Memory Scapes

What is Memory Scapes?

The concept of Memory Scapes extends the meaning and usefulness of everyday objects by adding a digital dimension.It creates a bridge between the physical and digital worlds, allowing personal experiences and memories to be integrated into everyday life, and objects can be provided with useful information. This enables a wide range of applications, for example for education, for storing recipes in kitchen utensils or for memories of specific places.

Saving Media

Users have the option of selecting the media for the objects manually. Alternatively, there is the option of using artificial intelligence (AI) to recognize new related images and suitable WARDs (objects).The AI can not only make suggestions for storage, but can also assist with manual selection by providing useful prompts.

Media Playback

The media stored in the objects can be played back in various ways. For example, they can be displayed on a large screen in landscape mode, or alternatively MemoryScapes allows you to immerse yourself completely in your memories using virtual reality (VR). There is also the option of projecting the saved experiences onto the wall as a poster using augmented reality (AR). These different playback options help to ensure that the memories can be experienced in an individual and impressive way, and the media can be interacted with at any time. This enables the user to move or copy media to another WARD during the playback process, for example.

Implementation with Unreal Engine 5

As part of the project, we worked intensively with the Unreal Engine 5 and developed a prototype for the user interface interaction. A fascinating vision for the future is the possibility of using WARDs (digitized memories) to store memories entirely in virtual reality, send them or store them in an individual virtual space. This opens up the possibility of conveniently taking memories with you and experiencing them via VR glasses.

Gaussian Splatting

Gaussian splatting is another technology we have been working on intensively, and in our example from the video (sunglasses) we were able to generate a relatively precise 3D model of the glasses from over 200 images from all possible perspectives.

Difficulties that still arose in our example were:

- Shiny surfaces

- Close-ups

- Shots from every perspective

- Enough images

-> All these points should soon no longer be an obstacle due to the rapid progress in technology.