The future of AI-based image creation

At KTM, the integration of generative AI image generation tools is currently under discussion. This is precisely because we sell highly emotional products and need to ensure that the spirit of the brand comes across in the images.

Gernot Leitner
Team Leader Master Data Management, KTM AG

In a world where reality and digital creation are becoming increasingly blurred, artificial intelligence (AI) has already taken its place. It is revolutionizing the way we create images, write texts and translate content. AI-based image creation tools open the doors to a wide range of possibilities. Designers and creatives can create engaging images in a fraction of the time previously required, something that was unimaginable just a few years ago. From creating photorealistic scenes to abstract works of art, the possibilities are almost endless.

Challenges

While the benefits of these innovations are obvious, they also harbor dangers. The automation of design processes through AI can lead to an alienation from traditional artistic skills and, with the wrong information, can also deliver the wrong results. However, if you are aware of the risks and take countermeasures, the potential of AI image generation can be fully exploited. It allows you to expand your own creative boundaries, discover new forms of expression and, above all, save time and money.

Generative AI in practice

To increase efficiency and versatility in image generation, crossbase has tried generative AI tools such as DALL-E 3, Adobe Firefly and Artsmart and successfully integrated the former into its own platforms crossbase.web and crossbase.pro. This integration opens up completely new functions for crossbase users to bring their creative ideas to life. The connection of further tools is already being planned, depending on the availability of an interface from the provider. In future, users will be able to choose which engine they want to work with or whether they want to use all three for image generation, depending on the licenses. All three tools deliver surprisingly good results.

Simple prompt instructions such as "blue sky, sun and snow-covered mountains" can be used to generate suitable backgrounds for cropped product photos and areas in images can be quickly filled with objects. This function in particular also comes into question for a long-standing crossbase customer, KTM AG. In addition to AI-supported functions that have already been implemented, such as data quality checks or rule-based motorcycle classification, this function could also save a lot of time. When using a cropped motorcycle, for which only the background is generated via a prompt, the actual motorcycle is not altered. In the case of fully AI-generated images, however, it must be checked internally with the marketing department whether the brand can be represented in accordance with the CI and CD specifications.

Outlook

The connection alone is not enough. AI-generated media in particular should be labeled accordingly in order to be able to trace their origin. crossbase is continuously developing its solution to ensure that these and other system requirements are comprehensively covered and do not generate any additional work.

AI at KTM AG

In this interview, Gernot Leitner, Team Leader Master Data Management at KTM AG, reports on how artificial intelligence is used in the company's work.