Loading…
It needs WebAssembly and WebGL. Please update your browser or enable hardware acceleration, then reload.
A photograph is not a paragraph. It carries where it was taken, when, and on what, and every one of those fields travels with the file into every message and every shared album. That is why it matters more here than anywhere else in this workspace that nothing you open is uploaded: the whole application is WebAssembly, it runs on your own device, and there is no server to send a picture to.
Every result is re-encoded from pixels, so the metadata does not survive — and that is worth *saying* rather than leaving to be discovered, because it cuts both ways. It is the whole point of the tool that strips a location, and it is a surprise to somebody who wanted their copyright field kept. Each result says which just happened.
Blurring and pixelating are not redaction. A weak blur and a coarse pixelation are both *averages of the original*, which is exactly what makes them reversible — published work has recovered faces and number plates from both. The censoring tool offers four methods, says which two actually remove what was underneath, and puts the number on the result when a setting is too weak to hide anything.
There is no sample photograph, because this workspace ships no image files at all. The example button draws a test pattern instead and says so, which turns out to be the better fixture: a generated image is the same every time, so a comparison between two settings is a comparison between two settings.
Rotate and flip
A quarter turn loses nothing; any other angle resamples, and the result says which.
Fix photo orientation
Turns a photograph the way its own EXIF flag says, and says when there is no flag to obey.
Mirror symmetry
Folds one half onto the other, or unfolds the whole beside its reflection.
Pad to a ratio
Adds margins until the image fits a frame, and crops nothing at all.
Crop to a circle
Takes the middle square first, which is what an oblong photograph in a round frame needs.
Round the corners
Rounds any of the four corners, and warns before a format that would fill them in black.
Add a border
A border of any width and colour, with an inner line that goes on first.
Add a background
A colour, a gradient or a checkerboard under whatever transparency the image has.
Polaroid frame
The instant-camera shape, with the deep band at the bottom left empty.
Add a watermark
Lays a second image over the first, once or tiled. It marks with an image, never with text.
Censor an area
Four methods, and only two of them actually remove what was underneath.
Pixelate an area
Blocks of one colour, with a warning when there are enough of them to read.
Blur part of it
A blur measured against the size of what it hides, because a weak one can be undone.
Convert to grey
Five weightings, and a pure blue comes out anywhere from 18 to 85 depending which.
Invert the colours
The one operation here that is exactly reversible, and it leaves the alpha alone.
Make a colour transparent
Cuts a flat background out of a logo, with a soft edge so the outline is not jagged.
Dither to few colours
Error diffusion or an ordered matrix: one looks better, the other tiles and compresses.
Colouring page
A photograph as black lines on white to colour in, with the threshold in your hands because no single value suits both a face and a hedge.
Add grain
Film grain or sensor noise, and the one effect here that adds colours on purpose.
Glitch effect
Channels split and bands slid: a drawing of a fault, in a file that is not damaged.
Check the dimensions
Size, ratio, format, colours, transparency and metadata, without changing the file.
Pick a colour
The colour at a point in every notation, with whether black or white text sits on it.
Extract a palette
The dominant colours, with each one the average of its group rather than a bucket's middle.
Average colour
Averaged in linear light, which is correct, with the ordinary answer beside it for comparison.
Print size and DPI
What the pixels mean on paper, both ways round. It does not write the density field.
Image to Base64
A data URL, a CSS rule or an img element, with what the third it costs actually is.
Image to ASCII
Characters instead of pixels, with the cell counted as twice as tall as it is wide.
Where the photo was taken
Reads the position out of the metadata and hands back a copy without it. No map is fetched.
Merge images
Side by side or stacked, and nothing is scaled unless you ask, so two screenshots still line up.
Split into a grid
Equal tiles, with the last one in a row left as the remainder rather than padded.
Make a collage
A grid of photographs, each fitted whole into its cell rather than cropped to fill it.
Contact sheet
A printable page of thumbnails, sized by the sheet rather than by the pictures.
Build a sprite sheet
Frames packed into one sheet, every cell the same size, with the coordinate map beside it.
Slice a sprite sheet
Frames of a fixed size, counting the margin and the spacing, and saying what was left over.
Photo mosaic
One picture rebuilt out of many, and it says when the tinting has become an overlay.