Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Expandora: Broadening Design Exploration with Text-to-Image Model

DaEun Choi, Kihoon Son, Hyunjoon Jung, Juho Kim

TL;DR

Expandora addresses the limitation that text-to-image (T2I) exploration often lacks divergence and non-linear iteration by introducing a structured input interface and a mindmap-like visualization, powered by an LLM-based pipeline. In a within-subject study with 8 designers, Expandora increased prompt diversity and the number of prompts attempted, and raised user satisfaction, while not delivering a CSI improvement due to insufficient convergent support. Key innovations include concreteness-based highlighting to guide exploration, adaptive prompt suggestions, and a non-linear exploration history that supports branching and revisiting ideas. The work demonstrates the promise of non-linear, intention-informed interfaces for T2I-assisted design, while highlighting the need to integrate convergent workflows to fully support creative outcomes.

Abstract

Broad exploration of references is critical in the visual design process. While text-to-image (T2I) models offer efficiency and customization of exploration, they often limit support for divergence in exploration. We conducted a formative study (N=6) to investigate the limitations of current interaction with the T2I model for broad exploration and found that designers struggle to articulate exploratory intentions and manage iterative, non-linear workflows. To address these challenges, we developed Expandora. Users can specify their exploratory intentions and desired diversity levels through structured input, and using an LLM-based pipeline, Expandora generates tailored prompt variations. The results are displayed in a mindmap-like interface that encourages non-linear workflows. A user study (N=8) demonstrated that Expandora significantly increases prompt diversity, the number of prompts users tried within a given time, and user satisfaction compared to the baseline. Nonetheless, its limitations in supporting convergent thinking suggest opportunities for holistically improving creative processes.

Expandora: Broadening Design Exploration with Text-to-Image Model

TL;DR

Expandora addresses the limitation that text-to-image (T2I) exploration often lacks divergence and non-linear iteration by introducing a structured input interface and a mindmap-like visualization, powered by an LLM-based pipeline. In a within-subject study with 8 designers, Expandora increased prompt diversity and the number of prompts attempted, and raised user satisfaction, while not delivering a CSI improvement due to insufficient convergent support. Key innovations include concreteness-based highlighting to guide exploration, adaptive prompt suggestions, and a non-linear exploration history that supports branching and revisiting ideas. The work demonstrates the promise of non-linear, intention-informed interfaces for T2I-assisted design, while highlighting the need to integrate convergent workflows to fully support creative outcomes.

Abstract

Broad exploration of references is critical in the visual design process. While text-to-image (T2I) models offer efficiency and customization of exploration, they often limit support for divergence in exploration. We conducted a formative study (N=6) to investigate the limitations of current interaction with the T2I model for broad exploration and found that designers struggle to articulate exploratory intentions and manage iterative, non-linear workflows. To address these challenges, we developed Expandora. Users can specify their exploratory intentions and desired diversity levels through structured input, and using an LLM-based pipeline, Expandora generates tailored prompt variations. The results are displayed in a mindmap-like interface that encourages non-linear workflows. A user study (N=8) demonstrated that Expandora significantly increases prompt diversity, the number of prompts users tried within a given time, and user satisfaction compared to the baseline. Nonetheless, its limitations in supporting convergent thinking suggest opportunities for holistically improving creative processes.

Paper Structure

This paper contains 31 sections, 2 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Background
  3. Creativity Support in the Idea Exploration Process
  4. Intent Expression with Generative Model
  5. Formative Study
  6. Findings
  7. Expandora
  8. Features
  9. Structured Input for Exploration Intention
  10. System Suggestions for the Expanded Prompts
  11. Iterative Exploration Workflow
  12. Technical Detail
  13. Concreteness
  14. Prompt Suggestion Algorithm
  15. Evaluation
  16. ...and 16 more sections

Figures (2)

  • Figure 1: Usage flow of Expandora. (a) The user begins by entering a prompt, and the system generates four initial images. Additionally, the system highlights the concreteness of each word in the prompt. The system also provides tools for further diversification, allowing users to select a specific part of the prompt to refine, choose how to diversify it, and adjust the desired level of novelty. (b) Based on the input, the system suggests four modified prompts that refine the specified part. Users can either select a prompt to generate new images or request additional suggestions. (c) For further exploration, users can click "Edit Prompt" to return to the interface in (a) and continue diversifying, enabling iterative and targeted exploration.
  • Figure 2: Examples of images generated by P1 and P2 using Expandora and the baseline. Images are displayed in the order they were generated, with the corresponding design topics noted below each set.