My stance on AI

I wrote this statement for two reasons:

1.  People occasionally ask me if I use AI. It’s obviously not something I hide, but some people are all too eager to dismiss my work as lazy or without value because of it. If you want to know if I use AI, you’re gonna get the whole answer. Not just the simplistic view.

2. The debate lacks nuance. It isn’t as simple as ”AI is bad”. AI is a tool. It is what you make of it. I believe I make a unique and high quality product and AI is only one of the tools I use and not even the most important one.

Who am I?

I’ve been a creative professional for nearly 20 years and am currently working as a designer for a triple A game company. Alfies is a side business or hobby. Almost certainly many of the images I’ve made in my career has been used to train todays AI engines. 

My position on AI:

There’s a whole lot of lazy AI images out there, most of it is boring and polluting. But my tokens are neither lazy nor “just AI”. I use AI as a tool, not dictation. If you have a moral stance against AI, I respect your opinion, my tokens are clearly not for you. But dismissing my work as “just AI” is reductive and incorrect.

AI images are cheap. For Alfies to have a reason to exist, I have to add value in other ways than just typing a prompt into an AI. That's why I put a ton of effort into the other areas of design, areas that are deeply underexplored by other token makers. And this is why my tokens are different and has (in my opinion) a unique sense of world building and concept.

My Process (including what role AI plays in it):

1. Concept / Idea (Tool: The notes app in my phone)

Behind every collection lies a concept. It may be an idea like "What if Magic was invented by someone else, somewhere around 1970-80?" That's the main idea behind Alfie's Adventure. Or "What if there was a Wrestling Federation in the Multiverse" or "What would a cat & dog show look like in the multiverse?" or “What if there was a Supermarket for planeswalker?”. These are ideas that you can find as collections on the store.

This idea guides every part of the token: The frame, the text, the image. Concepts are hard and behind every collection in my store there's a trail of dead concepts that didn't work out.

2. Research (Tools: Google, books, pinterest, visual archives etc)

To answer the questions within any concept, like "How would a collectible card game look 40 years ago?" I use a lot of time to find resources and references of colors, typography, language and general aesthetics that the collection can be based on. This informs the execution later down the road.

3. Art Direction (Tools: Illustrator, Photoshop)

This is where I spend the most time. Where most other token makers simply have “this is my style” as their art direction, I put a lot of effort into how to visually build worlds that can represent the concepts I have come up with. The Supermarket collection is based on references of how grocery shop posters used to look - colored broad markers on white paper. The Street Clans art direction is based on Y2k aesthetics, modern anime and cyberpunk. This art direction becomes the foundation of everything in the collection and is the reason every collection has their own visual identity. I am driven by ideas and concepts. This is all manual work, no AI in sight.

4. Frame, icons & general layout (Tools: Photoshop & Illustrator)

Most approach this part as a frame just being something on top of the image to put text on. As if the frame is an afterthought. I feel the opposite. Making a good frame and layout is very hard and is an area where I can really differentiate my tokens from my competitors. For reference I spent several months on the Adventure frame. No less than four times did I decide “This is it. I have it” before I decided to trash it and start again. This is all manual work, no AI in sight.

5. Central image (Tools: Midjourney & Photoshop)

In this step I have a good idea of what I'm looking for in images and I'll have a sketch of a layout and frame ready. In the centre of the frame I’ve usually made space for the image. Here I generate images in Midjourney, which is an AI. Eventually I find a setting that matches my vision and I pick my out favourite generations. Finding a good setting and/or prompt for a new concept can take weeks, often I never find it at all. When I generate images for new tokens in an established concept it takes a lot less time, generally between 1-4 hours, to find a good candidate.

Why does it take that long? I’m just writing text right? Because most AI generations looks like crap and I don’t want crap. Making something that fits my vision and demands requires effort.

6. Edit & cleanup of image (Tool: Photoshop)

At this point I'll test images in my layout. I'll edit flaws and unnecessary details or parts of the image. AI generates a lot of artifacts, phantom limbs or muddy/unclear details, those need to be removed or redone. Sometimes I paint over those parts myself, sometimes I edit together several different images. There's very rarely an image coming out of Midjourney that doesn't need one or more editing passes. Sometimes I miss details, I'm only one guy after all.

7. Polish (Tools: All of the above)

I now probably have a fairly decent token. But it'll still need polish to become it's best self. I'll make revisions on both image and layouts back and forth until I'm fairly convinced it's the best it can be.

8. Print (Photoshop and InDesign)

Now the image needs to be processed for print. This means adjusting colors and sizes so that it can be printed without being offset, cut, too dark, too bright or have any other technical flaws. I pick a paper stock that goes with the concept. For example for Alfie's Adventure I found a paper stock that had a rough and vintage look and feel to it that perfectly matches the concept of a TCG from the 80's.

AI FAQ

- Why do I use generative AI?
To me, this isn't strange. I've been a designer for decades and have always sourced images from other places. From stock sites, from colleagues etc. I'm not an illustrator or a fine artist, I do the other stuff, like layouts, typography, concept, art direction. The idea of a token just being an image with a hastily created frame on top doesn’t apply to me. I use my energy on the work around the image and as a result my tokens are different.

- Why don't I just learn how to draw?
I do what I love, not what other people say I should love.

- AI is unethical and bad
This is a viewpoint shared by many and if that's the case for you I respect your opinion and urge you to move on to another seller. I personally don't share that viewpoint, even though images I have created in my career almost certainly have been used to train the AI we see today. Photography happened, the computer happened, now AI has happened. This is what technical evolution does. Artists will be as important in the future as they are today, but their everyday duties may change. My job will likely also change, but I'm getting ready for it.

- Why don't you say upfront that your tokens are made using AI?
If you have a moral stance against AI, you’re not my target audience. If you’re not my target audience it doesn’t make sense to form my messaging for you. I’m not trying to reach you, I’m trying to reach those who appreciate my work.

Thank you for reading and have a nice day!