Inverted Filter
An inverted filter flips the whole photo horizontally, swapping image-left and image-right while keeping the same shapes and proportions. Compare both orientations below and save the version you need as a PNG. Check the supplied photo first: a camera can save either orientation, so a flip alone cannot identify an unmirrored view.
One photo. Both orientations.
Choose a photo to compare its original pixels with a horizontal flip. No face detection, score or sign-up is needed.
JPG, PNG or WebP · Up to 20 MB · You can also drop a photo here.
Your photo stays in this page. Nothing is uploaded.
Compare the familiar view with a reversed view.
Original and horizontally flipped
If downloads are blocked, open the image and use your browser’s Save Image option.
A selfie may already be mirrored by its camera. These labels describe the supplied image and its flip; neither is the exact moving, three-dimensional view another person sees.
Clear photo erases the previews and releases prepared files. Downloads contain your chosen photo.
Photo privacyOne horizontal flip. The same photograph.
“Inverted face filter” commonly means a left-to-right reflection of the complete picture. Here, it does not mean a color negative, an upside-down rotation or a generated face. Text and the background reverse along with the portrait; the filter does not move individual facial features.
| Choose | What happens | What you learn |
|---|---|---|
| This inverted filter | The whole photo is reflected horizontally. | How the same image looks in its opposite orientation. |
| Symmetrical face generator | One chosen half is copied across a line. | How a constructed left-left or right-right portrait looks. |
| Face symmetry checker | Selected detected landmark pairs are compared. | Defined image measurements, with a stated method and limits. |
Was your selfie already mirrored?
A front-camera preview and its saved file can have different orientations. “Original photo” in this tool means the supplied image as your browser decodes it. The label does not certify how the camera saw the scene or what edits happened earlier.
Use ordinary printed text
Look for a readable sign, shirt logo or card facing the camera. Backward letters can identify reflection when you know how the actual object reads. A caption added after capture cannot establish the scene’s orientation.
Separate selfies from reflections
A photo taken into a mirror records a reflected scene, including the room and phone. A direct front-camera photo depends on its save setting. “Front camera” alone does not identify which orientation you have.
No reliable clue?
Call the results the supplied view and its flip. A hair part, facial feature, filename or preferred appearance cannot reliably establish whether an unfamiliar file was previously mirrored.
Check your camera’s save setting
On iPhone, Apple calls the control Mirror Front Camera under Settings → Camera. Pixel Camera calls it Save selfie as previewed. Both allow a saved selfie to match a mirrored preview. Follow the Apple selfie instructions or Google Pixel instructions for your device; menus can vary by software version and app.
A practical check: take a direct selfie holding a card with a short printed word, then inspect the saved file. Change the relevant setting and repeat. Compare the saved files with the physical card; do not rely only on the live preview.
Compare orientation before changing the photograph
- Use one file. Choose a JPG, PNG or WebP, or select Try the sample. Both previews use the same decoded image, keeping expression, lighting and camera position fixed. The sample is an AI-generated fictional adult.
- Track an object, then the whole image. In the sample, follow the hair part and background as they swap sides. With your photo, use a known object or printed word. Image-left means the left edge of the displayed picture, not the person’s anatomical left.
- Describe the operation. Ask what changed position and what retained its size. The flip moved existing differences across the image. It did not create a new eye height, change a jaw width or provide a facial symmetry score.
- Choose an output for a purpose. Save the flipped view to reverse an orientation, or save the comparison to show both. Select a save button, then Download PNG or Open image. Keep the camera’s original file separately.
Three checks to take away
Use these as a brief inspection checklist. Your checks are not submitted to or stored by this site.
For two actual portrait angles, use the Good Side Finder. To choose among the photo experiments, return to Mirror Face Tools.
Where does each pixel go?
In a working image that is W pixels wide, a pixel column numbered x from zero moves to W − 1 − x. Its row stays the same. For a six-pixel row, column 2 moves to column 3: 6 − 1 − 2 = 3.
The filter uses canvas translation and negative horizontal scaling. A reflection preserves lengths and angles in the working image; it adds no correction for camera distance, head turn or facial depth. Applying the same reflection twice to the same working pixels restores their positions. Reimporting a resized or re-encoded file is a separate processing step.
What size is the saved PNG?
| Output | Dimensions | Keep in mind |
|---|---|---|
| Original or flipped PNG | The working image size: at most 1600 pixels on the longest edge. | Both dimensions are scaled together and rounded. Smaller images are not enlarged. |
| Comparison PNG | 1600 × 1000 pixels. | A separate layout fits both views inside labeled panels. |
For example, a decoded 4032 × 3024 photo becomes 1600 × 1200 in the working image. Save original PNG means a PNG of that working view; it is not a byte-for-byte copy of the camera file or a promise to preserve its metadata. Select and retain your source file separately when you need it.
Why can a flipped face feel unfamiliar?
When one orientation is familiar, its reverse can feel surprising. Mita, Dermer and Knight’s 1977 study used mirror and unmirrored prints from the same photograph. Among photographed women who brought a friend, 21 of 33 preferred their mirror print; among those who brought a partner in the second study, 20 of 28 did.
The first proportion alone did not reach conventional significance. Friends’ and partners’ initial preferences for the unmirrored print also did not reach the authors’ significance thresholds; partners’ preferences over repeated trials did. These small, specific samples are consistent with a familiarity explanation for preference. They cannot tell you which view everyone finds attractive or explain every reaction to a filter.
Camera distance is a different question
A flip retains the photograph’s perspective. Ward and colleagues’ 2018 geometric model predicted a higher apparent nose-to-cheekbone width ratio at 12 inches than at a distant viewpoint, with little difference at five feet. It used average dimensions rather than testing this filter or every camera.
To explore perspective, take a separate eye-level photo farther away with similar lighting and expression. Cropping afterward can match framing; it cannot change the camera position used for the original capture. The selfie perspective guide explains the distinction. A reflection is a photo operation, not a test of health, attractiveness or how much other people notice a difference.
Inverted filter questions
Is the inverted filter accurate?
It performs a horizontal reflection of the browser-decoded working image. That is an image operation with a defined output, not a validated simulation of social perception. The previews cannot automatically identify whether your source file was already mirrored.
Does this filter make my face symmetrical?
No. It reflects both sides together and retains their differences. To construct a portrait from one side, use the symmetrical face generator. To inspect detected landmark relationships, use the face symmetry checker.
Is this the same as a TikTok inverted filter?
This page performs the horizontal reflection commonly meant by that phrase, using still photos. A social app may apply other processing or camera settings. Matching the flip operation does not guarantee identical images from two different apps.
Does this tool upload a photo or need an AI model?
The selected image is decoded, drawn and exported in your browser. This flip tool requires no face detection, face model or account. The website still makes ordinary page and asset requests; choosing the sample fetches that sample image. See the privacy policy for site analytics and privacy settings.
What happens when I clear the photo?
Clear photo empties the previews, cancels pending work and releases prepared download links. It cannot remove files you already saved, copies you shared or your camera’s source file. Downloaded PNGs contain the selected picture.
Why will my photo not open?
Use JPG, PNG or WebP, up to the tool’s 20 MB file limit and no more than 50 megapixels. Export HEIC as JPG first. A corrupt or unsupported image may not decode. Try another file or the sample; large images that open are resized before comparison.
Sources and how this page was checked
Source review: November 4, 2026. We checked the filter’s decoding, reflection, resizing, export and clear behavior in its implementation. The illustrations are original explanations, not research data or photographs of a participant. The familiarity study has a narrower scope than claims about how everyone sees you.
- Mita, Dermer & Knight (1977): Reversed Facial Images and the Mere-Exposure Hypothesis - original methods and preference tables.
- Ward et al. (2018): Nasal Distortion in Short-Distance Photographs - a geometric perspective model.
- MDN: Canvas scale() - reflection with negative scaling.
- Apple: Take a selfie with your iPhone camera and Google: Take a selfie on your Pixel - device-specific save settings.
Editorial policy · Measurement methodology · Return to the photo comparison