In an earlier post titled Mirrors of the Psyche I touched upon my newly-formed concepts of Authenticity, Accuracy, and Originality. I then wrote some rambling thoughts as to how they might correlate with each other, and how the messy business of AI alteration fits into the picture. Now with the recent explosion of interest into AI arts - with the emergence of projects like Dall-E and Midjourney - I thought of revisiting this topic, and going over the relationship what an artist might have towards these new inventions.
First, it might be pertinent to write a few words concerning this new feature, Text-2-image, which allows AI to conjure forth new images simply from text prompts. In short: type a word or a sentence for the AI to process, perhaps put in some parameters such as "beautiful" or "award-winning" (couldn't help but laugh at that one when I saw it recommended), and lo-and-behold, you get an intricate representation of human-like art, albeit often in a weirdly distorted form. With multiple attempts and tweakings of parameters one can, nevertheless, create very impressive images that seemingly reach the level of master painters. Even more, by adding something like "Van Gogh", the AI can mimic the aforementioned artist and twist its digital brush to resemble his style. A mockery of man's capabilities, or a new dawn for the world of arts as a whole? Or perhaps neither, but a flash-in-a-pan style of spike in interest, something that will soon fade like fifteen minutes of fame. Only time will tell. In any case, this was not the point of this post, as the whole "seemingly radical" new phenomenon is not much of interest to my own project, for I consider these "text-2-images" prompts like fast food of the art-world. It allows for creation of images in an instant, without effort, similarly as if one was to take a job at McDonalds and subsequently claim himself as a master chef. In the creation of (purely) AI generated art, there is no human intention or direction beyond the command one gives for the algorithm to work its magic. Like man shouting for a dog to fetch its stick, the dog obeys, and brings the stick obediently upon the throw. The man can then smile and say "see how well the dog brings the stick", but no effort has the man made to better himself, or made any attempt at conveying any deeper meaning concerning the human condition, as he has merely subjugated this task for someone else. Anyway, I digress, as I tend to do, for AI alteration is not entirely without its merits. It can work as a valuable creative tool when used correctly - well, at least this is my subjective opinion. I have explained reasons for this in previous posts, so no need to repeat all that. What's interesting, now, is that I've noticed when "text-2-image" is used in conjunction with pre-existing images, it allows subtle "enhancement" of the image, not altering it 'too much', but such that the objective quality of it heightens. This, I think, also enhances the "accuracy" part of the trinity of "Authenticity, Accuracy, Originality" - without sacrificing either of the two remaining aspects. Furthermore, these Text-2-image alterations are quite finicky, and often result in worthless failed experiments, and so, it isn't such that there's absolutely no skill required in its use... perhaps yielding some alleviation in assenssing as to how lazy one is in his art-exploration path. * * * Going further with the "Authenticity, Accuracy, Originality" -issue, it can be said that, with enough AI-alteration and skillful editing, the "Accuracy" part can be honed to near perfection, as evident by the beginning example of this post. Of course this is just my own opinion, but I think that "maximum accuracy" for this particular art piece has been reached with this newest iteration. This boost in accuracy mainly concerns the Objective-side (again see my earlier post about subjective-objective divide about art-exploration), but the Subjective side can also later be perfected in editing, giving it the necessary artistical (human) intent. With this, I think one can get as close as possible in honing both sides of the Accuracy -issue without spending one's whole lifetime in mastering the world of visual arts, and doing everything the old-fashioned way. Side-note: and even if spending one's full lifetime in practising, it could very well be that necessary Accuracy still remains unreachable, depending on one's mental and physical predisposition. How does this Text-2-Image experimentation relate to other parts of the trinity then? Well, it obviously does not affect Originality much, if the source image remains the same - that is, if it's taken from somewhere else than the artist's own collection. It does alter the characters or actors - if there are any in the image - and so it might give some room to argue in a court, but that's about it. What of Authenticity? That's a bit harder to assess, but for what I can tell, the original images and edits were "intentioned" with enough Authenticity in mind already - and if one uses just "Text-2-Image" without any initial image, then naturally there's zero authenticity or originality involved. In conclusion, I can say that if "Text-2-image" is used solely by itself, it can serve as a quick but soulless fast-food dispenser of pleasant imagenary, but if used in conjunction - as an additional tool - with pre-existing pieces that are "good" but nevertheless lacking final touches, it can be an interesting feature that yields much-needed aid. Indeed, like a welcome new tool to a toolbox, something for fine work, it can enhance Accuracy of an art-piece, bringing it closer to perfection. After all, any aid is permitted when so much is at stake. For Sophia.
0 Comments
Leave a Reply. |
A.K
Gothic fiction novelist Archives
July 2023
Categories |