Best Practice Guide

Introduction

This guide is intended to inform new users in the best practices to use when creating image targets for use in augmented reality (AR) applications. This document includes:

• An overview of basic augmented reality concepts,
• A list of best practices to follow for producing image targets for AR,
• And a list of common issues that can occur when producing and placing image targets.

Following practices listed in this guide when creating image target design and placement will improve performance of your AR content.

Concepts

• Augmented Reality (AR)
  • A technology that superimposes virtual hyper-real elements over a user’s view of the real world, typically using a camera feed.

• Image Target
  • An image or graphic used to identify and localise AR content by an AR camera. See “What is an Image Target?” below for more info.

• AR content
  • Content that is displayed on-screen when an AR camera identifies an Image target. Examples include AR video, but could also include 3D models and animation, interactive content, etc.

• AR Camera
  • A camera feed that has been programmed to detect digital signatures from image targets and display AR content. Typically found in smartphone apps like Weavar.

• Recognition
  • The ability for an AR camera effectively to identify an image target when scanning an environment and provide its matching AR content. Eg. an AR video.

• Tracking
  • The ability for an AR camera to keep the AR content positioned on its matched image target while the camera viewport or the image target is in motion.

• Digital Signature
  • A collection of points programmatically gathered from an image target used for recognition and tracking of that image target.

Icon Legends

The following icons will be used to identify quality of different image targets shown as examples in this document.

Best Practices

This section will detail best practices when designing and placing image targets for use in AR applications.

What is an Image Target?

An image target is any unique combination of images, graphics and text. Image targets are used to recognise and track your AR content when viewed with an AR camera. Adhering to the following tips will produce the best performance for recognition and tracking. 

Uniqueness

Uniqueness is a core requirement for an image target. When generating image targets, key points from the image are gathered to form a digital signature. This digital signature is used to pair the image targets with the correct content. Image targets that are too similar without unique features such as text copy changes or unique graphics will cause a poor experience as an AR camera will struggle to differentiate the two different digital signatures and lead to poor recognition and tracking.

Colours and Contrast

Any colours can be used when generating image targets although having a strong contrast between elements will help produce a rich digital signature.

Size
Size of the AR content is directly linked to the size of the image target, with identical targets producing the same content proportional in scale to the matched image. Eg. A poster sized image target will produce a larger piece of AR content than a business card sized target when appearing in the AR camera view.

The bounds of the content sizing are based on what can clearly be seen by an AR camera. Smaller image targets require users to be quite close to the image to reasonably scan, where as extremely large targets may require users to view the image at extreme perspectives, which can hinder the AR cameras ability to see image targets clearly. 

Sharpness

Elements on image targets should be as sharp as possible as blurry images produce poor digital signatures and consequently affect recognition and tracking.

Text
Text will produce good detail for a rich digital signature and improve tracking, be careful though, text should be used in combination with graphics and imagery as lettering will commonly give similar signatures when perceived by an AR camera. See Logos and Common Elements for more information.

Graphics and Images
Graphics and images provide good detail for constructing a digital signature, especially when used in conjunction with text. Graphics that are commonly used across image targets can confuse recognition and tracking. See Logos and Common Elements for more information. Duplicate image target with exact copy will also produce duplicate information in the generated digital signature also reducing AR recognition and tracking quality.

Logos and Common Elements

Logos and Graphics common across sets of image targets can cause problems in generating unique digital signatures as the points gathered from the common logo/graphic will appear similarly across multiple digital signatures causing instability in image recognition. To rectify this issue, we recommend the minimising shared elements across similar image targets. If the logo/graphic in question is a major element in the image target design we recommend removing the element when submitting the image target to Weavar. Any final print can include this element, but by removing it from the submitted image target essentially “hides” the element from the generated digital signature lowering duplicity and consequently improving recognition.

Common Issues
Common issues encountered when designing image targets are listed below as well as common environmental issues that can occur when image targets are placed in-situ.

Shininess/Matte
Shininess can cause recognition and tracking instability as the reflecting light can obscure the image target when viewed by an AR camera. 

• Avoid glossy print materials when creating image targets for best performance.

Warping
Image targets perform best when viewed flat, depending on the size of your image target, warping may obscure parts of the image when viewed by an AR camera, eg. a cloth banner sagging under weight or a handheld marker bending due to paper/card quality.

• Ensure your image targets are flat where possible to ensure recognition and tracking works reliably

Minor Changes/Revisions
Revisions to image targets can differentiate the digital signature enough to cause reduced tracking quality but can in extreme cases cause issues in recognition of the target when viewed by an AR camera. 

• Ensure the image target design is final when going to print and do not make major revisions to the image target once submitted.

Overlayed/Foreground objects/items
When creating and placing image targets items in the foreground of an AR camera will hinder the recognition of the image target.

• Ensure there is nothing that would appear in the foreground of an AR camera when attempting to scan the image target. Eg. foliage when scanning an image target sign in green areas or other pedestrians walking over image targets placed on the ground in high traffic areas.

Perspective
Larger image targets (eg, ground vinyls larger than 3x3m) can cause problems when a user can’t adequately scan the image target within a reasonable perpendicular angle. 

• Ensure that image targets can be scanned at a perpendicular angle with no greater than approximately 45 degrees of variance.

Camera View
Image targets must fill approximately 50% of an AR camera’s viewport to reliably recognise the digital signature and display the matched AR content.. 

• Ensure image targets are not placed in positions where a user will struggle to get an image target to fill at least 50% of a camera’s viewport