The short answer
A raster image (also called a bitmap) is a grid of pixels. Each pixel stores one colour, and the picture is whatever those thousands or millions of dots add up to. PNG, JPG, WebP, GIF and BMP are all raster formats.
A vector image stores instructions instead of dots: "start here, draw a curve to there, close the shape, fill it navy". The computer redraws those shapes at whatever size you ask for. SVG is the most common vector format on the web; AI, EPS and many PDFs can also hold vector artwork.
So the core difference between raster and vector is this: a raster image records what the picture looks like at one fixed size, while a vector image records how to draw it at any size.
How raster images work
Zoom far into a photo and you eventually see little squares. That is the pixel grid. A 1200×800 image has 960,000 of them, and that is all the information it will ever contain. Its resolution, the pixel width and height, is baked in when the file is created.
When you display a bitmap larger than its native size, the software has to invent the missing pixels by blending neighbours. Edges turn soft, then jagged, then blocky. Shrinking is safer but still throws detail away. This is why a 200 px logo copied from a website looks fine in an email signature and terrible on a poster.
Raster's strength is the flip side of the same coin: because every pixel can be any colour, a bitmap can hold any image at all. Skin tones, foliage, fog, film grain and soft shadows are effortless for a grid of pixels and almost impossible to describe as a handful of shapes.
Common raster formats
- JPG / JPEG: lossy compression tuned for photos. Small files, but it leaves faint blotches around sharp edges and has no transparency.
- PNG: lossless, supports transparency. The usual choice for screenshots, logos on websites and anything with crisp edges.
- WebP: a newer web format that can be lossy or lossless, with transparency.
- GIF: limited to 256 colours, mainly remembered for simple animations.
- BMP: an old, usually uncompressed bitmap format with large files.
How vector images work
A vector file is a list of geometric objects. Each object is a path: a sequence of points joined by straight lines or curves, closed into a shape and given a fill colour (or a stroke). The points are stored as coordinates, and curves are stored as a few control points that define a smooth mathematical arc.
Because the file describes geometry rather than pixels, it has no fixed resolution. Display it at 16 px or 16 metres and the software recalculates the outline exactly, so edges stay razor sharp. This is what people mean by resolution independence, and it is the main reason designers insist on vector logos.
Vectors are also editable as objects. You can click a single letter and change its colour, move one shape without disturbing the others, or pull a curve into a new position. In a bitmap, the same edit means repainting pixels by hand.
What is an SVG file?
SVG stands for Scalable Vector Graphics. It is an open standard, and an SVG file is plain text written in an XML-style markup that every modern browser can display directly. You can open one in a text editor and read it, place it on a web page with an img tag, or open it in Inkscape, Illustrator, Affinity Designer or Figma for editing. Cutting machine software and many laser cutting programs import it too.
Other vector formats
- AI: Adobe Illustrator's own working format.
- EPS: an older PostScript-based format still found in print workflows.
- PDF: a container that can hold vector shapes, text and raster images side by side. A PDF is only "vector" if the artwork inside it is.
Renaming a file from .png to .svg, or placing a bitmap inside an SVG or PDF, does not make it a vector. The pixels are still pixels. Only shapes described as paths scale cleanly.
Raster vs vector at a glance
| Raster (bitmap) | Vector | |
|---|---|---|
| Made of | A grid of coloured pixels | Paths: points, lines and curves with fills |
| Scaling up | Goes soft, then pixelated | Stays sharp at any size |
| Best at | Photos, soft gradients, texture | Logos, icons, type, flat illustration, diagrams |
| File size depends on | Pixel dimensions and compression | Number of shapes and nodes |
| Editing | Paint and erase pixels | Select, move and recolour individual shapes |
| Typical formats | PNG, JPG, WebP, GIF, BMP | SVG, AI, EPS, vector PDF |
| Cutters and plotters | Must be traced first | Cut lines can come straight from the paths |
The difference between PNG and SVG
PNG and SVG are the pair people most often confuse, because both are used for logos and icons on websites and both support transparency. The difference is the one in the table: a PNG is a fixed grid of pixels, an SVG is a set of shapes.
In practice that means:
- A PNG logo needs a separate, larger file for every place it is shown big. One SVG covers every size.
- An SVG logo can be recoloured in a vector editor, or with CSS on a website, without redrawing it.
- For simple artwork an SVG is often smaller. The balloon illustration at the top of this page is 800×600 px; its PNG is about 28 KB, and the traced SVG, 18 shapes with 525 nodes, is about 16 KB.
- For a photo the opposite is true. A PNG or JPG stores it efficiently, while a vector version would need thousands of tiny shapes and still look posterised.
When to use raster and when to use vector
Keep it raster
- Photographs of people, places and products.
- Painted or textured artwork where brush marks and soft blends are the point.
- Screenshots, which are pixel-exact by nature.
- Anything with smooth gradients that must look photographic.
Make it vector
- Logos and wordmarks, which must work on business cards, websites and signs.
- Icons and simple illustrations built from flat colours.
- Cut files for vinyl cutters such as Cricut, and outlines for laser cutting and engraving, where the machine follows paths.
- Large-format print: banners, vehicle graphics, screen printing separations.
- Line art and signatures you want to place cleanly on any background.
A good test: if the image could be painted with a few pots of flat colour and a steady hand, it will make a good vector. If it needs a camera, leave it as a bitmap.
Converting raster to vector
Going from vector to raster is trivial: any program can export pixels at whatever size you choose. Going the other way needs tracing (also called vectorizing). A tracer looks at the pixels, decides which ones belong together, follows the boundaries of those areas and fits straight lines and curves along them.
Image Tracer does this in your browser. You open a PNG, JPEG, WebP, GIF or BMP, and it is traced on your own device, never uploaded. It has two approaches:
- Black & white turns the image into a mask of ink and background using a Threshold, then traces the ink. Ideal for logos, line art, text and signatures.
- Colour reduces the image to a small palette (2 to 32 colours, set with Number of colours) and traces each colour region.
Tracing works best on images that were "vector-like" to begin with: flat colours, clear edges, decent resolution. It cannot add detail that the pixels do not contain, and it produces flat fills, not gradients. A photo will come out as a stylised, poster-like picture rather than a faithful copy. The practical side is covered step by step in How to make an SVG from an image.
What is inside an SVG made by Image Tracer
Because SVG is text, you can look inside. A file from Image Tracer is deliberately plain, which makes it easy to edit and predictable in other software.
- One root
svgelement whose width, height andviewBoxequal the pixel size that was traced. The viewBox defines the drawing's coordinate system, so the artwork scales cleanly to any display size. - A list of
pathelements. Every visually separate shape is its own path, so you can select a single letter or region in an editor. - A
fillcolour on each path: black in Black & white mode, the measured palette colour in Colour mode. There are no strokes and no gradients. fill-rule="evenodd", which tells the renderer that an outline drawn inside another outline is a hole. That is how the counter of an O or the gap in a ring is cut out of a single path.- Path data made of straight lines and quadratic Bézier curves. A quadratic Bézier is a curve defined by a start point, an end point and one control point that pulls the curve towards it. Corners between straight edges are kept sharp rather than rounded off.
In Colour mode the Output setting decides how regions are arranged: Separate shapes writes every region as its own cut-out path with shared boundaries and no overlaps, while Stacked layers writes one path per colour, stacked dark to light. The details of every setting are in the documentation.
Frequently asked questions
What is a vector image, in one sentence?
An image stored as shapes described by points, lines and curves, which the computer redraws at any size, instead of as a fixed grid of pixels.
Is a PNG a vector file?
No. PNG is a raster format: it stores pixels. It supports transparency, which makes it popular for logos, but it still goes blurry when enlarged. SVG is the common vector equivalent.
Is JPG raster or vector?
Raster. JPG is a compressed bitmap format designed for photographs. It cannot store paths or transparency.
Can every image be converted to vector?
Any image can be traced, but the result is only useful when the image is made of clear shapes and flat colours, such as logos, icons, line art and simple illustrations. Photos become poster-like, with flat colour areas instead of smooth tones.
Why do cutting machines and laser cutters want SVG?
They follow outlines. A vector file already contains those outlines as paths, while a bitmap has to be traced first to find them.
Does converting to SVG improve image quality?
It makes edges sharp at every size, but it cannot add detail that was not in the original pixels. Start from the largest, cleanest version of the image you can find.
Last updated
