Iconix: Controlling Semantics and Style in Progressive Icon Grids Generation

TL;DR

Iconix introduces a two-axis progressive icon design workflow that decouples semantic richness from visual complexity to generate a coherent grid of icons from a single concept. It combines semantic scaffolding with a chained image-conditioned generation and a progressive simplification pipeline, producing a navigable icon continuum across detailed to abstract representations. In a within-subject study (N=32), Iconix improved usability, reduced cognitive load, and enhanced structured design exploration and diversity versus a baseline ChatGPT workflow. The work contributes a formal method for progressive icon grids, a practical co-creative system, and empirical evidence on its impact, with implications for scalable visual abstraction and human–machine collaboration in design.

Abstract

Visual communication often needs stylistically consistent icons that span concrete and abstract meanings, for use in diverse contexts. We present Iconix, a human-AI co-creative system that organizes icon generation along two axes: semantic richness (what is depicted) and visual complexity (how much detail). Given a user-specified concept, Iconix constructs a semantic scaffold of related analytical perspectives and employs chained, image-conditioned generation to produce a coherent style of exemplars. Each exemplar is then automatically distilled into a progressive sequence, from detailed and elaborate to abstract and simple. The resulting two-dimensional grid exposes a navigable space, helping designers reason jointly about figurative content and visual abstraction. A within-subjects study (N = 32) found that compared to a baseline workflow, participants produced icon grids more creatively, reported lower workload, and explored a coherent range of design variations. We discuss implications for human-machine co-creative approaches that couple semantic scaffolding with progressive simplification to support visual abstraction.

Figures (27)

• Figure 1: The Iconix's interface showcases an example of generated results for "fast food". The system comprises five coordinated modules: (A) Input Concept; (B) Related Concept Analysis (showing candidate ratings); (C) Semantic Exploration (visualizing associations and prompts); (D) Visual Simplification; and (E) the Dual-Axis Icon Grid for comparing design trade-offs.
• Figure 2: Overview of the Iconix pipeline. The pipeline consists of four stages: (a) Concept Ideation identifies related concepts from the input; (b) Semantic Exploration analyzes relations and constructs three-level prompts to generate image exemplars; (c) Visual Simplification transforms the exemplars into icon mask drafts of varying complexity; and (d) Style Refinement optimizes the drafts to generate multi-style icon grids.
• Figure 3: A series of simplified images clustered into nine groups and visualized in 2D via PCA. Red crosses denote cluster centers, highlighting the progression of simplification.
• Figure 4: Workflow comparison between (A) Iconix and (B) the Baseline. Both systems share the same major steps for icon generation. The key difference is the linear, selection-based progression of Iconix compared to the cyclic, prompt-based iteration required in the Baseline. Curved lines indicate iterative operations, while modules with diagonal hatching indicate optional actions. (C) is the Baseline (i.e., ChatGPT) interface used during the study.
• Figure 5: The user study procedure. The study followed a within-subjects design where participants completed two design tasks using both Iconix and the Baseline, with the order of conditions and tasks counterbalanced to ensure fairness.