What a pixel comparison actually checks
A visual diff lays two versions of the same image over each other and marks every pixel that is different, so if V2 of a homepage only changed the button text, you see a small red patch on the button and a changed share well under 1%.
The tool looks at colour, not meaning. These words help you read what it shows.
- V1 and V2
- The version your client saw in the last round and the new version you are about to send. V1 is always the base the other one is measured against.
- Changed pixel
- A dot whose red, green, blue or transparency value moved further than the sensitivity allows. One changed word in a headline can flip a few thousand of them.
- Changed share
- How much of the whole picture is made of changed pixels, shown as a percentage. It tells you how big the change is, not how important it is.
- Sensitivity
- How far a colour can drift before it counts. At 0 every tiny shift counts; higher values skip noise from compression and soft edges around text.
- Slider view
- V1 on the left and V2 on the right with a divider you drag across, good for showing a client the before and after of one area.
- Overlay view
- V2 faded on top of V1. Anything that moved a few pixels shows up as a ghost or a blur, which makes layout shifts easy to spot.
- Highlight view
- Every changed pixel painted in a strong pink-red over a faded grey copy of V1. This is the view you can download as a PNG.
Sample V1 and V2 pairs
These pairs are made-up examples, not measurements. Each row shows an image size, how many pixels changed, the share the tool would report, and what that share usually means in a review round.
| Example pair | Image size | Changed pixels | Changed share | What it usually means |
|---|---|---|---|---|
| Button text edited | 1,440 × 900 | 4,200 | 0.32% | A small, local change, such as a word, an icon or a colour. |
| Logo swapped in the header | 1,440 × 900 | 26,000 | 2.01% | A visible change to one area, such as a section or an image. |
| Hero photo replaced | 1,440 × 900 | 380,000 | 29.3% | A large change across several areas of the image. |
| Mobile screen, new menu icon | 390 × 844 | 1,100 | 0.33% | A small, local change, such as a word, an icon or a colour. |
| Poster, headline colour changed | 2,480 × 3,508 | 410,000 | 4.71% | A visible change to one area, such as a section or an image. |
| Same file exported twice | 1,200 × 800 | 0 | 0% | No pixel changed at this sensitivity. |
When the changed share misleads you
Both images must line up
If V2 was captured one pixel lower, or the page scrolled a little, nearly everything below that point counts as changed. Take both screenshots at the same width and from the same scroll position.
Different sizes get stretched
When V2 is a different size, it is stretched to match V1 before the comparison. The resize itself softens edges, so expect a light scatter of changes that are not real edits.
Compression adds noise
JPG files and screenshots shared through chat apps lose detail each time they are saved. Raise the sensitivity a few steps, or compare PNG files, before you trust a small changed share.
Big images are scaled down
Images longer than 4,096 pixels on one side are reduced first, so phones and tablets do not run out of memory. A one-pixel hairline change can disappear in that step.
Size is not importance
A price that changed from 49 to 94 is a tiny share of the image and a huge change for the client. Use the highlight to find edits, then read them yourself.
Questions about comparing two versions
Are my images uploaded anywhere?
No. Both images are opened by your browser and compared on your own device. Nothing is sent to a server, and closing the tab clears them.
Why does the tool say almost everything changed?
Usually the two images are offset: one was taken at a different scroll position, zoom level or window width. Line them up, or crop both to the same area, and compare again.
Which sensitivity should I use?
Start at the default of 10. Drop to 0 when you compare two exports of the same design file and want to catch every change. Raise it towards 20 or 30 for JPG photos or screenshots that went through chat apps.
Can I show the result to my client?
Yes. Download the highlight PNG and attach it to your round email, or use the slider view on a call. It is a quick way to say “here is everything that changed since V1”.
Can I compare two different file types, like a PNG and a JPG?
Yes. Both are turned into plain pixels before the comparison. Expect a little extra noise from the JPG, and raise the sensitivity a few steps to hide it.
Does it work on a phone?
Yes. Pick both images from your photo library or files. Very large images are scaled down first so the comparison stays fast and the page does not crash.
How is this different from comparing in a design app?
There is nothing to install and nothing to sign in to, and it works on any two images, including screenshots of a live site, a PDF page you saved as an image, or a client’s photo of a printed proof.