Cartoon Effect: Turn a Photo Into a Cartoon
Flat cel-shaded colour with real ink outlines, built the right way round: the outlines are measured on your original photo, before the colours are flattened, so you never get false lines drawn across an empty sky.
Drop images here
or click to browse · or paste with Ctrl+V
JPG · PNG · WebP · GIF · BMP · AVIF
Output
Thicker lines read better on large prints and small thumbnails alike.
Fewer levels means flatter, more cel-shaded colour.
Median smoothing removes speckle without blurring the boundaries.
Presets
Your photos are never uploaded. The smoothing, the outline detection and the export all run inside this browser tab.
People also use
Everything this tool does
Outlines found before quantising
Edges are measured on your untouched original, so quantisation boundaries never become false lines. Tested on a smooth gradient with no objects: zero outline pixels, against 1,280 when the order is reversed.
Median smoothing, not blur
Median takes the middle value of each neighbourhood rather than the average, so speckle is discarded outright while the boundaries you are about to ink stay perfectly sharp.
Adjustable line thickness
One to three pixels, grown by proper dilation rather than by lowering the threshold, so thicker lines follow the same shapes instead of picking up new noise.
Three to twelve colour levels
Three gives hard cel shading, twelve is a gentle flattening that still reads as an illustration rather than a photo. Endpoints stay anchored so black stays black.
Ink-only mode
Drops the fill entirely and gives you black lines on white, which is what colouring pages, tracing workflows and vector conversion actually need.
Drag to compare
Before and after share one frame with a slider between them, so you can see exactly which details the outlines kept and which the flattening removed.
Six purpose-built presets
Classic cartoon, comic book, cel shaded, subtle illustration, sticker and ink only, each with its own outline and colour balance.
Batch with one ZIP
Run a whole set through identical settings and download everything as a single archive assembled in your browser.
Transparency preserved
The alpha channel is untouched for PNG and WebP output, so cut-out subjects stay cut out.
Nothing is uploaded
No server, no queue, nothing to delete afterwards. Close the tab and every trace of the image goes with it.
What each control changes
Outlines and colour are two independent halves of the effect. The outline controls decide where lines are drawn and how heavy they are; the colour controls decide how flat the fill behind those lines becomes.
| Setting | What it does | Try | Good for |
|---|---|---|---|
| Outline 20-40 | Only strong boundaries become lines | thickness 2 | Portraits, clean look |
| Outline 60-80 | Most detail becomes a line | thickness 1 | Busy scenes, comic density |
| Thickness 1 | Single pixel lines | large images | Fine detail work |
| Thickness 3 | Bold heavy ink | small images | Stickers and thumbnails |
| Levels 3-4 | Very flat, cel-shaded | smoothing 2-3 | Classic cartoon look |
| Levels 8-12 | Subtle flattening only | smoothing 1 | Realistic illustration |
| Ink only | Black lines on white, no fill | outline 50-70 | Colouring pages, tracing |
2
output modes
3-12
colour levels
0
files uploaded
1-3px
line thickness
ZIP
batch download
Free
no account needed
FlipMyFormat vs other cartoon tools
| Capability | FlipMyFormat | Typical free tool |
|---|---|---|
| Files stay in your browser | ||
| Outlines measured before quantising | ||
| Median smoothing that keeps boundaries sharp | ||
| Separate outline and colour controls | ||
| Ink-only outline mode | ||
| Batch ZIP download without paying | ||
| No watermark on the output |
How to get a cartoon that reads as drawn, not filtered
A convincing cartoon is two separate jobs that happen to be done in the same pass, and the mistake almost everyone makes is treating them as one. The first job is the fill: real cartoon colour is flat, so the photograph has to lose its gradients. That is what the smoothing passes and the colour levels do together, and they work best as a pair. Smoothing alone leaves a soft photograph, and flattening alone leaves ugly stair-stepping where a gradient crosses a level boundary, because there is still noise sitting on top of it. Two or three median passes followed by four to six levels is the combination that reads as cel shading. The second job is the ink, and it has almost nothing to do with the first. Outlines are found by measuring how fast brightness changes in the original photograph, then drawn on top of the flattened colour. Outline sensitivity decides how small a change counts, and it is the one control worth spending time on: too low and your subject loses its own internal detail, too high and every fold in a shirt and every leaf on a tree becomes a line, which turns the picture into a scribble. Thickness is a viewing distance decision rather than an aesthetic one. Set it to one for large images that will be seen full size, and to three for stickers, avatars and anything that will be shrunk, because a one pixel line simply disappears at thumbnail size.
- 1
Add a clear photo
Drag it in, click to browse, or paste with Ctrl+V. Portraits and subjects with clean separation from the background cartoonify far better than busy scenes.
- 2
Set outlines first, then colour
Tune outline sensitivity and thickness until the lines land where you want, then drop the colour levels and add smoothing passes for the flat fill.
- 3
Export as PNG
Press Cartoonify, then download individually or take the batch as one ZIP. PNG keeps the flat areas clean and the outlines hard.
Why the order of operations decides the result
Outlines before quantising, never after
It is tempting to flatten the colours first and then look for edges, because the flattened image has such obvious boundaries. That is exactly the trap. Once colour has been forced into a handful of levels, every place a smooth gradient crosses a level boundary becomes a hard step, and an edge detector cannot tell that step apart from a real object outline. The result is lines drawn straight across an empty sky, along a wall, over skin. This tool measures edges on your untouched original and only then flattens the colour underneath them. On a smooth synthetic gradient with no objects in it at all, that ordering produced exactly zero outline pixels, which is the correct answer for a picture containing no objects. Detecting edges after quantising instead produced 1,280 line pixels on the same 64 by 64 test, every one of them tracing an invisible band boundary.
Median smoothing, not blur
Flattening needs the noise gone first, but a blur is the wrong tool because it softens object boundaries at exactly the moment you are about to draw ink along them. A median filter takes the middle value of each small neighbourhood instead of the average, which means a single stray bright pixel is discarded entirely rather than being smeared into its neighbours, while a genuine boundary between two regions survives untouched because most of the neighbourhood still sits on one side of it. That is why the passes here are median rather than gaussian, and why you can stack several of them and still get crisp edges to ink.
What people use it for
Profile pictures and avatars
Thickness three with few colour levels stays readable when a platform shrinks your picture to a small circle.
Stickers and merch
Flat colour with heavy outlines is exactly what print-on-demand sticker workflows are built around.
Colouring pages
Ink-only mode gives black outlines on white paper that print cleanly and leave space to colour in.
Comic panels from photos
High outline sensitivity with four colour levels gives the dense line and flat fill look of printed comics.
Kids' party invitations
A cartooned family photo prints well and is far more forgiving of cheap paper than a photograph.
Presentation and blog art
A cartooned image sidesteps the stock-photo look and keeps a slide readable at the back of a room.
Game and app assets
Cel-shaded flat colour matches a stylised art direction better than a photograph ever will.
Tracing into vector
Ink-only output has clean two-tone boundaries, which auto-trace tools convert far more reliably than a photo.
Six tips worth knowing
Set the outlines first and the colours second. The lines are what make it read as a cartoon; the flat fill is the supporting act.
Thickness is a viewing distance decision. Use one for large images, three for avatars and stickers that will be shrunk.
If the picture turns into a scribble, lower outline sensitivity. You are catching every fold and leaf rather than actual object boundaries.
Pair low colour levels with two or three smoothing passes. Flattening without smoothing first leaves stair-stepping where gradients cross a level boundary.
Ink-only mode is the one to use whenever something downstream needs a shape, such as a colouring page or an auto-trace into vector paths.
Export PNG. Flat colour with hard black lines is the exact combination JPEG handles worst, and PNG is usually smaller for it too.
Troubleshooting
Lines are drawn across the empty sky
That is the classic symptom of edges being detected after the colours were flattened, and it should not happen here. If you see it, the lines are following real gradients in the sky, so lower outline sensitivity.
The whole image is covered in lines
Outline sensitivity is too high, so every texture counts as a boundary. Bring it down until only object edges survive.
Colours look stair-stepped
You flattened without smoothing first. Add one or two smoothing passes, or raise the colour level count.
The outlines vanish when I shrink the image
One pixel lines do not survive downscaling. Raise line thickness to two or three before exporting anything that will be displayed small.
Frequently asked questions
How does a photo become a cartoon?
Two things happen at once. The colours are smoothed and then flattened into a small number of levels, which gives the flat cel-shaded fill, and separately the outlines of objects are detected and drawn on top in black ink.
Why does the order of operations matter?
Because flattening colour creates hard boundaries wherever a smooth gradient crosses a level. An edge detector run afterwards cannot tell those apart from real outlines, so it draws lines across skies and walls. Detecting edges on the original avoids that entirely.
How do you know that actually works?
We tested it on a smooth synthetic gradient containing no objects at all. Detecting edges on the original gave zero outline pixels, which is the correct answer. Detecting them after quantising produced 1,280 line pixels along invisible band boundaries.
What does outline sensitivity do?
It sets how large a brightness change has to be before it counts as an object boundary. Low values keep only strong outlines, high values turn every fold, leaf and strand of hair into a line.
What line thickness should I use?
It depends on the final size, not on taste. One pixel is right for images shown full size. Three is right for avatars, stickers and thumbnails, because a single pixel line simply disappears when the image is scaled down.
Why median smoothing instead of blur?
A blur averages, which softens the object boundaries you are about to draw ink along. A median takes the middle value of the neighbourhood, so stray pixels are discarded outright while genuine boundaries survive intact.
What do colour levels do?
They set how many distinct values each colour channel can take. Three or four gives hard cel shading, eight or more is a gentle flattening that still reads as an illustration rather than a photograph.
What is ink-only mode?
It drops the colour fill completely and gives you black outlines on a white background. That is what you want for colouring pages, for tracing by hand, and for feeding into an auto-trace tool that converts to vector paths.
Why does my cartoon look like a scribble?
Outline sensitivity is too high for your photo, so texture is being treated as object boundaries. Lower it until only the real edges survive, and consider adding a smoothing pass.
Why are my colours stair-stepped?
You have flattened without enough smoothing first, so noise sitting on a gradient is being pushed to different levels pixel by pixel. Add a smoothing pass or two, or raise the level count.
Which photos cartoonify well?
Portraits and subjects that separate clearly from their background. Even lighting helps a lot. Busy scenes with lots of similar detail become a tangle of lines, because there is nothing to tell the tool what matters.
Does it work on group photos?
Yes, though you may want a lower outline sensitivity than for a single portrait, since more faces means more competing detail. Check the preview at the size you intend to display it.
Are my photos uploaded to a server?
No. Smoothing, edge detection and export all run in your browser using the canvas API. Nothing is sent anywhere and the tool keeps working if your connection drops after the page loads.
Which format should I export?
PNG. A cartoon is flat colour with hard black lines, which is precisely what JPEG handles worst, adding ringing along every outline. PNG stores it exactly and is usually smaller for this kind of image.
Is transparency preserved?
Yes for PNG and WebP output. The alpha channel is untouched, so a cut-out subject stays cut out. JPG has no alpha, so transparent areas are filled with the background colour first.
Can I process several photos at once?
Yes. Add as many as you like and the same settings are applied to each. Download individually or take the whole set as a single ZIP built in your browser.
Is this the same as an AI cartoon generator?
No, and the difference matters. This is a deterministic image filter: the same photo and settings always give the same result, nothing is invented, and nothing about your picture is sent to a model. An AI generator redraws the image and can change faces.
Does it add a watermark or need an account?
No to both. No sign-up, no email wall, no daily limit and no watermark on the output.