davidd's A DOLL A DAY 2024

[davidd]

I wonder what the second wheel is for? 

[insert long diatribe about AI "art", and whether something that takes 10 seconds to generate off a few words has any soul at all in comparison to a doll photo setup that someone spent a couple of hours or more on, doing wardrobe and setting and lighting and so on, and put actual human thought and time and effort and creativity into]

The "Victorian steam yacht" looks more like a "steampunk airship gondola" to me, so I presume the "second wheel" is to adjust fore-and-aft balance, or pitch. 

Does AI art have as much "soul" as tedious miniature photography set-ups? That's an interesting question for which there is no easy answer. There's probably less swearing involved (although maybe not, considering the hand issues). For most casual AI users, yes, there's not much thought going in to the process. People dream up cutesy or quirky one or two sentence descriptions and paste them in to see what the AI app will spit out.

More serious users, however, learn how to use key words to influence output; how to properly sequence terms in a prompt; and they may spend hours or days or longer fine-tuning the AI apps to generate the output they want to see. The apps can in fact be used as tools to create a specific type of output. "Models" can be created based upon a library of sample images (or color or textures or lighting styles). Auxiliary coded algorithms called LORAs can be used to fine-tune "models." Further AI-enhanced tools allow images to be combined and areas of images to be replaced, improved, or enhanced using a process called "in-painting." The AI apps are basically "Photoshop on steroids," offering the option to edit and enhance existing images as well as to create images from scratch using only words.

One of my Instagram contacts, who now posts mostly AI images, started out doing doll photography. Then he began enhancing his doll photos with AI. Then he began working on computer models based on his doll photos. Then he began creating computer models of the various outfits the doll wore. Then he researched different scenarios, both historical and fictional, and began altering the models and LORAs to fit the doll in to settings based on history and literature. He began ex-perimenting with having his doll "interact" with historical figures. So in that case, yeah, there's definitely "soul" (and a lot of time-consuming effort) involved in the AI images.

If I hadn't inadvertently become acquainted with (in an online sense) that particular guy back when he was doing doll photos, I would not have any idea whatsoever about how AI imaging worked, and I would assume "you just type in some words" and create soulless, effortless images. Which is, for the most part, still largely true. But not always.

[EXAMPLES: here's the doll with a model airplane:
https://www.instagram.com/p/C6yJQzXLTD_/

Here's the AI illustrated version of the doll hanging out with Amelia Earhart:
https://www.instagram.com/p/C4KRXyZuIvT/?img_index=1

Kinda cool, huh?]

If you want to go on a "long diatribe," here's something for you: combine your diatribe from last week about real estate listing descriptions with your diatribe about AI! I just got off the phone with the real estate agent with whom I am working on the Florida house. In jest I said something about how I should have looked at his other listings to get a better idea of his creative writing talent. He laughed, but then told me he works with a "listing coordinator" (or some such title) to develop the descriptions "and we use AI."

That goes a long way toward ex-plaining why all real estate listings sound the same anymore, doesn't it? They're not even being written by somebody who has seen the property! The listing descriptions are "coordinated" by somebody using AI tools.

I'm beginning to think that most "writing" for public consumption (newspapers, broadcast media, magazines, newsletters, advertising, and blogs) is largely done by AI apps now. I am acquainted with a woman who used chatGPT to write articles for her company newsletter. I'm guessing that most "content" is now either AI generated or AI  enhanced.

Is AI writing "soul-less?" In my opinion, yes. No question. Yes.

But then, some writing, factual writing, is supposed to be largely soul-less, or at least neutral in tone without bias or opinion. Sadly, very little AI generated text is truly neutral. It's soul-less, because there's no thinking involved. But the programmers behind the scenes definitely have their fingers on the scales when it comes to neutrality. Have a chat with chatGPT or another AI app some time, and take a "politically incorrect" stance on an issue as you phrase your input, and see what happens. 

Is AI image generation "soul-less?" Most of it, yeah; it's people typing in silly prompts to see what happens. There are, however, a few people working to learn everything they can about the image generation technology so they can use it for specific purposes or to generate desired results. There are people working behind the scenes on the apps to make them faster and more detailed and more responsive to user input.

However, the image generating apps are extremely dangerous. "The camera doesn't lie" has never been entirely true, but today, there is not a single image that can be trusted. Not one. The image generating and image altering technology is far too powerful already. Now even the consumer level AI image apps are getting involved in video generation. AI apps create music and voices... and can mimic human voices and speech with remarkable accuracy. Anything seen on a screen or heard through a speaker is suspect.

We are living "in the matrix" whether we want to or not.

Sure, we could "unplug." That's easy enough. If we're not looking at screens, we're not seeing fake images (unless we pick up a magazine or newspaper). Unfortunately, the people in charge, the politicians and bureaucrats, are looking at screens, and they make the decisions that affect us all. What recourse do we, as individuals, have against "the state" making decisions based on AI altered information? What defense do we, as individuals, have against so-called "voters" who cast their votes based on the AI-generated content the media is continually hammering them with?

I tell myself it is worthwhile to at least be somewhat aware of the state of AI in pop culture - image apps, writing apps, etcetera - so I will (maybe) recognize them when they are being used or so I will be aware of when they might be used. But really, it doesn't matter. There's nothing I can do about things.

Additionally, I think the "free" and "consumer level" apps, much like video games, streaming video, and social media, are another form of electronic heroin to keep us addicted to our screens and to keep us from genuinely interacting with other people who are still thinking; and to part us from our money. (Most people aren't thinking anymore, by the way; or so it seems to me. When was the last time you talked about "an issue" with somebody who didn't simply parrot a "party line" identical to what their chosen media outlet tells them to think? When was the last time you had a conversation with somebody who could offer concrete factual reasons supporting an opinion or belief?)

So yeah, I often want to "unplug" and go for a walk in the hills or stare at the river or take a nap; but what will I find when I come home or wake up? Grocery prices are up another 20% over last week, utility prices are the same, the income tax and property tax have taken an enormous jump, and some *sshole is telling me I have to wear a mask and submit to an injection if I want to keep my job or ride the bus or shop for groceries. Even unplugging doesn't work, because eventually we get hungry or thirsty and when we wander back to civilization to grab a Coca Cola or a coffee we find that we can't afford the soda or the coffee and we are shunned because we do not carry the mark of the beast.

And all this because I asked a stupid online game to draw me a picture of a Pullip doll.  

I shoulda asked chatGPT to write this rant internal dialog for me. I wonder what it would have looked like?

And with that: how 'bout another fake Pullip picture:

---

1 August - A Doll A Day 2024:

.

.

.

.


Another AI image generator prompt for a Pullip doll, this time asking for a Victorian style doll wearing a long skirt with a narrow waist.
.
The AI did a surprisingly realistic render of the ungainly wrist joints of a typical Pullip doll. Again, the oversize head, large eyes, and pouty lips are plausible, and a few random stray hairs are visible.
.
I am somewhat concerned about the "faux text" at the bottom right of the third image. I did not include "text" as part of the prompt, nor did I supply a "seed image." The text certainly supports the case that the AI apps are scouring the Internet for image sources, including images likely bearing copyright disclaimers or source credit.
.
AI app: OpenArt.ai