Ideation phase of the industrial product design process involves drawing multiple forms of sketches, collecting feedback from engineering teams and iterating over the design through sketches and colored renders. These activities involve notable manual effort and time. Recently, Generative AI has demonstrated extensive capabilities towards image generation and has also been adopted to aid in the product design and visualization phases. However, popular general-purpose tools built around such foundational GenAI models typically are unable to cater to the finer domain specific aspects. In this paper, we present an approach to extend Generative AI models to encompass new domain specific visual aspects and components, so that these models can be used for quick iterations over sketches and colored renders by synthesizing specific styles, features and components into an existing sketch, as directed by the designer. The proposed approach allows for rapid visualization and digital prototyping of design ideas as the AI models abstract away typical activities around creating sketch and render feature and design variations, viewpoint variations and drawing/erasing specific components in a sketch. This paper demonstrates the results of the approach on actual data and analyses the efficacy.

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

Using Generative AI for Rapid Digital Prototyping of Industrial Design Sketches and 3D Renders

  • Ashish Jha,
  • Vijayasri Kalyan,
  • Ninad Kulkarni,
  • Ujwal Bhate

摘要

Ideation phase of the industrial product design process involves drawing multiple forms of sketches, collecting feedback from engineering teams and iterating over the design through sketches and colored renders. These activities involve notable manual effort and time. Recently, Generative AI has demonstrated extensive capabilities towards image generation and has also been adopted to aid in the product design and visualization phases. However, popular general-purpose tools built around such foundational GenAI models typically are unable to cater to the finer domain specific aspects. In this paper, we present an approach to extend Generative AI models to encompass new domain specific visual aspects and components, so that these models can be used for quick iterations over sketches and colored renders by synthesizing specific styles, features and components into an existing sketch, as directed by the designer. The proposed approach allows for rapid visualization and digital prototyping of design ideas as the AI models abstract away typical activities around creating sketch and render feature and design variations, viewpoint variations and drawing/erasing specific components in a sketch. This paper demonstrates the results of the approach on actual data and analyses the efficacy.