Flip Image
Flip JPG, PNG, WebP, GIF images horizontally or vertically in one click. Lossless mirror tool runs 100% in browser — no upload, instant download.
About Image Flipping
Image flipping is a geometric transformation that mirrors a picture across its horizontal axis (left-right swap) or vertical axis (top-bottom swap). It is the most common pre-processing step in photography, graphic design, and computer vision because every consumer camera by default produces a mirror image of what the lens captures through its front-facing sensor. This tool performs the flip directly in your browser using the Canvas 2D API, preserving every pixel and color value — the operation is mathematically lossless. The resulting image keeps its original resolution, bit depth, and aspect ratio; only the pixel order changes.
Why do selfie photos look mirrored on phone screens but normal when saved?
Every iPhone, Pixel, and Galaxy front camera shows a live preview that is horizontally flipped because humans are accustomed to seeing themselves in a mirror — an un-mirrored preview feels wrong because parting hair on the left appears on the right. Apple introduced an option in iOS 14 (Settings, Camera, Mirror Front Camera) that keeps the saved photo flipped to match the preview; older defaults flip back when saving so text on shirts reads correctly. If your saved selfie shows you backwards compared to what you saw on screen, apply a horizontal flip in this tool and it will look right. The pixel data is identical either way; flipping is purely a viewing convention.
Is flipping really lossless, or does it degrade the image like rotation?
Horizontal and vertical flips are mathematically lossless because every output pixel is an exact copy of a source pixel at a mirrored coordinate — no interpolation is required. Rotation by an arbitrary angle (say 7 degrees) needs interpolation because output pixels rarely land exactly on source pixels, which softens edges. But 180-degree rotation and the two flips are special cases: they only reorder pixels, never blend them. The only quality loss comes from re-encoding the result if you save as JPEG, because JPEG compression is itself lossy. To stay 100% pristine, export as PNG or WebP-lossless after flipping. Some tools also perform lossless JPEG flipping via the jpegtran utility, which manipulates DCT blocks directly without re-encoding.
How is flipping different from rotating an image 180 degrees?
A 180-degree rotation is mathematically equivalent to applying a horizontal flip AND a vertical flip — both operations together. A single horizontal flip produces left-right mirror, a single vertical flip produces upside-down with reversed left-right, and 180 rotation produces upside-down with normal left-right. Test it on text: rotating 'HELLO' 180 degrees gives 'OllƎH' upside-down; horizontally flipping 'HELLO' gives backwards 'OLLEH' right-side-up; vertically flipping gives 'HELLO' upside-down with the H still on the left. Use rotation when you want to read upside-down content the right way up, and flips when you specifically need mirrored output.

Should I flip before or after cropping, resizing, or applying filters?
Order matters in image pipelines but flipping is commutative with most operations. Flipping commutes with brightness, contrast, saturation, hue, and most color filters because each output pixel only depends on the corresponding input pixel's color. It also commutes with resizing as long as the resize uses symmetric kernels (Lanczos, bicubic, bilinear). It does NOT commute with directional operations like motion blur, asymmetric crops, perspective transforms, or watermarks: a logo placed in the bottom-right before flipping will end up in the bottom-left. Best practice: apply directional edits last so the flipped version is the final pixel state.
How do flipped images affect EXIF orientation metadata?
The EXIF Orientation tag (number 274 in the EXIF spec) encodes eight transformation states: 1=normal, 2=horizontal flip, 3=180 rotate, 4=vertical flip, 5-8 combine rotation with flip. Cameras and phones write this tag rather than altering the pixel grid, so a portrait photo from your phone is often a landscape pixel array with Orientation=6 (rotate 90 CW). When you flip in software, well-behaved tools bake the transform into the pixel data and reset Orientation to 1; poorly-written tools edit the tag without touching pixels, leaving an inconsistency that breaks browsers and image libraries differently. This tool re-renders pixels and strips the orientation tag, guaranteeing the saved file displays identically everywhere.
What is the use of flipping in machine learning training data?
Horizontal flipping is the most common data-augmentation technique in computer vision. By including each training image and its mirror, you effectively double the dataset and teach the model that a cat facing left is the same class as a cat facing right — a property called translational invariance. Libraries like TensorFlow and PyTorch ship random-flip layers (RandomFlip in Keras, RandomHorizontalFlip in PyTorch) that flip 50% of images during training. Vertical flipping is used less because most natural scenes have a clear top-bottom asymmetry (sky above, ground below); flipping them vertically produces unrealistic inputs. The exception is satellite, microscopy, and medical imaging where there is no preferred orientation, so both flips are applied.
Can flipping fix scanned documents or photos that came out backwards?
Yes — this is one of the most common reasons people flip images. Photos scanned with the wrong side facing the glass, photographed in a mirror, or captured through transparent material (a coffee shop window, a film negative scanned emulsion-side-up) all come out reversed. A single horizontal flip restores the correct reading order: text becomes legible, faces look like the original person, and asymmetric details (a parting in hair, a watch on the wrist) end up on the correct side. For scanned 35mm film negatives the rule is: if text reads backwards in the scan, flip horizontally; if your subject's wedding ring jumped from left hand to right, flip horizontally. For double-sided document scans where alternate pages are upside-down, batch-process the odd or even pages with a 180-degree rotation instead.
