A white android squirrel creates birds with a stylus in front of a mountaneous town with ancient Chinese architecture as an anthropomorphic rabbit and fox visit a temple to her left.

Magic Stylus


Tyson Tan. Motherfucking Tyson Tan. The darling of the free software darlings. A well-deserved reputation, no doubt, which is especially tough given his universal popularity. First he makes the KDE mascot, and then the Krita mascot. What's next? Rice Krispies? The KKK?

It's hard to sum up my feelings towards Tyson Tan, beyond having an incredibly fun name to say, but if I was given ten million dollars to design my new tech startup, my team of artists would be Tyson Tan, Pan-Pizza, Shane Frost, the guy who made Helvetica, the guy who does Seth Godin's books, and the Black Eyed Peas. You probably don't know any of them (who the fuck are the Black Eyed Peas?), so it's fortunate that I'm talking about Tyson Tan, the man who can slam jam.

It's rare for an artist like Tyson Tan to appear to have done absolutely nothing wrong (except for being a filthy android fucker), a reputation as spotless as the Black Eyed Peas. Not only is he an exceptional artist, which probably amounts to his ludicrous two million page views on DeviantART, making him the Black Eyed Peas of furries (but still has nothing on the guy who worked on Banjo-Kazooie who draws tits on squirrels with NINETEEN MILLION VIEWS), he also is a honoured contributor to the free software movement, as well as being unfailingly polite. This must come from his culture of...

"People's Republic of China."

Oh no. I'm so sorry.

Wait, how the fuck does he know Japa-

Okay, Tyson Tan, the man with a plan. To be honest, it wasn't a matter of if I would feature his work, but which one to feature. All of his work is like going to a country that nobody has discovered, and yet more technologically advanced than any in the world, still maintaining its natural beauty due to its obscurity. It's sexy without turning anyone into sex objects. It's novel without being alienating. It's furry while casting off that dirty-furry reputation. Make no mistake: his work is instantly recognisable, and though one could dedicate an entire book of criticism as to why that is, sometimes you just have to see it to believe it. Like the Black Eyed Peas, only good.

Sometimes I wonder what we did to deserve Tyson Tan, being one of the most simultaneously unique, sincere, and professional artists alive. He breaths the Degenerate spirit - and though I have not offered him a Degenerate Apprenticeship yet (does the bloke even need one?), he is certainly on the shortlist. All could stand to learn something from him. All should learn from him.

...which brings me to GRADIENTS.

The cost of making work with such detail is many-fold, which doesn't even include an asshole like me violating the Sharealike license (whoops!). While they make for excellent compression testing material given their wide variety of colours, themes, shades, and setpieces, all of that also means it depends heaviy on having a high resolution and a huge colour palette. And you know what that means? Compression Hell!

It's a miracle how modern technology has advanced to the point where anybody with a good eye and a copy of GIMP can make what I'm making right now, where algorithms are extraordinary things and all that matters is knowing how to exploit them. It's less a skill-based discipline and instead an intelligence-based one, because while it does take some degree of skill to alter images to look good, conform with the artist's intentions, and maintain the perfect balance between filesize and quality, it's really a learning process to be able to do all of that.

I spent the majority of last night testing out compression schemes on this image seeing the different styles and sizes produced by altering the image in many different ways. Okay, first we index the image and then apply the resize. And then we resize first, and index it after. Riveting stuff, I tell you. But the fact is that these minor changes in order of operations affect the output of the final piece drastically.

Although indexing and resizing should be theoretically perfect because we're only throwing away information and not creating new information out of thin air, we have to be practical. If such perfection existed, I wouldn't be here. I guess the reason so few people aren't into the discipline is because compression, for 99% of people, is good enough. Nobody is clamouring to adopt WebP because JPEG and PNG are already really good for their respective purposes of realistic and lossless encoding. GIF is a piece of shit, though, which makes me sad APNG never caught on because Google is afraid it will sully their Chrome crapware.

Technology never becomes perfect - the general public only cares about what works well, which is why you probably use incandescent lightbulbs instead of LEDs. LED lightbulbs are better in every possible way; lifespan, brightness, efficiency, durability, and sometimes you can even change its colour. But the modest price increase for LEDs seem excessive when their dollar store lightbulbs are doing just fine. It's hard to care about things like "energy efficiency" when it doesn't affect you personally.

As a result, we see excellent technological advances, like the Vorbis codec (known as .OGG), dwindle in obscurity because the general public understands that MP3 is "good enough". Disregarding the historical and modern-day practices of adoption which made MP3 a really big deal, the plain fact is that Vorbis is just plain better. It's like the LED lightbulbs of audio schemes, where MP3 is the Black Eyed Peas. But nobody is clamouring for Vorbis because nobody cares enough about the upgrade. Am I saying that people will consume any old crap for no real reason? Yes, that is indeed what I'm saying. For proof of this, go on the YouTube front page.

So the original image was huge here, a literal wallpaper's worth of bytes, the 3MB filesize staring at me, threatening to eat up even more of my dwindling bandwidth, with not a single regard for compression at all. But who could blame it? Today's compression algorithms just aren't prepared for the sheer depth-of-technique that exists in today's artwork, with all its alpha channels and gradients and billions of colours and what have you. They're stuck in the past, dad, and the only way to take the art of the future and shove it into the past, is to apply the tried-and-true techniques to it until it breaks.

Squeezing in this image to fit just under the 10kB limit required extensive knowledge of how to manipulate the algorithms, and a lot of testing. The end result is a lightly recoloured, cropped, and resized imaged indexed down to sixteen colours. The miracle of this is not that it's possible; it's easy to take any random garbage and fit it under an arbitrary limit. The miracle is that it looks somewhat decent while doing so.

Tyson Tan? More like Tyson Tan... out of Tan.

You know, that cost me brain cells to write.

Date: 2017–02–09. Size: 9,893 bytes. Colours: 16.

Upscaled Dimensions: 1000×500. Original Dimensions: 250×125.