A Guide to AR Image Targets
Have you ever tried to use an augmented reality app and watched as a character glitched out and randomly slid all over your table? Chances are, that app failed to correctly make use of image targets. Keep on reading, and you’ll learn how to create the perfect target that helps your AR app keep its eyes on the prize.
What is an Image Target?
An image target is a picture that an AR application uses as a physical anchor to map digital images on top of. This anchor is an important part of the AR experience, because it is what allows the user to move the camera around and still have the digital object stay in the correct position in 3D space, creating the illusion of the object existing in the real world.
An image target can be used as an anchor for AR experiences. For example, a base plate or stage for a 3D character, like this frog popping out of a coloring book page:
Or it could serve as the foundation for an entire 3D scene, like this one:
In cases like those, the image target would typically be printed on paper or a card, and placed on a flat surface to create the impression that these 3D scenes are happening on top of the user’s table or floor. However, an image target may be used as an anchor for any kind of AR display, on any surface. In this example, a piece of graffiti art that covers an entire wall has been made into an image target, so that the AR app can play an animated digital version of the art directly onto the wall:
What Makes a Good Image Target?
Technically any picture on any surface can be used as an image target, but there are certain qualities an image can have that can either increase or decrease its readability for AR apps. If an image has poor readability, the program will lose track of the anchor, causing the AR projection to disappear or display incorrectly. To avoid such problems when designing or choosing an image to be a target, you need to know what qualities a good target has.
An effective image target contains as many of these features as possible:
Balance between both large and small shapes.
Avoid making everything the same size in the composition. One possible solution here is to have one big shape be the subject of the image, and include several smaller shapes as background details or patterns.
Multiple varying shape types.
The image should include some shapes that have round edges, some with sharp corners, and combinations between the two. For example, image targets that include text are effective because that naturally creates a lot of unique shape combinations.
Contrast between dark and light areas.
The most extreme example of this is to have a point in the picture where a white shape is placed directly next to a black shape, but this works any time two contrasting color values meet.
Variety and contrast are the recurring themes for all of these points. If the image has a lot of different things happening visually, it will make it easier for the program to recognize it as a unique target and distinguish it from the surrounding background environment.
What Makes a Bad Image Target?
Mostly, the opposite of the good points. A general lack of visual variety will make an image harder to read. This includes:
Low color contrast.
If putting an image through a black and white filter just turns the whole thing into one shade of grey, the app will have a hard time making anything specific out. This also applies to gradients that gradually shift from one color to another- sharp edges of contrast are better.
Too many similar shapes.
If your image is nothing but shapes of the same type and size, it’s not going to make for a marker that’s unique or easy for the program to recognize.
Too similar to other targets (When creating more than one for the same app).
If an app makes use of multiple different image targets, then it’s important for each individual target in the set to be as visually distinct from the others as possible. If they look too similar, the program might pull up the wrong AR element because it’s mistaken one target for another.
Reflective Surfaces
Most target images are printed on some physical material. When doing this, matte materials like standard paper and cardstock are a better option than anything glossy. These targets will need to be viewed through a camera with as much clarity as possible, so materials that create strong reflections and distort the camera’s view of the image are not recommended. This also means that solely relying on digital screens to display your target image is a bad idea.
Examples of Good and Bad Image Targets
This illustration has been run through a program that tests the effectiveness of AR image targets, and has proven to be a good one. The yellow crosses indicate points of interest that the program has picked up on. As you can see, the crosses are most heavily concentrated on the points of color contrast, especially where black and white shapes are right next to each other. Additionally, there’s a good amount of variety in shape types and sizes here, from the large and complex characters, to the smaller snowflake shapes, to the tiny background elements, so this target is easy to read from any angle and is distinct from other targets in its set.
Intentionally designing illustrations to meet good image target criteria is a reliable method of getting a usable target, but it’s not the only way. This photo of leaves also makes for a good target, because the edges of the leaves naturally create a lot of uniquely shaped points of color contrast for the program to latch onto. The strong shadows towards the top of the composition are especially helpful.
On the other hand, this photo of a basic painted wall makes for a very poor target. There is not much variation in shape, size, or color, so there is nothing for the program to work with, and it seems to have locked on to two tiny random spots as points of interest. There’s no chance that an AR program would be able to reliably read this image.
Closing
The most important things to keep in mind when designing or choosing pictures to be AR image targets are visual variety and contrast. Look at your image and try to break down the shapes and colors of the composition by the criteria we’ve listed here. If your image successfully matches all or most of the points discussed in this guide, then you have a reliable and high-quality image target. If your target meets little to none of the criteria, you’ll need to fix the image by incorporating more of these elements, or select a different picture entirely. In what ways can you improve the visual variety of your image targets and enhance your users’ AR experience?
Feel free to share this post to help educate other AR developers and clients.
