#making | Explore Tumblr posts and blogs

talenlee · 20 days ago

Text

Life With Generative Tools

In 2023, back when my posts were still being shared to Twitter because the API wasn’t paid-only, I wrote an article about the potential ramifications of generative art media going forward. My concern in the immediate was that the tools weren’t going to go away, but also the potential harm to artists was as much about general economic precarity and not people using fanart to make their D&D characters. I further added to this with a consideration of how I wanted to avoid using generative art in my game development because I didn’t want what people would say about it. That is, a social pressure about the art is what keeps me from using it, not a personal philosophical disposition. I’m an artist who already works with collage and constraints, this feels like a handy way to have something I can play with.

Well, it’s been a year and change and a sort of AI Art Apocalypse has happened, and if you’re not aware of it, it’s because you’re someone who avoids all of the pools that have been so thoroughly pissed in that they are now just piss. If you’re at all related to any part of the internet where people share a bunch of images – which is to say a lot of social media – then you’re already dealing with the place crawling with generative images. Whether it’s a fanart booru, or big sites like facebook and twitter, or god help you deviantart, there is a pretty clear sign that anywhere that opened the door to generative art became a space overwhelmingly for generative art.

I teach about this subject now and I have had some time with it in a situation away from the internet, and I’d like to give you some insights into what this stuff is for, what it does, why you shouldn’t use it, and ways it can be useful.

Content Warning: I’m going to be talking about these tools as tools that exist and leaving the philosophical/ethical arguments about ‘art theft’ and their genesis aside. I’m not including any examples. No shrimp jesus jumpscare.

You might notice I’m saying ‘generative art’ and not ‘AI art.’ Part of this is because I don’t want to buy into the idea that these tools are ‘artificial intelligence.’ Ironically, ‘AI art’ now has less of an implication of being ‘Artificial Intelligence’ and is much more of an implication of ‘it’s ugly shiny art of shrimp jesus with badly spelled signs.’

I want to focus for this conversation on generative graphical tools, and I want to do that because I don’t have much experience with the other types. The textual generators offer me something I don’t really need? I already make a ton of words of dubious quality. Those are actually the things that concern me because their natural aesthetic is authoritive and comprehensive and that’s why it’s a problem that they’re being used to present any old nonsense that may just be straight up wrong. I don’t use those tools and I avoid the platforms that use them so I’m not familiar with them.

Things Generative Art Is Good For

I already use art I don’t own, a lot, for playing. Every day for the past three years I’ve shared a custom Magic: The Gathering playing card, a game I don’t own the rights to, using a card face I don’t own the rights to, and artwork from an artist on Artstation whose artwork I did not pay for or even ask for. This is generally seen as a totally reasonable and acceptable form of playful, transformative media generation and I at no point pretend I have any rights to the material. If I take a picture of someone famous and put a speech bubble over their mouth saying ‘I drink farts,’ if I, as tumblr says, play with jpgs like dolls, that is by no means being done with rights and permission.

Which means we’re already aware that there’s a way of playing with images that both violates copyright but is generally okay to do.

The metric I use for this is if the thing you’re using generative art for doesn’t matter, then it doesn’t matter. If you’re not going to try and claim money, if you’re not going to put it on a marketplace, if you aren’t going to try and claim ownership and profit off generative material, I think you’re probably fine. I mean probably, if you’re using it to say, generate revenge porn of a classmate that’s an asshole move, but the thing is that’s a bad thing regardless of the tool you’re using. If you’re using it to bulk flood a space, like how Deviantart is full of accounts with tens of thousands of pictures made in a week, then that’s an asshole move because, again, it’s an asshole move regardless of the tool.

If you’re a roleplayer and you want a picture of your Dragonborn dude with glasses and a mohawk? That’s fine, you’re using it to give your imagination a pump, you’re using it to help your friends visualise what matters to you about your stuff. That’s fine! It’s not like you’re not making artistic choices when you do this, cycling through choices and seeing the one that works best for you. That’s not an action deprived of artistic choice!

There are also some things that are being labelled as ‘AI’ which seem to be more like something else to me. Particularly, there are software packages that resize images now, which are often calling it ‘AI upscaling,’ which it may be using some variety of these Midjourney style models to work, but which serves a purpose similar to sequences of resizes and selective blurs. There are also tools that can do things like remove people from the background of images, which is… good? It should be good and easy to get people out of pictures they didn’t consent to be in.

Things Generative Art Is Bad For

Did you know you don’t own copyright on generated art? This is pretty well established. If you generated the image, it’s not yours, because you didn’t make it. It was made by an algorithm, and algorithms aren’t people. This isn’t a complicated issue, this just means that straight up, any art you make at work that’s meant to be used for work, shouldn’t be used because people can just straight up use it. Logo design, branding, all that stuff is just immediately open for bootlegging or worse, impersonation.

Now you might think that’s a bit of a strange thing to bring up but remember, I’m dealing with students a lot. Students who want to position themselves as future prompt engineers or social media managers need to understand full well that whatever they make with these tools are not things that will have an enduring useful application. Maybe you can use it for a meme you post on an account, but it’s not something you can build branding off, because you don’t own it. Everyone owns it.

From that we get a secondary problem, because if you didn’t own it, its only use is what people say or think when they look at it, and thing is, people are already sick and tired of the aesthetics of generated art. You’re going to get people who don’t care glossing over it, and people who do care hating it. Generative art as a way of presenting your business or foregrounding your ‘vibes’ are going to think that your work is, primarily, ‘more AI art’ and not about what it’s trying to communicate. When the internet is already full of Slop, if you use these tools to represent your work, you are going to be turning your own work and media presence into slop.

What’s more, you need to be good at seeing mistakes if you’re using these tools. If you put some art out there that’s got an extra thumb or someone’s not holding a sword right, people will notice. That means you need to start developing the toolset above for fine-tuning and redrawing sections of artwork. Now, that’s not a bad thing! That’s a skill you can develop! But it means that the primary draw of these tools is going to be something that you then have to do your own original work over the top of.

The biggest reason though I recommend students not treat this work like it’s a simple tool for universal application is that it devalues you as a worker. If you’re trying to get hired for a job at a company and you can show them a bunch of generative art you’ve made to convince them that you’re available, all you are really telling them is that you can be replaced by a small script that someone else can make. Your prompts are not unique enough, your use of the tool not refined enough that you can’t just be replaced by anyone else who gets paid less. You are trying to sell yourself as a product to employers, and generative art replaces what you bring with what everyone brings.

They make you lazy! People include typos in the generative media because they’re not even looking at them or caring about what they say! And that brings me to the next point that there are just things these tools don’t do a good job doing, and that’s stuff I want to address next in…

Things That Are Interesting

Because the tools of generative art create a very impressive-seeming artistic output, they are doing it in a way that people want to accept. They want to accept them and that means accepting the problems, or finding a way to be okay with those problems. People who don’t care that much about typos and weird fingers and so on, because you know, it gets me a lot of what I want, but it doesn’t get me everything, and I don’t know how to get the everything.

If you generate an image and want to move something in it a little bit, your best way to do that is to edit the image directly. Telling the software to do that, again, but change this bit, this much, is in fact really hard because it doesn’t know what those parts are. It doesn’t have an idea of where they are, it’s all running on an alien understanding of nightmare horror imagery.

What that means is that people start to negotiate with themselves about what they want, getting to ‘good enough’ and learning how to negotiate with the software. My experiments with these tools led to me making a spreadsheet so I could isolate the terms I use that cause problems, and sometimes those results are very, very funny. In this, the tool teaches you how to use it (which most tools do), but the teaching results in a use that is wildly inappropriate to what the tool promises it’s for.

One of my earliest experiments was to take four passages from One Stone that described a character and just put that text straight into midjourney to see what it generated based on that plain text description. Turns out? Nothing like what I wanted. But when I treated it like say, I was searching for a set of tags on a booru system like danbooru or safebooru… then it was pretty good at that. Which is what brings me to the next stage of things, which is like…

These things were trained on porn sites right?

Like, you can take some very specific tags from some of the larger boorus and type them into these prompt sites and get a very reasonable representation of what it is you asked for, even if that term is a part of an idiolect, a term that’s specific to that one person in one space that’s become a repeated form of tag. Just type in an artist name and see if it can replicate their style and then check to see what kind of art that artist makes a lot of. This is why you can get a thing that can give you police batons and mirrored sunglasses just fine but if you ask for ‘police uniform’ you get some truly Tom of Finland kind of bulging stuff.

Conclusion

Nobody who dislikes generative art is wrong. I think there are definitely uses of it that are flat out bad, and I think it’s totally okay and even good to say so. Make fun of people who are using it, mock the shrimp jesuses, make it very clear you’re aware of what’s going on and why. There’s nothing wrong with that.

I do think that these tools are useful as toys, and I think that examining the art that they produce, and the art that the community around them are exalting and venerating tells us stuff. Of course, what they tell us is that there are a lot of people out there who really want porn, and there are just as many people who want the legitimisation of impressive seeming images that they don’t care about what those images are doing or what they’re for.

Now part of this defensiveness is also the risk of me being bitten. If I buy stock art that isn’t correctly disclosed as being generative art, then I might make and sell something using generative art and now I look like an asshole for not being properly good at detecting and hating ‘AI art,’ and when I’ve say, made a game using generative art that then is integrated into things like worldbuilding and the card faces, then it gets a lot harder to tear it out at the roots and render myself properly morally clean. I’m sure a bunch of the stock art I used before 2020 was made algorithmically, just pumped out slop that was reprocessing other formula or technical objects to fill up a free stock art site like Freepik.

Which is full of generative art now.

You won’t hurt yourself by understanding these things, and people who are using them for fun or to learn or explore are by no means doing something morally ill. There are every good reason to keep these things separated from anything that involves presenting yourself seriously, or using them to make money, though. If nothing else, people will look at you and go ‘oh, you’re one of those shrimp jesus assholes.’

Check it out on PRESS.exe to see it with images and links!

#Making #Media

144 notes · View notes

kipplekipple · 1 year ago

Text

Every skill you learn, everything you MAKE, is stairs. You learn one skill and then another and then a third and every larger skill - drawing or painting or sewing or crochet or 3D printing or writing - is made of those steps, like a stair.

Talent is being able to learn some of those steps more easily. And of course sometimes there are barriers - I'm a wheelchair user and there is no ramp to pole vaulting - but it's so important that you remember that's it STEPS.

Especially if you're neurodivergent, because a lot of us have executive dysfunction, which can really fuck up your ability to turn stairs into steps, mentally. And then when you say you can't draw, what you actually mean is you can't climb a flight of stairs in one step.

Sometimes the stairs are in the Goblin King's domain and the way you get from one step to another is a bit different - or sometimes it's just hard to make out what the next step is. But you're building! And similarly, some of your works are just steps, instead of finished stairs on their own. That's okay!!! That's when you tap their imaginary shoulders with your imaginary sword and declare them to be Just A Sketch After All. You're learning how to build the next step, even if you end up tearing it down and starting over.

And it's all like that! Every single project is a flight of stairs! When I started painting with watercolours I didn't know how to layer things or how to blend colours that were wet. But now I know I start a painting and the process of painting is ALSO steps.

It lets you be a little bit more patient with yourself. And it lets you realise that:

A) if you can't immediately skip a bunch of the steps because of talent, doesn't mean you can't build the steps one at a time, and

B) if you CAN skip a bunch of the steps, doesn't mean there's something wrong if you can't skip them all.

I have a huge number of skills. I'm not bragging - I'm autistic and ADHD and I'm like House except my puzzles are how to make stuff (and I work hard not to be an arsehole about it). People always tell me I'm talented. I'm not. Talent means very little or nothing to me, because it's not something you choose or make happen. It's like saying I have brown hair - completely neutral.

What I have instead is hyperfocus, and special interests, and the luxury of access to a community makerspace. I'm in an ideal situation to build stairs.

You can build one flight. It can be crooked, it can be a loop de loop, the stairs can be anything you need them to be. You're the only one who ever has to go up them.

But next time you look at a flight of stairs you've built, or even a stepladder (my daughter would say, "you never knew your real ladder")... And you think it's terrible... Remember that usually means you haven't finished building all the steps yet.

Be the M C Escher you need in your life.

#talent #art #disability #neurodiversity #creativity #making

136 notes · View notes