The Art That Poses Back: Assessing AI Pastiches after Contemporary Artworks

TL;DR

The artists'comments revealed limitations of ChatGPT's pastiches after contemporary artworks, which were perceived by the authors of the originals as lacking dimensionality, context, and intentional sense, and seeming more of a paraphrase or an approximate quotation rather than as a valuable, emotion-evoking artwork.

Abstract

This study explores artificial visual creativity, focusing on ChatGPT's ability to generate new images intentionally pastiching original artworks such as paintings, drawings, sculptures and installations. The process involved twelve artists from Romania, Bulgaria, France, Austria, and the United Kingdom, each invited to contribute with three of their artworks and to grade and comment on the AI-generated versions. The analysis combines human evaluation with computational methods aimed at detecting visual and stylistic similarities or divergences between the original works and their AI-produced renditions. The results point to a significant gap between color and texture-based similarity and compositional, conceptual, and perceptual one. Consequently, we advocate for the use of a "style transfer dashboard" of complementary metrics to evaluate the similarity between pastiches and originals, rather than using a single style metric. The artists' comments revealed limitations of ChatGPT's pastiches after contemporary artworks, which were perceived by the authors of the originals as lacking dimensionality, context, and intentional sense, and seeming more of a paraphrase or an approximate quotation rather than as a valuable, emotion-evoking artwork.

Figures (9)

• Figure 1: Bar charts comparing average cosine distances for Orig$\rightarrow$Past1, Orig$\rightarrow$Past2, and Past1$\leftrightarrow$Past2 comparisons across all five models.
• Figure 2: Distance distributions for each of the five models. Note the tight, low-distance grouping of AdaIN-Style versus the wide, high-distance spread of DINOv2 and VGG19, illustrating their respective discrimination power.
• Figure 3: We computed pairwise correlations to see if models that rank one pastiche as very similar to the original (low distance) also rank others similarly. This scatter plots visualize the agreement for all model pairs.
• Figure 4: Visual example of the "Compositional Gap" (High DINO, Low AdaIN). This case illustrates pastiches that successfully mimic the texture and color palette of the original (low AdaIN distance) but fail to capture the structural composition (high DINO distance).
• Figure 5: Visual example of Structural Alignment (Low DINO, High AdaIN). Representing the inverse of the "Compositional Gap," this case shows where the AI successfully replicates the spatial composition and geometric blocking (low DINO distance) but deviates in texture or color statistics.