How to use an image perspective corrector
Load a JPG, PNG, WEBP, or BMP, choose Deskew (simple rotation) or Perspective (four corners), place the guides, and click Apply. The Corrected image preview uses your output-size and fill settings, and Export corrected image saves a PNG. The page reads the selected file with a local object URL and processes it with browser image and canvas APIs.
- Drop an image here or click browse files, then choose a JPG, PNG, WEBP, or BMP file.
- Choose Deskew (simple rotation) for one rotation, or choose Perspective (four corners) to place four guides separately.
- Drag the dashed-line handles or the four corner handles. Turn Grid overlay on when its lines help you place a guide.
- Choose an option under Output canvas size and Fill exposed areas.
- Click Apply, check the Corrected image preview, then click Export corrected image.
How to deskew an image
Deskew (simple rotation) applies one rotation to the image. Move Straighten angle or enter a value in Angle (deg); the code clamps the value between -45 and 45 degrees. To estimate the rotation, place the two dashed-line handles along an edge and click Use dashed line as angle.
How to correct trapezoidal perspective
Perspective (four corners) uses four independent handles. Place them on the four corners you want to map, then click Apply. Auto-detect edges (assist) estimates a starting quadrilateral from edge contrast and leaves the guides for you to check. Reset to image corners restores the initial guide positions near the image edges.
Choose output size and fill exposed areas
Output canvas size controls the corrected image dimensions. Auto-fit to guides uses the average guide width and height in Perspective (four corners), while Deskew (simple rotation) uses the source dimensions. Preserve original image size keeps those source dimensions. Custom size shows Width and Height fields for pixel dimensions; an empty or invalid field falls back to the matching source dimension.
Rotation and warping can leave pixels outside the source image. Fill exposed areas gives you four choices:
- Transparent leaves those pixels transparent.
- White fills them with white.
- Edge-extend samples the nearest source edge.
- Crop to valid area trims the canvas to sampled source pixels.
Where image perspective correction comes up
Use Deskew (simple rotation) when the image needs one overall turn. The dashed reference line can give you an angle for a visible straight edge.
Use Perspective (four corners) when the area to correct has four corners that need separate guide placement. The tool maps those guide points into the output canvas, then applies the chosen fill policy to pixels outside the source image.
When another image edit tool fits better
This page sits in Image Edit Tools. Choose Image Rotator when you need rotation, flipping, or auto-orient behavior without correction guides.
Choose Image Cropper to trim an image with a freeform, aspect-ratio, or social preset selection. Choose Aspect Ratio Checker & Cropper to inspect an image ratio before cropping.
Frequently asked questions
How do I use an image perspective corrector?
Load a JPG, PNG, WEBP, or BMP, choose Deskew (simple rotation) or Perspective (four corners), position the guides, and click Apply. Export corrected image saves the result as a PNG.
What is the difference between deskewing an image and correcting perspective?
Deskewing rotates the whole image by one angle. Perspective correction moves four selected corners independently, which corrects a rectangular area whose edges do not appear parallel.
Can Auto-detect edges (assist) correct a photo by itself?
Auto-detect edges (assist) estimates a starting position from edge contrast. Check and adjust the four guides before clicking Apply.
What happens to empty areas after image warping?
Fill exposed areas applies the selected Transparent, White, Edge-extend, or Crop to valid area option. Crop to valid area reduces the canvas to the sampled region containing source pixels.
Are images uploaded when I use this image perspective corrector?
No. The script reads the selected file with a local object URL, processes it with browser image and canvas APIs, and creates the PNG download in the browser. It does not send the image to a server.