How to Vectorize Images
For logos and other digital drawings, this is useful if you only have a bitmap available, but need a vector version of it in order to print it, scale it, or edit it. Vector art can be scaled to any size without any pixelation or blurriness. For photos, vectorization is more of a stylistic effect, somewhat like the rotoscoping effect in the recent movies "A Scanner Darkly" and "Waking Life".
It is traditionally done by hand, with the artist carefully redrawing every detail, but for a couple decades now, there have been a number of attempts to produce an automatic computer tool for performing these conversions. The biggest name tools out there today are Adobe's LiveTrace, which is built into Illustrator, and Corel's PowerTrace, which is built into CorelDraw. Neither of them compare to the quality or ease of use of Vector Magic.
Compared to the other tools, Vector Magic is best at capturing small details, especially in anti-aliased logos and other similar images. It also makes many fewer mistakes with most images, and doesn't look as wobbly as Adobe's LiveTrace or as faceted as Corel's PowerTrace. Take a look at this link to see some comparisons between VM and the other major players:
Vector Magic's user interface is much more intuitive and slick than the other tools. It uses a simple wizard that guides you through the process, asking simple multiple choice questions about your input image and your desired output. In contrast, the other tools use confusing dialog windows with technical control knobs like "path fitting," "minimum area," "blur," and "corner angle." Not only are these unclear in their meaning and purpose, they also don't correspond to the types of thing that the user wants to control, so the process of finding the best settings (which still might not yield a very good result) can be time-consuming and frustrating.
You can try VM out online at:
When you first come to the site they give you two free conversions. After that, you can either buy a subscription for $7.95/month or buy the desktop application for $295. They do offer educational discounts for the desktop application. If you do any serious amount of vectorization, this software is well worth the price.
Overall, Vector Magic has an easier interface and much better results than any other vectorization tool on the market. Delivering it online is also a nice way to let people try it out without having to even download any software.
Ever need to convert from a bitmap image to vector art? Say a client gives you their logo as a BMP or JPG but you need it in EPS or SVG to do your job. For a long time now, there have been software tools to help in this process, but all of them have done such a poor job that most people prefer to just redraw the art themselves rather than clean up the messy result produced by the software.
That has changed with the release of Vector Magic. VM is a software product and online service that grew out of a Stanford University research project. Some guys in the artificial intelligence lab in the computer science department came up with this completely new way to vectorize images. It works much better than existing tools. To see how much better, check out:
It doesn't work on everything, and despite coming out of the artificial intelligence lab, it is not nearly as clever as a human. But it does work really well on medium- and high-resolution images without too much noise. In fact, on digitally rasterized bitmaps, I bet it does better than most humans would do in recovering the original vector image.
They have priced this as a professional tool, which makes sense considering that it is mostly professionals who have ever even heard of vectorization. But if you even need to vectorize a few images per month, I bet it is still profitable to sign up for a subscription or buy the desktop application.
One of the nicest things about VM from my perspective is the care they have taken in the design of the graphical user interface. It is clear that they have put a lot of time into distilling the interface down to the bare essentials.
Beyond the first two conversions, which are free, you can either buy a subscription at $7.95/month, or the desktop application at $295. Both the online and the desktop editions work on both Mac and PC.
Not all of you will have heard of this, but there is a process that is done in the graphic design and print industries called "vectorization." It is also sometimes called "tracing" and the prefix "auto-" is sometimes stuck in front of either word. This is the process of converting a bitmap image - described by a grid of tiny little pixels - into vector art, where the shapes are described with mathematical formulas.
The main benefits of vector art are that it can be scaled without causing pixelation or blurriness, and that it can be edited in a much more intuitive way than pixel-based images.
Anyway, this process is typically done by hand because the automatic tools for doing this just don't work well enough. In fact, a lot of designers find it is faster to just redraw an image than to clean up the garbage most auto-tracing tools produce.
But all that has changed. A couple of researchers out of the AI (that's artificial intelligence) lab at Stanford University have figured out a new way to do automatic vectorization that works a lot better than existing tools. It doesn't work on every image - some images are just too small for a computer to figure out what is going on - but it does work on enough images that it is actually useful.
In technical terms, it works best on medium- to high-resolution images that were digitally rasterized from a vector art original, and that do not contain very many shading gradients. It can handle some noise, but the quality of the result does degrade accordingly.
In basic english, don't expect miracles. If you'd have to make some educated guesses while redrawing it, chances are the computer will not make very good guesses. But if it is a clean image that is big enough to see all the little details easily, this tool should do the trick.
Anyway, they have a website where you can try it out online without downloading any software:
Beyond the first two conversions, which are free, they charge a subscription fee. If you don't like subscriptions, you can also just buy the desktop application, which works on both Mac and PC. The prices are reasonable for any professional who does this task even semi-regularly, as it saves hours and hours of time. And hand-tracing is not exactly fun work!
I'm going to talk a little about vector art and bitmaps and how they relate to each other.
A bitmap is a regular image, like a JPG from your camera, or a PNG on a website. Bitmaps are described by a grid of tiny colored squares called pixels. If you want to increase the size of a bitmap, you have to come up with new pixels to fill in the gaps between the original ones. This process is called "interpolation" and it leads to the blurriness or pixelation that you see when you zoom into a bitmap or scale it up to a larger size.
Vector art, in contrast, is described by the mathematical formulas of the shapes that make it up. As a result, the vector art can be viewed at any size without any blurriness or pixelation. It also means that you can change the image in more useful ways. For example, if your vector image contains a circle, you can easily just change the size of that circle, or even transform it into an ellipse (a type of oval). That would not be possible in a bitmap image.
Converting from vector art to a bitmap is easy as pie. In fact, any program that displays vector art has to convert it to a bitmap just to display it on the screen. The technical word for that process is "rasterization." The opposite process, that of converting from a bitmap to vector art is not so easy. In fact, it is a somewhat ill-posed problem in that there are lots of vector images that "match" any given bitmap, so it is impossible to say which one is correct.
The "standard" way to convert from a bitmap to a vector representation is for a human designer to just redraw the art in a vector editor. This is time consuming and frustrating, but it leads to consistently good results, and it is what people are used to.
There are also a number of software packages for doing this - Adobe has a tool called LiveTrace that is built into Illustrator and Corel has one called PowerTrace that comes with CorelDraw. But these and the other tools on the market just don't deliver the level of quality that is needed, so automatic vectorization has gotten a bad name. And deservedly so. Those tools suck so much that it is often faster to just redraw the art than to clean up the mess that LiveTrace et al. are able to produce.
The good news is that some new research out of Stanford University has recently changed that for a large group of images. It is called Vector Magic (VM) and they've taken it commercial at:
It is not a miracle worker, so don't think you'll get a great vectorized result of your scan of a cocktail napkin, but the new tool does work remarkably well on medium- and high-resolution bitmaps that were originally vector art at some point.
Anyway, you don't have to take my word for it. For some comparisons, check out:
And you can try it yourself for free. Just upload an image to their website and it will guide you through a wizard, leading to your vectorized result. The online tool is really slick, but you can also download a desktop version of the software that is free to try (you need to buy if you want to save the resulting files, but you can inspect them in detail with the trial).
Vector Magic is a new software tool and online service that grew out of a research project in the Artificial Intelligence Laboratory of Stanford University. The tool converts bitmap images into vector art and it kicks the pants off of all the existing tools that claim to do this, even those by graphics heavy-weights Adobe and Corel.
You can try it out for free at:
And if you're curious to see a side-by-side comparison, just check out:
I suppose they cherry picked those results a little, but seriously, if you've used Live or Power trace to any extent, you'll see the difference in quality right away.
Every now and then some new technology comes along that shakes up the established order of things. Sometimes these technologies come in big fields (Google revolutionizing search, for example) and more often they come in small niche areas.
In this case, I'm talking about graphic design. A new technology out of Stanford University for converting bitmap images into vector art has recently been commercialized under the name "Vector Magic."
Even with their billions in revenue, Adobe--the 800 pound gorilla in the graphic design field--has not been able to produce anything even close to the quality of VM. And this is not for lack of trying. Between Streamline and the more recent LiveTrace, which they featured very prominently in the release materials for Illustrator CS2, Adobe clearly knows that this is an important feature for graphic designers and print professionals.
But it turns out that converting from bitmaps to vector art is just a really hard problem. Really really hard. And not just for rich primates. There are probably 30 software tools on the market that attempt to solve this problem. And until VM, they've all fallen far, far short.
You can check out the difference in quality here:
And you can try out the tool itself on the main page:
They offer two free conversions to new users, and then you can select from a subscription or a desktop application version of the software. Both are try-before-you-buy and if you do any serious amount of vectorization, the prices are very reasonable, especially considering that it is a professional tool.
All my hyperbolic praise aside for a moment, VM doesn't solve every vectorization problem. Without solving the AI problem entirely and making a computer that is as smart as a human, some images are just too small or too intricate for a computer to understand what all the pixels mean.
But for a large number of images--especially bitmaps that were at some point in the past digitally produced from vector art originals--VM does a great job. It doesn't do as well on scans and small noisy images, but hey, we also don't have a cure for the common cold.
I've also played around with using it for photos and it has some neat effects on some photos, depending on what you are going for. When using it on photos it reminds me a bit of those Photoshop filters, even though it is doing something a lot different.