Core Concepts
Leveraging generative AI to create customizable and data-driven color-changing textures on 3D objects while addressing the material constraints of photochromic systems and the design requirements for data-encoded textures.
Abstract
This paper discusses the potential of using generative AI to create customizable color-changing textures on 3D objects using photochromic materials. The authors identify three key challenges:
Material Constraints:
Available Color Space: Photochromic materials have a more limited color space compared to the RGB color space used in typical image generation. Generative AI models need to be constrained to this limited color space.
Color Application Time: The time required to apply different colors varies, so the generative model needs to consider this factor to create time-efficient patterns.
Type of Light Source: The light source used (e.g., projector, LED) imposes different constraints on the texture resolution, speed, and geometry, which the generative model should account for.
Data-Encoded Texture Generation:
Identifying Viable Regions: The generative model should identify the visible regions of the 3D object to place data-encoded information.
Adjusting Visualization Size and Orientation: The generated textures need to be legible and properly oriented on the 3D object.
Generating Texture Style based on User Data: The generative model should be able to create data-driven texture styles, rather than relying solely on user-provided text or image prompts.
The authors propose augmenting existing generative AI models to address these challenges and enable the creation of customizable, data-driven, and physically realizable color-changing textures on 3D objects.