Visit grebenshyo’s MakersPlace profile
Brady Walker: Can you introduce yourself to our readers?
grebenshyo: I am known as grebenshyo in the art world, a moniker I’ve chosen to align with the ethos of “crypto art.” While my real identity remains undisclosed, I am currently based in Zurich, Switzerland.
My artistic journey began with graffiti and evolved into formal art education. Over time, I’ve transitioned towards the digital medium. After dabbling in mixed media, I eventually devoted my work entirely to digital art, specifically focusing on 3D and AI in recent years.
My creations reflect my unique perspective, underscoring the relationship between the observer and the observed. In this sense, you could say that my art communicates about communication itself, unfolding in a self-referential and recursive fashion. It is in the deviation from this cycle that art – and life for that matter – truly manifests itself, although this is not something we can necessarily control. Broadly speaking, my approach to art is improvisational. However, I often delve into scientific themes such as neurology and optics in my work.
BW: What has AI allowed you to create that you never would’ve otherwise?
g: AI’s standout quality in art creation lies in its speed. Although its distinctive quirks and glitches set it apart, fundamentally, AI—despite its profound capabilities—only duplicates the tasks it’s been programmed to perform. From a conceptual standpoint, it doesn’t produce more novelty than what has been input by the programmers, thereby challenging the concept of its “intelligence.”
AI systems, whether they work via “prompting” or “image-based input,” quickly hit their limits. While AI is undoubtedly a permanent fixture in our future, it doesn’t substitute every facet of art. However, it does supply a vast pool of “assets” that artists can tap into. Moreover, its self-generating (autopoietic) nature ensures it evolves concurrently with the input it receives, rendering its potential virtually limitless.
This entire response serves as an example, as it is nearly a direct rephrase of my original input—twice as long—generated straight from ChatGPT.
BW: Can you describe your process from pre-idea to finished piece?
g: My artistic process is systemic, where each piece has to fit within the overarching framework of my practice. It has to align with my project goals, which heavily incorporate aspects of neuroaesthetics, and must also consider practical necessities such as the steps required for completion and the financial viability of the process.
With respect to AI, I’ve conducted numerous raw studies using popular image synthesis generators like Midjourney, DALL-E, and Stable Diffusion. I’ve also delved deeply into a technique known as outpainting, a process that involves generating sections of images by extending and refining a central core input. However, these are not straightforward “push-button” tasks. They largely involve self-trained models derived from years of my past artwork, predominantly 3D. Broadly speaking, I view my work as a comprehensive project that involves a variety of tasks and disciplines.
From a more practical standpoint, apart from the standard AI workflows, I’ve learned to curate a large library of diverse image inputs to take advantage of the tool’s rapid iteration capabilities. This practice fosters a dynamic interchange between specific concepts and references of any nature, a facet of the process that I find invigorating. Moreover, tools like ChatGPT and other large language models (LLMs) have become a fundamental part of my creative toolkit. They are particularly beneficial for providing coding assistance, which is often required when crafting 3D and generative art.
BW: Do you have any practical tips for going from a total AI art beginner to a true AI artist? (Take the term “true AI artist” however you like)
g: Thank you for your thought-provoking question. Despite common assumptions, I do not identify myself as an “AI artist”. Even though I incorporate AI in my work, I do not subscribe to the prevalent idea of “AI art”, where artists abandon traditional mediums to use only corporately-provided generative models. This narrative often leads to misconceptions about the capabilities of image synthesis tools. It’s crucial to understand that AI generation fundamentally depends on randomness. Any guidance is not derived from universal truths, but rather from a specific set of preconditions established by those who manage the system – either the artist, when providing external inputs, or the programmer overseeing the software’s operations.
By way of analogy, we don’t categorize artists as “oil paint artists” – the tools themselves do not define the art. The act, signified by the “-ing” ending (as in “painting”), is vital to distinguish the mechanical from the creative aspects in any artistic approach. Recognizing this distinction is critical for understanding the role of AI in art.
In essence, the journey is the path itself.
BW: Can you give our readers a few prompt snippets or style prompts to experiment with?
g: Certainly. My first minted AI piece included the entire prompt in its description, a conceptual approach I’ve repeated with pieces across various platforms—you’re very welcome to check it out!
Moreover, I think that image-to-prompt and image-to-image generative techniques largely reduce vocabulary’s importance, shifting focus to the mathematical construct where input meaning becomes less crucial than one’s contributions to the process. For instance, Midjourney’s “blending” feature, a so-called “image-to-prompt-to-image” operation, requires no words at all (at least on the front end)! Furthermore, when training my own models, like using the “Dreambooth” technique, I even shape the system’s “lexicon” by inventing “words” to impart concepts to the program. A prompt from my custom model might look like this:
{super dynamic, dramatic light, toon color scheme, stylized 3d rendering of exploding (((zwx))) fluffy clouds device with cables, concept art in style of graffiti artist daim, moebius, studio ghibli, katsushiro otomo, masamune shirow, james jean and sat one, intricate, highly detailed, unreal engine, trending on artstation, 64K, HD, HQ}
I know, the clunky example above might appear outdated in the face of the ongoing discourse about the “ethics of AI art,” though I want to share exactly that in order to emphasize how, personally, I am not striving to be an AI art “purist” in search of the “perfect” prompt. Instead, my primary interest lies in comprehending the process and directing my attention toward a form of art that focuses less on the “what” and more on the “how.”
