responsive. On iPhone, open it in Safari and tap to select from your Photos library. On Android, use Chrome and pick from your gallery. No app installation, no account, and the image is processed locally on your phone — nothing sent to any server. Whether you searched for how to flip an image in Canva, PowerPoint, Photoshop, on iPhone or on Android, this page handles all of it without requiring a download or sign-up.

The mathematics of image rotation

Every rotation is governed by the 2D rotation matrix. To rotate a point (x, y) by angle θ around the origin, the new coordinates are: x′ = x·cos θ − y·sin θ and y′ = x·sin θ + y·cos θ. This formula maps every source pixel to a new destination position. 90° rotations are a special case because cos 90° = 0 and sin 90° = 1, which collapses the matrix to x′ = −y, y′ = x — a pure integer rearrangement with no floating-point arithmetic required. That is why quarter-turn rotations are lossless: every source pixel maps exactly onto a destination pixel with no ambiguity.

For arbitrary angles — say 7° to fix a tilted horizon — the transformed coordinates are rarely whole numbers. A source pixel at (100, 200) might map to destination (98.3, 213.7), which falls between four real pixels. The algorithm must then interpolate a blended color from those neighbors, the simplest method being bilinear interpolation. This blending is why non-90° rotations produce a very slight softness; it is the same effect Photoshop exhibits, and it is mathematically unavoidable. Saving the result as PNG avoids adding JPEG compression on top of the resampling blur, keeping the output as sharp as possible.

The pivot point also matters. Rotating around the image's center keeps the bounding box symmetrical and is what virtually all user-facing tools do — the image stays centered and the canvas expands uniformly in all directions to contain the rotated corners. Rotating around the top-left corner (the mathematical origin) shifts the image visually and produces an off-center result. 180° rotation is equivalent to flipping both axes simultaneously: the rotation matrix at θ = 180° gives x′ = −x, y′ = −y, which is identically the composition of a horizontal flip (x → −x) followed by a vertical flip (y → −y). That identity is why a double-flip equals a 180° turn.

EXIF orientation and the auto-rotation problem

Smartphone camera sensors are physically mounted in landscape orientation. When you hold the phone in portrait mode and shoot, the sensor still captures a landscape-format image — but the camera firmware writes an EXIF Orientation tag into the file's metadata instructing display software to rotate the pixels before showing the image. The tag uses integer values: 1 = normal (no rotation needed), 3 = 180° rotated, 6 = 90° clockwise, 8 = 90° counter-clockwise, plus four mirrored variants (2, 4, 5, 7) for front cameras that flip horizontally. A portrait photo taken on an iPhone typically carries tag value 6, meaning 'rotate 90° clockwise to display correctly.'

The chaos arises because different software treats the EXIF tag differently. Browsers like Chrome and Safari read it and rotate the displayed image automatically — so photos look correct. Older tools and some image-processing libraries ignore the tag and display the raw pixel layout — so a portrait photo appears sideways. Others strip all EXIF metadata during processing, which discards the rotation instruction entirely and leaves the pixels in the wrong orientation with no tag to guide display. This discrepancy is why the same image can look correct in one application and sideways in another. The robust solution is to bake the rotation into the pixel data — physically rearrange the pixels so the image is stored correctly, then set the EXIF Orientation tag to 1 (normal). After that, every application displays it correctly regardless of whether it reads EXIF.

This is exactly what this tool's auto-fix orientation feature does on load: it reads the EXIF tag, applies the corresponding rotation or mirror to the canvas pixels, and resets the tag to 1 before any further edits. The result is a file that is correct at the pixel level, not merely annotated with instructions that some applications will ignore. Professional export pipelines — including Adobe Lightroom's 'Edit in Photoshop' round-trip and most CMS image processors — do the same thing for the same reason: downstream systems cannot be trusted to honor EXIF orientation.

Why flipping images matters: psychology, selfies, and machine learning

A horizontal flip reflects every pixel across the vertical center axis: for a pixel at column x in an image of width W, the output column is W − 1 − x. A vertical flip applies the same logic to rows: y′ = H − 1 − y. These are exact integer mappings — no rounding, no interpolation, no quality loss — which is why flipping is always lossless regardless of the image format. The operations are conceptually the same as holding a transparency up to a light and turning it around, or reading text in a mirror.

The most common reason people flip a photo is the selfie mirror effect. Front-facing cameras capture what the lens sees, but people are accustomed to seeing themselves in a mirror — a reflected view. When a photo is displayed without flipping, the viewer's face appears to have its asymmetries reversed: a slight leftward smile now curves right, hair is parted on the unexpected side, and the image feels subtly 'wrong.' This phenomenon is documented in psychology research including the Mita, Derner, and Knight study (1977), which found that subjects rated their own mirror image as more attractive than the true image, while close friends and partners preferred the true image — because each group was more familiar with the version they usually saw. Horizontal-flipping a selfie produces the true-camera version; flipping it back produces the mirror version the subject finds more flattering.

In computer vision and machine learning, random horizontal flipping is one of the most widely used data augmentation techniques for training image classifiers. Because the semantic content of most natural images is identical whether read left-to-right or right-to-left — a dog facing left and a dog facing right are both dogs — flipping each training image doubles the effective dataset size at zero labeling cost and teaches the model not to rely on left-right orientation as a feature. Vertical flipping is used selectively, mainly for satellite and aerial imagery where a building photographed from above is equally valid upside down. For face recognition and text detection, vertical flipping is generally avoided because those tasks are orientation-sensitive.

How browsers flip and rotate images: Canvas, OffscreenCanvas, and the CORS restriction

CSS offers a zero-cost way to visually flip an image: transform: scaleX(-1) mirrors it horizontally without touching a single pixel. This is a GPU compositing operation that costs virtually nothing and reverses instantly. But it is purely presentational — the underlying pixel data is unchanged, and if you try to download the image, you get the original unflipped version. To actually change pixel data in a browser, you must use the HTML Canvas API: draw the image onto a canvas (optionally with a transform applied), then call canvas.toBlob() or canvas.toDataURL() to read the new pixel data out as a file. That canvas-based approach is what this tool uses for every flip, rotation, and download.

The ImageData API gives direct access to the underlying pixel buffer: a Uint8ClampedArray of RGBA values stored in row-major order — four bytes per pixel, left-to-right, top-to-bottom. Reading and writing this array allows arbitrary per-pixel transformations such as custom rotation angles with bilinear interpolation. For large images, these pixel-loop operations can be slow on the main thread and may block the UI. The OffscreenCanvas API — available in modern browsers — moves canvas rendering into a Web Worker, enabling the same pixel operations to run off the main thread so the page remains responsive while processing a high-resolution image.

One important restriction applies to any canvas-based tool that loads images: the CORS tainted-canvas rule. If a browser canvas draws an image fetched from a different origin without a permissive Access-Control-Allow-Origin header, the canvas becomes 'tainted' and getImageData() throws a security error — preventing any pixel readback. This is a deliberate browser security measure to stop scripts from reading pixel data from cross-origin images and leaking that data to a third party. This tool sidesteps the issue entirely because the image is loaded directly from the user's local device via a file input — a local file has no cross-origin relationship with the page, so the canvas is never tainted and pixel readback always works.