Wikitude announces updated SDK 8.5 with new image recognition features and improvements

May 24, 2019 – Wikitude has announced it has updated its SDK to Wikitude SDK 8.5. Developers are now able to recognize even more types of targets and can utilize new SDK features and performance updates. According to the company, the update offers even more stability to the platform and introduces two new Image Recognition features:

Transparent Areas in Image Targets

A new Transparent Area feature has been included to support special types of image targets that do not fit the typical rectilinear image shape. Examples of such images containing transparent areas beyond the main outlines include tattoos, stickers, logos, images with cutouts and basically any image file containing parts with alpha channel transparency.

Thanks to the new update, images with transparent areas can now be used as Target Images. Transparent areas will be ignored during the recognition and tracking process, and the feature works with Image Targets at Runtime, not with WTC files.

Image Targets at Runtime

Traditionally, Image Recognition requires developers to create an Image Target collection (WTC file), previous to the AR experience, to be used to detect and track images. As efficient and well-performing as this process may be – and it will continue existing as is, developers now have an additional option to create Image Targets.

With the new Image Targets at Runtime functionality, developers can create Image Targets on the fly, no need for pre-processing. The feature allows for the use of regular images (png, jpg) directly with the SDK as target images. Multiple images can be bundled into a zip file, there is no need for WTC file creation. Working with WTC files remains as is.

Being able to define Image Targets at runtime means users have the ability to easily create AR experiences on the spot. Ideal for quick and convenient testing during the development phase. The feature is also more practical for spontaneous use cases on the go: demos during networking, dynamic art projects, in a classroom environment, adding AR to game cards, and more.

Runtime Targets give Wikitude developers another target creation option. However, the traditional WTC file target collection methods continue to offer advanced functionalities, such as target image performance ratings and manual height value input fields, which are ideal for AR experiences that require fine tuning and precise input for even more accurate distance to target calculation and proper wearable device calibration.

The SDK 8.5 update also brings significant performance advancements to Image Recognition. According to Wikitude, not only are image targets able to be recognized faster, but they are now also recognizable from even further away – more than three meters away in fact. The company tested an A4 / US-letter sized image target and SDK 8.5 was able to recognize it from 312cm away, which marks a +40% increase in distance from previous SDK versions. Wikitude’s performance tests showed that, with SDK 8.5, image targets can be recognized even when they occupy only 1% of the device screen area, meaning the 8.5 update is ideal for use cases in which users do not have the target image within hands reach.

As well as the above noted upgrades, Wikitude SDK 8.5 includes a series of fixes and stability improvements. Developers will notice faster recognition when it comes to duplicate targets, and can expect substantial improvements in image recognition speed and tracking stability. Wikitude has also updated its iOS sample project, which is now based on Apple’s Swift 4.2 programming language, and is shipping sample applications that show the integration of the Wikitude SDK and how to use the APIs to create AR experiences.

Active Wikitude SDK subscribers are entitled to any and all SDK version updates that are released throughout their term. Non-subscribers are able to download a free Wikitude SDK 8.5 trial version for testing purposes.

Image and video credit: Wikitude/YouTube

About the author

Sam Sprigg

Sam is the Founder and Managing Editor of Auganix. With a background in research and report writing, he has been covering XR industry news for the past seven years.