PicoBlog

The other night, as we were playing Shanghai Rummy with friends, my daughter kept shouting out, “Mom, focus!”

The funny thing was, that the reasons I was less than focused on the game (I always lose this game anyway, so I’m not sure why my focusing was an issue), because I was showing our friends the program Fooocus.

Yep. That’s it’s name. Three ooos. No idea why.

Fooocus, for those who have not encountered it, is a free, downloadable, Art AI you can run in your own livingroom. It can be downloaded here.

I figured if I was going to talk about AI Art , I should probably have some idea of what I am talking about. And also, it would be a great thing to have more pictures of characters from my books. I use photos for inspiration when I write, so the more I have the better.

Having played around with Fooocus for nearly a week, I thought I would share my humble observations. I will also give some examples of what the program can do.

First: How it works

Fooocus is not like Midjourney or Grok or ChatGPT. These programs are public programs you use on Discord or Twitter (X), etc. Fooocus is a program you download onto your own computer and run yourself. (You do need a pretty decent video card.)

The program allows you to enter descriptions or upload photos, and to adjust how an uploaded photo influences an illustration. It also has a whole series of styles that can be adjusted in different ways—with more constantly becoming available.

The additional styles, etc., can be downloaded for free from a website called Civitai.com

AI and the Law:

The issue of whether or not AIs can draw from copyrighted image is currently being considered by the courts in a number of cases. To a degree, this issue is moot, because if you have downloaded material to your own computer, it’s not like the law can reach its long hand through the Internet into your computer and take it away.

But…

I recall Napster. I recall when the music industry thought it was a goner, and everyone said Napster could never be dethroned. But now? Spotify sells music, and Napster still exists—but it sells music, too. The large-scale theft of music has diminished. And Youtube with ads offers another possible way to make music as well.

This suggests that the Internet is easier to regulate than many of us wish.

Let’s look at Fooocus.

If I download something, it’s mine. But what if I want something new? Civitai is the chokepoint. If the law should come down against the current system, it is the places that are allowing the downloading of programs and materials that the law would go after.

My humble guess is that it would be easier to regulate, going forward, than we might expect, but it could not be done entirely.

AI and Artists:

AI’s drawing pictures are an utter blessing in so many ways. But there is one group of dear people who are not being blessed…Artists. Particularly, the struggling artists who really want to make a living doing something they love.

I know a lot of artists. It’s part of the territory for working in publishing. I’ve known professional artists, and those who would like to be professional, since the 1980s. And when I think about them, my heart just aches.

But actually getting to use an AI gave me a bit of home as well as my fear and sorrow on the artists’ behalf.

Right now, the AIs are a blow to artists, but there are many ways in which AIs may offer artists either help or new opportunities.

First, if the artists puts together a model set (called things like LoRAs and Checkpoints, etc.) for an AI, they could have the AI do some of their work. This would not be a huge help to, say, a portrait artist, but a cartoonist could have the AI help with the kind of thing that a professional lets an In-Betweener do.

(Obviously, an artists could let the AI do some or all of their work, but I am more concerned about the ones who don’t want to do this.)

But there may eventually be more opportunities. Right now, the styles for Fooocus and some similar programs are free. But I can think of a LOT of things that used to be free online and arebn’t now. If there came a time when this sort of material were for sale, it might be a great opportunity for artists—put together model sets on a given topic and sell them to be used with Art AIs.

I can even imagine some companies hiring an artist to make them a model set, so they could get their AI to do something the others could not do.

And if there were artists that offered a service like: “for a reasonable, fee we’ll turn your AI art into something that has a human face and doesn’t have extra people popping up in it for no reason,” I might be first in line.

So there may be a silver lining out there for artists. We shall see…

AI Whisperers:

I think we’re going to see a new job, too. AI whisperer. A person who has learned how to coax what they actually want out of a given Art AI…and who offers this service for sale.

This job is like part computer work, part graphic artist…and if the person is good at photoshop or, even better, an artist themselves, they could quickly fix the things that AIs get wrong.

What am I using it for?

Years ago, we began the habit of picking photos to represent characters. So I have hundreds and hundreds of photos, some of actors, some of random people I’ve found online, representing the characters in The Books of Unexpected Enlightenment.

With Fooocus, I can take these photos and dress them in the kind of outfits my characters wear.

This is tremendous fun…especially as it takes some cleverness to figure out how to get the program to do what I want…and it is not taking away from any artist, because I could never have afforded to do any such thing otherwise.

Also, I enjoy playing with this program for the same reason I love Photoshop so much. I just like seeing how images work together.

Concept Formatter:

Which leads to another future use of AI. For authors who do want cover art by an human artist, it could be really useful to work with an Art AI to work out their concept ideas. Play with it over weeks or months, until you have some illustrations that look, at least reasonably, like your character, your world background.

Then hand these to your cover artists and say, “I want a cover. This is my concept art. Can you give me something that looks a bit like this, but doesn’t have three arms and has these other characteristics?”

(I do this with my cover artist, but I use photographs. “Oh, this character should look like this actor or this random person I found online. Or this location should like 7:23 on this video. (I did that for the River Dart.)

What does it actually look like?

At some point, I may share some of the book-related images with The Roanoke Glass substack, but in the meantime…

Below are a series of examples. All these pictures were produced by the Fooocus AI—all using exactly the same prompt, but altering the styles. (Your email will probably cut this off, but you can see it on the website.)

Note the clock sporran:

Now, you may ask, “Um, Mrs. Wright…” (Actually, you probably are asking, “Um, Jagi,” but if you haven’t met me in person, you’re probably mispronouncing it, which is fine, but Mrs. Wright is safer. ) So…

“Um, Mrs. Wright, Why do these pictures all look like Gerard Butler?”

The answer is that I wanted to use the exact same prompt for all of the pictures, and the prompt I happened to use was:

Young Gerald Butler, long hair, wearing a kilt, smiling

(Though a few were done before I added the long hair.)

To which, you might reply, “Ahem…that’s not what I meant….”

To which, I might reply: The answer to that question would take a whole other article. ;-)

For non Butler-related examples, look here.

ncG1vNJzZmikmpa0qrjApqeloZedwaa%2B1qugoKCkY8C2rtKtmJyjXpi8rnvPaJ2op5%2BYwrR5zqdkmqFdlr%2B1