Introducing Auto Trace Illustration Mode

This one is for all of you who love an efficient workflow. It’s always nice to save time, right?

That’s why we developed Auto Trace, a powerful Linearity Curve (previously Vectornator) tool that turns raster images into clean vector paths in a fraction of a second!

While we previously launched Auto Trace with two different modes for different use cases— Sketch and Photography—the 4.9.0 update includes a new mode especially for illustrations.

Edit: Since publishing this blog, our Auto Trace feature received another major upgrade with a more powerful algorithm that enhances existing modes and introduces new ones. You can now instantly turn raster images into vectors using four different presets: Sketch, Photo, Illustration, and Basic Shapes (new). Learn more here.

Auto Trace Illustration Mode is designed to transform colorful raster artworks into editable vector illustrations.

Import, Auto Trace, Refine

No need to spend hours tracing over lines and shapes with the Pen Tool. You can now import any illustration from raster-based programs like Procreate into Linearity Curve, hit Auto Trace, and watch the magic happen.

Our software will automatically analyse your input image and send back a vectorized version. You can set the level of detail to low, medium, or high. And you can even control the minimum path size by setting the threshold on the slider. A higher value creates longer paths between points, resulting in a less detailed image, and a lower value will generate shorter paths between points, producing a greater level of detail.

Then, you can tweak your design to exactly how you want it. Use the Node Tool to adjust the paths, play around with color by filling in your shapes with different hues, and easily remove backgrounds.

Artwork by: Kelly Llanos @byllanos

Auto Trace is made even cooler by the fact it’s powered by AI. That means that the algorithm learned to produce results similar to what a human would manually create. And it could get even smarter with more data!

To find out more, we interviewed Marko—one of the brains behind the new Auto Trace Illustration Mode.

Hey Marko! Can you tell us a little bit about what you do at Linearity?

Hey, I’m Marko. I joined Linearity about 9 month ago as Machine Learning Engineer and I’m from Ukraine. My job is to work on AI that helps designers to be more creative and productive.

You recently played a big part in giving our Auto Trace feature a huge update. What was the idea behind the new Illustration Mode?

Most of our users use Linearity Curve to work on illustrations, and we are eager to help them be more productive. We also know that often this process starts with the existing artworks drawn in raster or some that people found on the internet. Therefore, we wanted to build a tool to help people trace their illustrations faster and be able to edit them afterward.

How did you go about developing the new mode and how does it actually work?

There is a saying that “a good programmer is a lazy programmer.” And I think this is especially true for machine learning engineers. We let a neural network do the job and learn how to vectorize images on its own by showing it thousands of examples traced manually by artists. With enough data, it figures out how to do it similarly to humans and becomes more robust towards noise and compression artifacts. An exciting part - it can get much better over time!

What challenges did you face when developing the feature?

The neural networks have millions of parameters and require a lot of computational resources. Also, to produce the best possible results, we need to work with high-resolution images. Therefore, making it work on mobile devices can be challenging and not always trivial.

The other one was coding in C++ and optimizing the pipeline for a high-performance.

It’s very exciting to see how powerful new processors become and how this enables AI in different workflows.

Can you explain the difference between Sketch Mode, Photography Mode, and Illustration Mode?

Each Auto Trace mode has pros and cons for different use cases. That’s why Curve automatically recommends the best suitable algorithm for every image.

Sketch Mode follows the contours of elements and works only with black and white images. Photography Mode tries to group similar colors into shapes without prior knowledge of what these shapes might look like.

Meanwhile, the new Illustration Mode tries to recognize paths and shapes that artists had drawn and fill them with colors. This makes results more editable and robust to the noise or compression artifacts.

What kind of illustrations work best with the new Auto Trace mode?

It works well with a flat design, polygon art, cartoons, comics, even a certain type of paintings (like Maria Prymachenko).

I also had a lot of fun tracing artworks from Instagram with the #vectornator hashtag and see how well it works.

There are two ways our users can control their Auto Trace results: the Min Path Size slider and the Level of Detail input. Can you give some advice on the how to get the best results?

Level of Detail defines a trade-off between the noise and how detailed the tracing will be. Depending on your use case and an illustration, you might want to change the setting to get the best result.

Min Path Size defines the minimum size of the element you want to keep on your artboard. For example, you might want to get only the basic shapes and draw smaller details on your own.

Thanks, Marko!

If you haven’t tried Auto Trace yet, update to 4.9.0 to enjoy the new Illustration Mode. And don’t forget the share your results with us on social media!

With Linearity 4.9.0, we also introduced a Merge Layers functionality. Find out more on our Learning Hub.