Reaction Stickers: your character, ten reactions

A wave, a laugh, a very sleepy face. Reaction Stickers turns a character reference into a sheet of ten expressions using the image-generation tool in your own chat app.

Try the demo. It includes a downloadable robot reference and a finished example, so there is something to start with even if you do not have a character ready.

The small version

Choose the everyday reactions or write ten of your own. Keep the large labels or turn them off. Copy the prompt, attach your reference in ChatGPT, and generate the sheet there.

The website prepares text. It does not receive your images or run an image model. Generation uses your own account’s available tools and limits.

Keep it

The demo also offers a downloadable skill and plugin package. Install the skill where your account supports it, or use the standalone prompt without installation. Setup instructions and the full source are linked from the project page.

The default result is one JPG preview in a two-column, five-row layout: hello, love, laugh, thanks, yes, no, sad, angry, thinking, and sleepy. Review it and ask for specific repairs before using it.

Individual transparent stickers and publishing a Telegram pack are separate steps. This first release keeps the workflow small: bring a character, make expressions, and keep the preview.

This essay is also part of The Singularity Log.

When Art Gets Cheap Enough to Waste

Disposable art was the point.

I have about ten people in my head who need Telegram stickers. This is not a sentence I expected to have an economic theory attached to it, but here we are.

Let’s say everybody gets ten stickers: happy, angry, sleepy, nope, hug, laughing, screaming, whatever particular emotional vocabulary makes them feel like themselves. That is one hundred pieces of artwork. At twenty dollars a piece, which is already bargain-basement pricing for custom character art, I am looking at two thousand dollars for Telegram stickers.

And that’s before anybody changes their outfit, before I want a picture of two people together, before I decide somebody’s ears are wrong, or before somebody wants a sticker involving a very specific inside joke that will be funny to approximately four people for the next six weeks. When you are a plural system, art becomes an n-squared problem. Ten people do not produce ten relationships. They produce forty-five unique pairs, plus trios, groups, friendships, conflicts, people who front together, and people who become different versions of themselves around one another. The visual vocabulary of a system is not a cast list. It is a graph.

I am not spending five thousand dollars illustrating that graph.

So I built a factory.

The factory

Over the last few days, I built a little system for producing Telegram sticker packs. You give it a character reference, work out their visual language, choose the reactions you actually use, block the poses, iterate until they feel right, generate the finished stickers, and package them up. The whole thing is deliberately mundane. I am not particularly interested in whether a computer can produce a technically impressive cartoon mouse looking exhausted. We crossed that bridge a while ago.

What interests me is that making the exhausted mouse is now cheap enough that I do not have to ask whether the exhausted mouse deserves to exist.

That is a much stranger change.

For most of my life, making an image came with an implicit economic question: is this idea worth making? Sometimes the cost was money. Sometimes it was time, training, equipment, or the social cost of asking an artist friend for a favor. Sometimes it was simply the distance between having an image in your head and possessing the ability to drag it into the world. Whatever form it took, scarcity put a little customs checkpoint between imagination and artifact. State the purpose of your image. Declare its value. Is this really worth doing?

That filter does not merely determine how much art gets made. It determines what kinds of art get made at all.

A hundred-dollar portrait makes sense. A hundred-dollar reaction sticker usually does not. A carefully commissioned illustration of two characters on their anniversary can justify the expense. A picture of one of them discovering that somebody drank the last of the oat milk probably cannot. The economics quietly sort ideas into categories: serious enough to become real, or too trivial to bother.

Then the economics changed.

Disposable art

A while ago, I started using the phrase disposable art for a lot of what I was making. I did not mean bad art, thoughtless art, or art without value. I meant art that was allowed to be temporary. Art you could use rather than preserve. Art you could send to somebody once, make into an icon for three months, use to settle an argument in a group chat, or create because somebody said something stupid at 1:14 in the morning and you immediately needed a picture of your fursona staring into the middle distance.

It is closer to a doodle in the margin than an oil painting, closer to a con sketch than a gallery piece, and in a lot of cases closer to speech than to an object you own.

That distinction matters because the conversation around generative images often assumes that every generated picture displaced some hypothetical commission. The question becomes, “Would you have paid an artist to make this instead?” For a lot of the images I make, the answer is simply no. Not because the artist is charging too much, and not because their labor lacks value, but because the image itself cannot support the economics of a commission.

I was never going to commission one hundred Telegram stickers. There is no missing two-thousand-dollar check floating around in the economy. There is only a folder containing one hundred images that previously would not have existed.

That is a different economic event, and I think it is the more interesting one.

Cheap tools create new behavior

When production costs collapse, we tend to imagine the old thing being made more cheaply. That certainly happens, and it creates all sorts of legitimate arguments about labor, markets, authorship, and compensation. But sufficiently cheap production does something else: it makes entirely new categories of use possible.

Photography did not merely give us less expensive painted portraits. Desktop publishing did not merely give us cheaper typesetting. Digital audio did not merely give us cheaper records. Once a medium becomes inexpensive enough to use casually, people start doing things with it that would have been economically absurd under the old cost structure.

They waste it.

They take seventeen photos of lunch. They record demos nobody will ever hear. They make memes that live for nine hours, fake movie posters for fictional sequels, birthday invitations, niche graphics for communities whose entire membership could fit inside a Cheesecake Factory, and elaborate visual jokes whose audience is four people and one confused group chat moderator.

On a spreadsheet, much of this looks like waste. Culturally, it is where an enormous amount of life happens.

Cheap tools let us spend creative effort on things that do not need to survive.

That is not a defect. It is one of abundance’s most interesting properties.

Images become language

This is the point I keep circling back to: when images become cheap enough, they stop functioning exclusively as art objects and start functioning as language.

A Telegram sticker is not usually something I contemplate. I deploy it. It is a sentence with ears. The image does not sit outside the conversation waiting to be admired; it participates in the conversation. It can answer a question, punctuate a joke, communicate affection, establish tone, or tell somebody that whatever they have just proposed is such an extraordinarily bad idea that words have temporarily failed.

Language requires abundance. Imagine charging twenty dollars every time somebody wanted to say “lol.” That is approximately the problem with trying to build conversational visual culture entirely out of bespoke art objects.

If every visual utterance has to justify the labor of a commission, visual language remains reserved for unusually important moments. Drop the marginal cost far enough and suddenly visual communication can become mundane. You can make six versions because the eyebrows are wrong. You can give different people different ways to react. You can throw something away because the joke did not land. You can experiment without first deciding whether experimentation is financially responsible.

That mundanity is the point. It means visual expression can become something you do, not merely something you purchase.

We have spent a remarkable amount of time asking whether generative media can make precious things. I am increasingly interested in what happens when it makes things that are not precious at all.

I suspect that may be the larger transformation.

Furry already understands disposable art

Furry fandom is an unusually good laboratory for this because it is both intensely visual and deeply social. It is one place where campfire logic is already visible. A character can be an avatar, a costume, an autobiography, a recurring joke, a social interface, or some glorious combination of all five that got out of hand sometime around 2007.

It is also an expensive visual culture. There are reference sheets, badges, icons, stickers, convention commissions, character redesigns, fursuits, prints, and enough other forms of personalized art to make a Visa card start sweating.

None of that means artists are charging too much. Artists should charge enough for their labor to be sustainable, and twenty dollars for a custom piece is already cheap. The problem appears when participation in a highly visual culture depends on repeated access to scarce custom labor. Stack enough individually reasonable purchases together and expressive bandwidth starts tracking disposable income.

Money becomes visible.

The person who can afford twenty commissions can show twenty different facets of their character. They can have seasonal outfits, relationship art, joke art, experimental redesigns, reaction stickers, and little moments from their imaginary life. The person who can afford one commission gets the reference sheet and learns to make it stretch.

This has always been treated as normal because, until recently, there was not much of an alternative.

Generative tools make the alternative obvious, and that is one reason the argument around them gets so hot. They interfere with a long-standing relationship between money and visual participation, a question I also explore in After Scarcity. If somebody who could never plausibly spend two thousand dollars on Telegram stickers can suddenly possess the expressive equivalent of two thousand dollars worth of Telegram stickers, then something real has happened. Not necessarily to the market value of an individual commission, but to the amount of visual self-expression available to that person.

Those are not the same question.

The n-squared problem

This is where being a system made the whole thing impossible for me to ignore.

If I were commissioning one portrait of one character, the existing model would be perfectly workable. If I wanted one special image of two people together, likewise. There are pieces I absolutely want another human being to make because another person’s interpretation is part of the value. I want to see what somebody whose work I love sees in a character I care about. That is not interchangeable with pressing a button until I get something attractive.

But my everyday visual needs are combinatorial. I do not just want ten portraits. I want the small, stupid, affectionate, fleeting moments among ten people. I want reactions. I want relationships. I want visual shorthand. I want things nobody in their right mind would commission because the idea is worth approximately forty seconds of joy.

Under the old economics, most of those images simply disappear before they are born.

Once production becomes cheap enough, they survive.

And then another hundred ideas show up behind them.

That is the part of abundance that interests me. It does not merely satisfy existing demand. It creates demand that scarcity had trained us not to notice.

So I built a sticker factory

Which brings me back to my ridiculous little machine.

I did not build it because Telegram stickers are important. I built it because they are not.

That is the experiment.

Can creative technology get cheap enough that I am willing to waste it? Can I make images too trivial to commission? Can I make something simply because it will make my wife laugh? Can ten different people have ten different ways to say “absolutely not”? Can I throw away a finished image because the expression feels a little off, without having to mourn the money I spent making it? Can visual art become casual enough to behave like conversation?

Apparently it can.

The sticker factory is not the end product of that realization. It is a tiny proof of concept. The more interesting thing is the change in the question. I used to ask whether an idea was important enough to turn into an image. Now I can ask whether the image would be useful, funny, affectionate, clarifying, delightful, or simply worth having for a minute.

That is an enormous expansion of the creative territory.

I used to think disposable art was a category of art. I think I had it backward. Disposable art is what happens when creative abundance crosses a threshold, when the cost of making an image falls below the value of the tiniest reasons we have for making one.

A joke. A mood. A memory. A flirtation. An identity. A Tuesday.

At that point, art does not become worthless. It becomes cheap enough to waste.

And once we can afford to waste it, we start discovering all the things we never thought were important enough to make.

That may have been the point all along.