Artist Styles
We’re about to dive into some of the most controversial parts of AI generated art. Starting with including specific artist styles in our prompts.
One of the main arguments against AI use in art is the uncertainty surrounding copyright infringement.
It is on thing when an artist’s work is used in training data without consent but if AI can be used to blatantly mimic and artist’s work closely there are understandable concerns about taking work away from living artists. Furthermore, it is likely to be decades before copyright law catches up with the advancements in AI.
In this exhibition, all the artworks in an artist’s style are intended for educational purposes only. We want to highlight the issue surrounding prompt engineers and encourage discussions around the pros and cons of using AI in art.
See if you can match the artist’s names with the artworks in this section. How closely do you think the images resemble the original artist’s style?
Vincent Van Gogh
Frida Kahlo
Edwin Landseer
Kehinde Wiley
Ed Ruscha
Vincent Van Gogh - Undertaking the ambitious task of recreating a famed painting of a café in Arles, France, this piece attempts to inject a contemporary spin into a classic masterpiece. Night-time renderings pose a distinct challenge – to depict the nuances of light as perceived by the human eye, a task often beyond the capabilities of a camera. In its attempt, the DALL-E model presents a commendable emulation of Van Gogh's distinctive brushwork. It successfully mimics the iconic painter's unique manner of portraying the warm, soft glow of nighttime illumination. These stylized strokes resonate with Van Gogh's expressive technique, weaving a textural tapestry that is remarkably reminiscent of the post-impressionist master's touch. While maintaining a clear dialogue with the original painting, the piece also introduces modern elements, bridging the gap between classic and contemporary art. Prompt: a cafe in Bristol at night in the style of Van Gogh Model: DALL-E v2.0 (2022)
Vincent Van Gogh - The updated model offers a different interpretation of Van Gogh's distinct style. Here, we witness more of the swirling patterns characteristic of the painter's final works, particularly those created during his time as a patient at the St Rémy asylum. The digital medium of DALL-E continues to struggle with fully emulating the impasto texture and richness of oil paint. This also triggers a significant dialogue concerning the copyright implications of AI-generated images. The obscurity surrounding the specific training data utilised to create these images poses a complex problem. But, when a certain artist's style is explicitly requested in the prompt, the lines of copyright infringement may potentially become clearer, and the issue more pressing. This consideration becomes particularly pertinent when considering living artists whose works and livelihoods could be undermined by AI reproductions. Prompt: a cafe in Bristol at night in the style of Van Gogh Model: DALL-E v2.Exp (2023)
Ed Ruscha - In 1963, the renowned American Pop artist, Ed Ruscha, curated a collection of images titled 'Twentysix Gasoline Stations.' This book is a photographic journey, capturing fuel stops along the iconic Route 66, exuding a palpable sense of Americana. Our goal here was to encapsulate the quintessential feeling permeating Ruscha's work. Indeed, this image manages to evoke a profound sense of desolation, emblematic of a solitary road trip across the expansive American landscape. The scene appears abandoned, bereft of attention, and seemingly forgotten by time - elements that heighten the raw, emotive appeal of the image. Prompt: an Americana image of a deserted petrol station in the middle of nowhere in the style of Ed Ruscha Model: DALL-E v2.0 (2022)
Ed Ruscha - Another fuel stop lost in time. The experimental model has created a more stylised photograph. It is also worth noting how DALL-E still struggles to portray coherent text. In both of these images we see a significant deviation from Ruscha's original approach - colour. Ruscha's monochromatic photographs carried a distinct vintage allure and a stark simplicity, qualities that both models, in their vibrant colour renderings, seem to overlook. Prompt: an Americana image of a deserted petrol station in the middle of nowhere in the style of Ed Ruscha Model: DALL-E v2.Exp (2023)
Kehinde Wiley - Kehinde Wiley, renowned for his portraiture, masterfully reimagines Old Master paintings by centring Black protagonists against ornate floral backgrounds. While Wiley's oeuvre predominantly highlights masculine figures, we sought to engage the AI in envisioning his work with a shift in focus towards a feminine subject. The 2022 model's attempt has a certain level of naivety but undeniably reveals the influence of Wiley's work, suggesting its presence in the model's training data. Wiley's signature style, with its unique blend of classical poses and vibrant patterns, is virtually unmistakable. Hence, even in its simplified rendition, the AI-generated piece manages to resonate with the compelling visual language established by Wiley. Nonetheless, the image opens up a discussion on AI's capability to interpret and adapt the subtleties of an artist's style, Prompt: a woman on horseback in front of a background of plants and flowers in the style of Kehinde Wiley Model: DALL-E v2.0 (2022)
Kehinde Wiley - The experimental model marks a significant departure from the previous image's naive execution, crafting a stunning reinterpretation of an Old Master painting. The floral backdrop in this rendering exudes a delicate grace, suggesting the artful influence of Wiley's skillful hand. This captivating image, however, raises deeper philosophical and ethical questions. It beckons us to question the very essence of creativity, originality, and the nature of artistic ownership. Should AI-generated images, like this one, that closely mimic a living artist's work be created? Is this an homage or a blatant infringement of copyright? Such questions will shape the future discourse of AI in the art world, challenging our understanding of the boundary between human creativity and machine learning. Prompt: a woman on horseback in front of a background of plants and flowers in the style of Kehinde Wiley Model: DALL-E v2.Exp (2023)
Frida Kahlo - Those familiar with Kahlo's work would readily identify these renderings as derivative. While the compositions and subjects align with Kahlo's iconography, the execution appears overtly cartoonish — a stark departure from the authentic aesthetic Kahlo embodied in her work. Such discrepancies could stem from the generality of the given prompt, or the potential scarcity of Kahlo's works in the model's training data. This observation raises a pertinent question: To what extent were women artists represented in DALL-E's learning process? The quality and diversity of an AI's training data significantly influence the output. As such, the underrepresentation of female artists could result in an unbalanced and skewed rendition of the art world. It brings to the fore the need for broader, more inclusive datasets to truly capture the depth and diversity of global artistic heritage. Prompt: the most surrealist image in the style of Frida Kahlo Model: DALL-E v2.0 (2022)
Frida Kahlo - The experimental model has created a much more striking composition but still a departure from Kahlo's style. Both of these creations lack the finesse and nuanced subtleties that so unmistakably characterise Kahlo's artistic hand. Although we cannot be sure if this lack in style is cause by underrepresentation, what might be the wider implications of a lack of inclusion of women in AI training data? Prompt: the most surrealist image in the style of Frida Kahlo Model: DALL-E v2.Exp (2023)
Edwin Landseer - If you've ever found yourself in London's Trafalgar Square, you've likely come into contact with Landseer's work—his sculptures of the regal lions that stand guard at the base of Nelson's Column. In the image at hand, there is an identifiable resemblance to a fox, demonstrating that the 2022 model exhibits a stronger proficiency in animal likenesses compared to its human renditions. Yet, the style of the work bears more resemblance to a pastel technique, a medium that Landseer is not known to have used, creating a discordance between the artist's trademark style and this representation. Furthermore, the image doesn't succeed in capturing the grandeur of Landseer's original painting, 'The Monarch of the Glen'— a component that was explicitly indicated in the prompt. Prompt: a majestic looking fox in the snow in the style of Monarch of the Glen by Edwin Landseer Model: DALL-E v2.0 (2022)
Edwin Landseer - This particular set of images beautifully illustrates the advanced capabilities of the new model in capturing the intricate nuances of both human and animal likenesses. The grandeur and sophistication of the original painting find resurgence in this strikingly realistic representation of a bushy fox. The fox, with its ears pricked in alertness, is rendered with such finesse that it stirs our curiosity. We find ourselves pondering over the unseen object or event that has captured the fox's attention. At the same time, we are captivated by the creature's stunning beauty. This evocative image is a testament to the AI's refined understanding and execution of complex organic forms. It not only captures the physical attributes of its subject but also imbues it with an emotional resonance, thus elevating the overall viewing experience. Prompt: a majestic looking fox in the snow in the style of Monarch of the Glen by Edwin Landseer Model: DALL-E v2.Exp (2023)