serbus

My thoughts concerning AI are diverse. There is anger, absolutely, because I have spent the last two decades learning how to be a good digital artist and AI is, quite frankly, much better at it than I could ever be... at least, in the right situation. I'll come back to that.

I'm an elder millennial, meaning that I was raised both with the culture of being accepting and loving of all people as well as a culture where computers were not widely available without great personal expense. We were poor, but my birth father never made good financial decisions. So by the time I was 11 we owned 5 computers and a 14k modem. 3 computers I built myself and one I had soldered my own graphics card for. Because in 1992 if you wanted a graphics card you had to assemble it yourself.

Meanwhile I was also an avid artist. I drew like crazy, I drew on everything, and I had a really good eye for details that most other artists in my classes missed. But I knew one thing very clearly, I was terrible at it. I saw classmates who could draw realism that blew my mind. Kids, 9 and 10 years old, making large illustrations and selling them to galleries already. This was about 5% of our class, so obviously I was in the majority of kids who just did art badly. But being stubborn, I was determined to do it well anyway.

Then came Photoshop. I distinctly remember being about 14 and my art instructor took us on a little field trip to the computer lab and showed us excitedly how to use it. The Spot Healing Brush was the most magical thing I'd seen in art up to that point. You could composite an image and print it out within minutes, saving hours of time where we would previously cut clippings from newspapers in order to get a good reference image for our paintings. It changed the game for kids like me who couldn't do traditional art at the same level as the kids with "talent" who were often also the kids with money. The backlash was equally passionate.

It was cheating and there were massive copyright concerns. Not to mention how frustrated the talented kids were that they now had to compete with the whole class. They still got A's, but everyone was getting A's and they hated feeling average. Many of them quit altogether. I should note, at this point we weren't using Photoshop for matte painting yet, just for references. Painting was still done by hand in the classroom. The wider industry moved much faster though. Adobe Flash came out and matte paintings were so much easier on the computer than painting on glass slides and we tumbled into a sort of art tool renaissance.

The issue of copyrights started being addressed politically for both art and music (a lot of people really wanted to shut down Napster). I kept painting. Oils were my favorite. And when I was 19 I was given my first tablet, an Intuos 3. Soon after I bought a Wacom Bamboo for portability. Tablets were amazing! I joined the thousands of artists who had transitioned to digital and my painting time went from 3 or 4 weeks to just a few days for a single artwork. These days, I spend an average of 50 hours on my best work and 4-6 hours on my sketch work.

Once the legal battles ended, copyrights didn't limit Photoshop usage AND the new rules gave photographers and artists much greater peace of mind. Essentially, any reference used in digital art had to be significantly altered in the final piece. If a digital work used less than 20% of a reference image then it was generally acceptable. So you could, say, take someone's face from a photograph then splice in a different person's eyes, lips, nose. Then add a couple pearls or something from of a jewelry design and reshape it to match the perspective necessary to make the composite image look good. And as long as you didn't use more than 20% of a reference and you made major changes, like painting over the image with a fancy photoshop brush for a cohesive painting, then you didn't need to worry about copyrights. More than that though, then a copyright contract needed to be made with the original photographer.

The first AI images started about 8 years later. We didn't call it AI back then. It was a simple software program that would scrape tags off of images submitted to a database of portraits of people. Then it would overlay about 20 of those images with a low opacity to build a final image. It started as a study of human beauty. Every single composite image was considered beautiful by participants, but the person in the image was not a real person. It helped researchers understand what and why we see beauty in faces.

So lets come back to the AI of today. It's trained to analyze and sort objects and art styles. It scrapes dozens of digital images off the internet to create a composite, but it also references large shapes and tries to match that shape with the composite of the art style. If you add the tag, "sheep" with the tag, "Jamie Wyeth" you'll probably get a decent rendering of a Jamie Wyeth styled sheep in the same position as a random photograph of a sheep found online. But here's the kicker, 2 tags means it's most likely only sampled from 2 images. 50% one image and 50% another. A good AI system will sample from dozens of images, but not all AIs are equal. And this is where copyrights are up in the air. We don't know which images are being sampled or how much percentage of the final artwork comes directly from those images... yet.

I suspect in the next few years, once our great leaders get involved, AI scraping behaviours will be ironed out and will absolutely protect artists as well as the people who utilize AI. There are other concerns as well. Will I be unable to find work as an artist in the future? Well the truth is, I've been unable to find work most my life. The art industry is heavily biased against women, minorities, and persons with disabilities. As much as they claim it's not, my lifelong experience is that it is. I'm not a black woman, but I have a black woman's name and since day one here on DeviantArt I often saw artists with common white names and far less skill than me getting jobs that I was applying for. It wasn't until I put an image of my face in my icon that I started getting work. Being a woman was a setback, having a minority name was a setback, and now in my late 30's I am completely disabled which has been the greatest setback so far. The game art industry in particular only hires people at full-time salary. Because they like paying people less than minimum wage for 70 hours of mandatory work. A form of modern day legal servitude. I can only work part-time, so it's easy to see that any hiring manager in the industry would consider me too much of a hassle to deal with.

One look at my gallery and you can see my work is all over the place. I'm a generalist, but I'm also very good at everything I do. Painting, though my passion, takes far longer than writing a GDD or coding a small game. So for me, AI changes everything. And it changes it even better BECAUSE I'm a skilled and capable artist. This is Photoshop all over again. With AI tools, I can be a 1 man studio which means that I can compete with the 5%. All those massive businesses that will never hire me are suddenly a little less massive. AI is another tool for me to be able to build the games I've always wanted to make. And soon, 3D AI will be the same.

But there's an art to it. Certain tools will allow you to include your own reference image in the composite. My favourite is "Dream by WOMBO". I have not yet tried DA's AI, but I do find a bit of irony that they would call it DreamUP... Ah, the potential lawsuits. XD

WOMBO's Dream has been publicly available for about a year and I think they had another year of testing before that. The more I use it, the better results I get. Not because the AI is getting "trained", but because I am understanding the AI better and how it works. There are certain things AI can't do well. Jewelry, details, consistently proportional faces... though maybe 1 in 20 renders can produce a half decent face. My process as an artist is this, I start with a copic thumbnail sketch, the same as I would with any concept image. I make a list of 12-20 different tags about this thumbnail. I scan the thumbnail and upload it to the app using the medium influence setting. I choose a particular style that I find is closest to my own style as an artist and render (I wish they would let me sample my own gallery). After maybe... 60 renders I find 2 or 3 that are closest to the final composition that I wanted. Then I take those compositions into photoshop and make a composite image. I adjust the colours, make minor changes, add missing elements, and then take it back into the AI app.

This is where things get fun. I now have a pretty strong composition, granted it's choppy and often has artifacts or weird details. But I upload that new image using the strongest influence setting. I then reduce my tags to the most essential 3 or 4 and render another 60 images to find that perfect one. At this point, the AI has sampled probably hundreds of different images combined for me to get this composite image. I worry this is very taxing on their servers, so I do pay for their service. The last AI render is based strongly on my own composition and personal input. The only step left is painting over the image with a fancy photoshop brush for a cohesive result. I then spend the next 8 hours making it mine. What used to take me 50 hours is now 10. Getting the same end result, no... a better result than if I had spent 50 hours. Because the AI catches things that I don't and every day I'm learning more from it. It's even making me a better artist. In a market that is increasingly flooded with skilled worldwide artists willing to do high detailed renderings for $15 simply because the dollar is 260x the value of the peso... Well, you can understand that being able to work faster gives me some breathing room.

Now there is a huge difference in an artist who uses AI as a tool and an artist who uses AI for an end result. I've seen some impressive images on the WOMBO gallery, but every single one has tell-tale signs that it was an AI rendering with very little human input. You can tell between the images that utilize the thumbnail method and the ones that don't. You can tell between an artists who knows the AI so well that they can get something impressive and beautiful from finding the perfect tags. There is an art to it, which is why, yes, I consider them artists. But I consider them artists in the same way that I consider a young, excited teen who loves anime and draws dozens of images of Naruto to be an artist. They are young, they are still learning, and their lack of knowledge and skill is easy to distinguish. As an elder, I'm happy to give a helping hand if they ask for it. And occasionally they teach me a thing or two, like how certain keywords do so much better than others.

What I'm most angry about in the end is not really AI, it's the industry. I hate the bias I always have to deal with and the constant competition. I hate that when I network I'm always getting propositioned for sex. I hate that my name makes business harder. Yes, initially I was terrified about AI. How could I possibly compete when I was already struggling so much? But after really thinking about it... this gives me a chance. And I'm not a bad artist these days. There's a wider picture which frees me up to make games I have never had the resources to make. As long as I take care to be sensitive about copyright issues by starting with my own work, committing to multiple renditions, and utilising a huge number of keywords to prevent over-reference to work that should absolutely be respected. I think that those who submit their artwork for AI learning should be given a royalty fee whenever it's used. That's not a thing yet on DA, so for now I'll allow specific images from my gallery to be scraped by the AI in the hopes that it will happen later. Maybe it could be another source of income for me in the future? I like that DA informs us which images were used which is a step in the right direction, so at least in that regard I consider it better than most of the AI programs out there.

AI is a tool. If you use it, you'll learn quickly how to distinguish the skilled artists. The industry will catch up too. The chaos we feel right now won't last forever. So I say, play with it and use the time we have right now to take advantage of things that will allow us to compete against monopolies. You can drop out, or you can enjoy the next renaissance. I respect and understand both positions.

Prices are for Creatures. Add $10 if you need a human character. Classic angels and demons are considered human characters. Traditional

Prismacolour Warm or Cool Gray sketch: $15

Pencil on Watercolour and White Acrylics Sketch: $35

Digital Flat full body: $75

Character with simple bg: $480

Just a Background: $165

Do's:

monsters

fantasy creatures

game related art

Don'ts: school kid characters adult content ------- GDD's: $0.03 per word GAD's: $0.05 per word (includes market research) Game Design Document, accompanying concept sketches: $10 per for grayscale, $30 per for colored. An average document for a small game generally contains 12+ gameplay concepts and 30+ grayscale asset sketches.

Project Manager hourly rate: $38 I prefer to use Perforce for Source Control because it's more reliable than Git and carries a few benefits that Git can't provide. It helps the team run smoother allowing everyone to work on the project at the same time with backups and contingencies when things go wrong. If your team is determined to use only Git, I charge an extra $10 hourly fee. I know it seems drastic, but Git requires a lot of extra tracking and backup on my end to ensure a smooth team dynamic.

Please give me the option to disable the scroll over popups. They never go away on Firefox and it's killing my browsing experience.

Also, fragments are the dumbest idea this site has had since eclipse. Stop making bad design choices! Just spend the money and hire a real interface design dev team.

CW: *Rant* The site's pretty and all, but whoever designed it had no idea about intuitive user experience. Everything is impossible to find and I'm more than frustrated every time I visit DeviantArt. If I didn't rely so heavily on the gallery feature I would never come back here. The community is non-existent, if you browse by newest there's nothing but child pornography and there's no regulation about it by the company. There is literally no way for me to search for stock anymore. Sure I can 'browse' stock, but I can't look for specific keywords, rendering it useless. Features that gamified the site are hidden and thus worthless. And the forum is still as bad as ever. What, DA? It's nearly 2021 and you still don't have a proper report feature? I still can't edit my posts? I mean, bbcode had that nearly 2 decades ago now! And the "senior member" feature is a joke. Senior members have only ever been given to the elitists who snuggle up next to the staff and not to actual members who make a difference helping the community. And now they're gonna start charging more for plus accounts claiming they will have "extra features", but not even bothering to tell us what those features will be. It'll probably be just the same features on a tiered scale. If they need to raise their prices they should just tell us the fucking truth and not be dicks about it. I get that DA need to be able to afford making the site better. But I've been a plus member for years and frankly, DeviantArt, you've been investing in all the wrong areas. The strength of this website was it's community and you destroyed it in favour of looking pretty and catering to pedophiles.

*End Rant*