Our user study leverages a unique approach by comparing a robust mimicry method directly against a baseline, focusing on participant evaluations of style fit and quality.
Participants were tasked with evaluating which image aligned more closely with the style of original artwork, enabling us to determine the effectiveness of the mimicry method.
We ensured robustness in our study design by randomizing image pairings and selection sequences, thereby minimizing bias and enhancing the reliability of participant feedback.
Unlike previous studies, our method highlights ongoing challenges faced by current AI models in accurately mimicking artistic styles, particularly from unprotected original works.
Collection
[
|
...
]