The synergistic effect of artificial intelligence technology in the evolution of visual communication of new media art

Shuicheng YanHeliyon 2024 · DOI

TL;DR

This paper introduces a novel AI-powered layout design method using Convolutional Neural Networks (CNNs) to enhance visual communication in new media art. The proposed system automatically generates design layouts, demonstrating superior performance in overall assessment, text readability, and visual path rationality compared to traditional design approaches.

What Problem Does This Paper Address?

The rapid advancement of artificial intelligence (AI) is increasingly influencing various creative fields, including new media art. While AI's potential in art creation and design is recognized, there's a need to systematically explore its synergistic effects on visual communication, particularly in layout design. Traditional layout design relies heavily on human intuition and experience, which can be time-consuming and may not always achieve optimal results in terms of efficiency and user engagement. This paper aims to bridge this gap by developing and evaluating an AI-driven layout design method that can automate and optimize the creation of visually appealing and effective designs for new media art.

Methods

The core of this research is an AI layout design model based on Convolutional Neural Networks (CNNs). The process involves:

  • CNN-based Layout Design Model: A CNN architecture is designed and trained to extract key features from input images and design elements. This model learns to understand visual elements, their relationships, and aesthetic principles.
  • Automatic Layout Generation: The trained CNN model is used to automatically generate layout designs. This involves extracting features from new input elements and analyzing their layout to produce design results.
  • Parameter Optimization: The model's parameters and network structure are continuously adjusted during training to optimize its performance.
  • Comparison with Traditional Methods: The AI-generated layouts are compared against traditional layout design methods (e.g., layout, comparison, harmonic composition) using a Likert 7-point scale.
  • Multi-dimensional Evaluation: The evaluation focuses on three key aspects: overall comprehensive assessment, readability of text information, and rationality of the visual path.

Datasets

Not explicitly detailed in the paper, but the study mentions collecting "a large number of layout design samples in various styles" for training the AI model. Additionally, 20 design students participated in evaluating the generated designs.

Key Insights and Ideas

  • Synergistic Effect of AI: The research highlights that AI, particularly CNNs, can synergistically enhance visual communication in new media art by automating complex design tasks.
  • Bridging Design Logic and Computational Logic: The paper emphasizes the transformation of abstract design logic into executable computational logic, enabling AI to understand and perform design tasks.
  • CNNs for Feature Extraction in Layout Design: CNNs are shown to be effective in extracting crucial visual features from images and texts, which are then used to inform layout generation.
  • Multi-dimensional Evaluation is Crucial: Evaluating AI design solely on aesthetics is insufficient; readability and visual path rationality are equally important for effective communication.
  • AI as a Creative Partner: Generative AI can serve as a powerful tool for designers, offering inspiration, automating repetitive tasks, and expanding creative possibilities.

Key Results

The proposed CNN-based AI layout design method achieved significantly higher scores than traditional layout design methods across three evaluation dimensions:

  • Overall Comprehensive Assessment: 5.95 (AI) vs. lower scores for traditional methods.
  • Readability of Text Information: 5.68 (AI) vs. lower scores for traditional methods.
  • Rationality of Visual Path: 5.74 (AI) vs. lower scores for traditional methods.

An ablation study indicated that removing the feature extractor component significantly reduced the model's accuracy, underscoring its importance. The study also noted that deeper and more intricate CNN model structures led to improved performance.

Limitations and Future Work

The authors acknowledge several limitations and suggest future research directions:

  • Interdisciplinary Knowledge Integration: The application of AI in visual communication design requires better integration of knowledge from computer science, art design, and cognitive psychology. Future work should focus on fostering interdisciplinary collaboration.
  • Computational Costs and Training Time: CNNs can be computationally intensive and require long training times for complex design tasks. Research into more efficient algorithms and model optimization is needed to lower barriers to AI design tool usage.
  • Ethical Concerns: As AI becomes more prevalent, ethical issues surrounding intellectual property, originality, and cultural adaptability of AI-generated art need in-depth exploration.
  • Broader Application Scope: Future research could expand the application of this AI layout design model to more art forms and explore the synergistic effects of different AI algorithms.

Summary updated

Abstract

This study aims to clarify the synergistic effect of artificial intelligence (AI) technology in the evolution of visual communication of new media art, thereby exploring an AI layout design method based on Convolutional Neural Network (CNN) in the practice of visual communication design. Firstly, this study designs an AI layout design model based on CNN, and trains and optimizes it with training data. Secondly, the automatic generation of layout design is realized by constantly adjusting the model parameters and network structure. Finally, various AI layout design algorithms are compared, and their effects and performances in layout design generation are analyzed. To verify the layout and composition matching model's performance, traditional layout design methods are selected for comparison (layout, comparison, harmonic composition, etc.). This study involved 20 design students as participants, evaluating them across three dimensions: overall comprehensive assessment, readability of text information, and rationality of visual path using a Likert 7-point scale. The results reveal that the proposed method's evaluation outcomes in these three aspects are 5.95, 5.68, and 5.74, respectively, higher than the traditional layout design methods. To sum up, the generative AI discussed here can automatically generate design elements and schemes through deep learning and big data analysis, thus providing a reference for the innovation of visual communication design.
BibTeX
@inproceedings{yan2024synergistic,
  author = {Shuicheng Yan},
  title = {{The synergistic effect of artificial intelligence technology in the evolution of visual communication of new media art}},
  booktitle = {Heliyon},
  year = {2024},
  doi = {10.1016/j.heliyon.2024.e38008},
}
Figure