image diff
☞ runs entirely in your browser — nothing is uploadedDrop two images here
or click to choose both at once — JPG, PNG, WebP
Now choose the second image to compare against.
OneImageTool
Free Image Tools
image diff
Compare two versions of an image and find every pixel that changed — even the ones you would never spot by eye.
or click to choose both at once — JPG, PNG, WebP
Now choose the second image to compare against.
pixel by pixel
Two exported screenshots, the same banner before and after an edit, a design and its implementation, a render from yesterday and today: they can look identical and still differ. This tool compares every pixel of image A with the pixel in the same position in image B, marks the ones that differ in red, draws a box around the area that changed and tells you how many pixels are affected. Even a single changed pixel shows up in the preview, because a whole cell is marked when any pixel inside it differs.
Three views help you read the result. Highlight shows the changes in red over a faded copy of A. Difference blend shows how much each colour changed: identical areas turn black and changes glow. Switching between image A and image B makes changes flicker in place, which the eye is very good at noticing. A tolerance setting ignores tiny variations such as JPEG noise. Both images are compared in your browser and never uploaded.
three steps
Select two JPG, PNG or WebP files at once, or add one and then the other. The first becomes A, the second B.
Changed pixels appear in red with a box around them, and the count and position are listed below. Raise the tolerance if compression noise is flagged.
Download the current view as a full-size PNG to attach to a bug report or a design review.
quick answers
They are aligned at the top-left corner and compared over the larger of the two sizes. Any area covered by only one image counts as different, and a note shows both sizes. Resize one of them first if you want to compare scaled versions.
Most likely one of them was saved as JPEG again, which changes many pixel values slightly. Raise the tolerance to 10 or 20 and only real changes remain.
The largest difference in any of the red, green, blue or alpha channels, on a scale of 0 to 255. At 0, a pixel counts as different if any channel changed by even one step.
Yes, including their transparency. Pixels that are fully transparent in both images count as the same even when their hidden colour values differ, because those colours are never visible.
Not usefully. It compares exact positions, so a shot taken from a slightly different angle differs almost everywhere. It is made for versions of the same image: screenshots, exports, edits and renders.
No. Both images are opened and compared entirely in your browser, and nothing is sent anywhere.