"Imagine a Dress": Exploring the case of task-specific prompt assistants for text-to-image AI tools
Letter Of Changes: We greatly appreciate the feedback from the anonymous reviewers. Below are the revisions we
have made to the paper:
• To address the design of prompt assistants, we have added a section mentioning the
implementation of a decision tree in the prompt assistance design. The text is discussed in
Section 3, "Method."
• In response to comments on statistical comparison methods, we previously compared
averages and standard deviations in Section 5, "Results," of our initial submission. We
have now improved our statistical analysis by incorporating an empirical cumulative
distribution function (C.D.F) figure, i.e., Figure 8, and corresponding text to diversify our
statistical comparisons.
Additionally, we would like to clarify that certain issues raised were addressed in our initial
submission:
• For the calibration of subjective scores, we developed an accuracy assessment system
featuring 11 standards to ensure precise calibration in our initial submission. This
methodology and its results are discussed in Table 2 and the accompanying text in
Section 5, "Results." The approach standardizes evaluations across individuals with
different standards, thereby making the evaluation process more fair and accurate.
• Concerning the performance differences between females and males in the dressgeneration
task, this analysis was included in Section 5, “results” and is also referenced in
Table 1.
• To address asking participants to design men’s garments. We chose to limit our design to
a single, highly-specific task by selecting one garment, as each individual garment type
can have a significant number of options and design elements. Future iterations of this
work may expand to include multiple garment/object types.
Keywords: generative AI, prompt engineering, image generation, fashion
TL;DR: A task-specific prompt assistant helps users assemble prompts for text-to-image generative tasks such as designing dresses, improving satisfaction and accuracy over those that do not use a task-specific assistant.
Abstract: In this paper we explore the impact of task-specific prompt assistants for text-to-image generative AI tools through a user study.
Participants were asked to recreate a dress with SDXL using either a prompt assistant tailored to dress design, or, no assistant at all.
A detailed analysis of the results and feedback suggests that for this specific task, a tailored assistant improves result satisfaction and accuracy.
This style of assistant helps users focus on the task by providing a detailed, visual and organized approach to describing the object---enabling faster production times and more accurate descriptions with less ambiguity.
Submission Number: 28
Loading