Thesis
Generative pictorial expressions: synthesizing ideas through text-to-image generation
Master of Science (M.S.), Drexel University
Sep 2022
DOI:
https://doi.org/10.17918/00001400
Abstract
Creative collaborations between human and machine have been sparked by new milestones in AI-generated digital art. This study focuses on a phenomenological methodology to evaluate human interpretations of these works. With the development of models like Dalle-2 and Midjourney, which can create incredible images, there have been many significant developments in the field of AI-generated art. However, most of the research in this area focuses on the technical and aesthetic features of these programs and has simultaneously sparked much debate into the meaning and artistic value of the works they produce. This study proposes a methodology for studying the more subjective value of these artworks. It focuses on how a piece of art can elicit a shared interpretation from a large group of viewers, as well as what related interpretations are available and how viewers come to their conclusions. The experiments have been conducted to assess the viewer's perception of the intention of the artworks by using phenomenology as the qualitative method. Stimuli consisted of three artworks that were created using descriptions of untranslatable phrases that were then relayed to a diffusion model image generator. The resulting digital art objects were then presented to a human audience for interpretation and deduction. The collected survey data showed a strong correlation with the original meaning, and in some cases, the untranslatable input phrase was identified. It was concluded that the artworks generated for this experiment expressed human thoughts through language and then transformed them into images inspired a wide range of diverse and imaginative interpretations. There were a wide range of interpretations that were heavily influenced by individual background knowledge and culture, and in many cases, they pointed in the direction of the intended meaning of the image.
Metrics
40 File views/ downloads
101 Record Views
Details
- Title
- Generative pictorial expressions
- Creators
- Anjelikal Rogers
- Contributors
- Emil Polyak (Advisor)
- Awarding Institution
- Drexel University
- Degree Awarded
- Master of Science (M.S.)
- Publisher
- Drexel University; Philadelphia, Pennsylvania
- Number of pages
- x, 66 pages
- Resource Type
- Thesis
- Language
- English
- Academic Unit
- Digital Media; Drexel University; Antoinette Westphal College of Media Arts and Design
- Other Identifier
- 991019135509504721