Two Crosses
one of my favorite photos I've taken. I love posterizing photos in this specific color palette
seen from China
seen from China

seen from United Arab Emirates
seen from United States
seen from T1

seen from India
seen from United States

seen from Malaysia
seen from China
seen from Malaysia

seen from France

seen from Yemen
seen from United States

seen from United Kingdom
seen from Malaysia
seen from China
seen from United States
seen from China
seen from United States

seen from United States
Two Crosses
one of my favorite photos I've taken. I love posterizing photos in this specific color palette
Can't stop listening to coded oxygen. Can't stop making these damn edits.
platonic MadWise headcanon - inspired by the fic I've been writing.
Max and Will meet up on the rooftop of the WSQK ... It's the summer before everyone leaves for college, post-graduation. Max had peer-pressured Will to steal alcohol. They drink into the night with the rest of the party, supervised by Robin. Will's moving to Montauk soon and has to sort through keepsakes after his shift at Melvald's. He finds Jonathan's Candid photos of him and the party. The whole party gathers and reminisces under the stars atop the roof. Max and Will are the first to shed tears, bonding over their shared trauma, knowing summer's end approaches. She scribbles a fond farewell on a Polaroid, snapped the day she woke up, "Kind of seemed like a sorcerer to me." She stops, crosses it out, and leaves something a little more real, "One day, Byers - we won't have to keep running up that hill."
ive been feeling a certain way lately
wonder if anyone else gets that too
Tidbit: The “Posterization” Effect of Panels Due to the Consequences of GIF Color Quantization (and Increased Contrast (And Also The Tangential Matter of Dithering))
There’s this misconception that the color banding and patterned dithering found in panels is an entirely deliberate, calculated effect Hussie manipulated the image into looking with some specific filter, but this isn’t the case, exactly. It wasn’t so much a conscious decision he took but rather an unavoidable consequence of the medium he partook in: digital art in an age where bandwidth and storage was at a premium.
Not to delve too deeply into the history and technicalities of it, but the long and the short of it is back in the early nineties to late aughts (and even a bit further into the 10s), transferring and storing data over the web was not as fast, plentiful, and affordable as it is now. Filesize was a much more important consideration than the fidelity of an image when displaying it on the web. Especially so when you’re a hobbyist on a budget and paying for your own webhosting, or using a free service with a modest upload limit (even per file!). Besides, what good would it be to post your images online if it takes ages to load them over people's dial-up Internet? Don't even get me STARTED on the meager memory and power the average iGPU had to work with, too.
The original comic strip's resolution was a little more than halved and saved as a GIF rather than a large PNG. That's about an 82.13% reduction in filesize!
So in the early days it was very common for people to take their scans, photographs, and digital drawings and scale them down and publish them as smaller lossily compressed JPEGs or lossless GIFs, the latter of which came at the cost of color range. But it had a wider range of browser support and the feature to be used for animations compared to its successor format, PNG ("PNG's not GIF").
You'd've been hard-pressed to find Hussie use any PNGs himself then. In fact, I think literally the only times he's ever personally employed them and not delegate the artwork to a member of the art team were some of the tiny shrunken down text of a character talking far in the distance and a few select little icons.
PNGs support semi-transparency unlike GIFs, which is why Hussie used them to preserve the anti-aliasing on the text without having to add an opaque background color.
While PNGs can utilize over 16 million colors in a single image, GIFs have a hard limit of 256 colors per frame. For reference, this small image alone has 604 colors:
For those who can't do the math, 256 is a pretty damn small number.
Smaller still were the palettes in a great deal of MSPA's panels early on in its run. Amazingly, a GIF such as this only uses 7 colors (8 if you count the alpha (which it is)).
Not that they were always strictly so low; occasionally some in the later acts of Homestuck had pretty high counts. This panel uses all 256 spots available, in fact.
If he had lowered the number any smaller, the quality would have been god-awful.
To the untrained eye, these bands of color below may seem to be the result of a posterization filter (an effect that reduces smooth areas of color into fewer harsh solid regions), but it's really because the image was exported as a GIF with no dithering applied.
Dithering, to the uninitiated, is how these colors are arranged together to compensate for the paltry palette, producing illusory additional colors. There are three algorithms in Photoshop for this: Diffusion, Pattern, and Noise.
Above is the original image and below is the image reduced to a completely binary 1-bit black and white color palette, to make the effect of each dithering algorithm more obvious.
Diffusion seemingly displaces the pixels around randomly, but it uses error diffusion to calculate what color each pixel should be. In other words, math bullshit. The Floyd-Steinberg algorithm is one such implementation of it, and is usually what this type of error diffusion dithering is called in other software, or some misnomer-ed variation thereof.
The usage of Pattern may hearken back to retro video game graphics for you, as older consoles also suffered from color palette limitations. Sometimes called Ordered dithering because of the orderly patterns it produces. At least, I assumed so. Its etymological roots probably stem from more math bullshit again.
True to its name, Noise is noisy. It’s visually similar to Diffusion dithering, except much more random looking. At least, when binarized like this. Truth be told, I can’t tell the difference between the two at all when using a fuller color table on an image with a lot of detail. It was mainly intended to be used when exporting individual slices of an image that was to be “stitched” back together on a webpage, to mitigate visible seams in the dithering around the edges.
To sate your curiosity, here's how the image looks with no dithering at all:
People easily confuse an undithered gif as being the result of posterization, and you couldn't fault them for thinking so. They look almost entirely the same!
Although I was already aware of this fact when I was much younger, I'm guilty of posterizing myself while editing images back then. Figured I may as well reduce the color count beforehand to help keep the exported GIF looking as intended. I view this as a complete waste of time now, though, and amateurish. Takes away a bit of the authenticity of MSPA art, how the colors and details are so variable between panels. As for WHY they were so variable to begin with, choosing the settings to save the image as requires a judicious examination on a case-by-case basis. In other words, just playing around with the settings until it looks decent.
It's the process of striking a fine balance between an acceptable file size and a "meh, good enough" visual quality that I mentioned earlier. How many colors can you take away until it starts to look shit? Which dithering algorithm helps make it look not as shit while not totally ruining the compression efficacy?
Take, for example, this panel from Problem Sleuth. It has 16 colors, an average amount for the comic, and uses Diffusion dithering. Filesize: 34.5 KB.
Then there's this panel right afterwards. It has 8 colors (again, technically 7 + alpha channel since it's an animated gif), and uses Noise dithering this time. Filesize: 34.0 KB.
The more colors and animation frames there are, and the more complicated dithering there is, the bigger the file size is going to be. Despite the second panel having half the color count of the first, the heavily noisy dithering alone was enough to inflate the file size back up. On top of that, there's extra image information layered in for the animation, leaving only a mere 0.5 kilobyte difference between the two panels.
So why would Hussie pick the algorithm that compresses worse than the other? The answer: diffusion causes the dithering to jitter around between frames of animation. Recall its description from before, how it functions on nerd shit like math calculations. The way it calculates what each pixel's color will be is decided by the pixels' colors surrounding it, to put it simply. Any difference in the placement of pixels will cause these cascading changes in the dithering like the butterfly effect.
Diffusion dithering, 16 colors. Filesize: 25.2 KB
This isn't the case with Noise or Pattern dithering, since their algorithms use either a texture or a definite array of numbers (more boring nerd shit).
Noise dithering, 16 colors. Filesize: 31.9 KB
Pattern dithering, 16 colors. Filesize: 23.1 KB
There's a lot more I'd like to talk about, like the different color reduction algorithms, which dither algorithms generally compress better in what cases, and the upward and downward trends of each one’s use over the course of a comic, but since this isn’t a deep dive on GIF optimization, I might save that for another time. This post is already reaching further past the original scope it was meant to cover, and less than 10 images can be uploaded before hitting the limit, which is NOWHERE near enough for me. I should really reevaluate my definition of the word “tidbit”… Anyway, just know that this post suffers from sample selection bias, so while the panels above came from an early section of Problem Sleuth that generally had static panels with diffusion dithering and animated panels with noise dithering, there certainly were animated panels with diffusion later on despite the dither-jittering.
Alright, time to shotgun through the rest of this post, screw segueing. Increasing the contrast almost entirely with “Use Legacy” enabled spreads the tones of the image out evenly, causing the shadows and highlights to clip into pure black and white. The midtones become purely saturated colors. Using the Levels adjustment filter instead, moving both shadow and highlight input level sliders towards the middle also accomplishes the same thing, because, you know, linear readjustment. I'm really resisting the urge to go off on another tangent about color channels and the RGB additive color model.
Anyway, there aren't any examples in MSPA that are quite this extreme (at least in color, but I'll save that for a later post), but an image sufficiently high in contrast can be mistaken for being posterized at a glance. Hence the Guy Fieri banner. In preparation for this post, I was attempting to make a pixel-perfect recreation of that panel but hit a wall trying to figure out which and how many filters were used and what each one's settings were, so I sought the wisdom of those in the official Photoshop Discord server. The very first suggestion I got was a posterization filter, by someone who was a supposed senior professional and server moderator, no less. Fucking dipshit, there's too much detail preserved for it to be posterization. Dude totally dissed me and my efforts too, so fuck that moron. I spit on his name and curse his children, and his children's children. The philistines I have to put up with...
In the end, the bloody Guy Fieri recreation proved to be too much for me to get right. I got sort of close at times, but no cigar. These were some of the closest I could manage:
You might be left befuddled after all this, struggling to remember what the point of the blogpost even was. I had meant for it to be a clarification of GIFs and an argument against using the posterization filter, thinking it was never used in MSPA, but while gathering reference images, I found a panel from the Felt intermission that actually WAS posterized! So I’ll eat crow on this one... Whatever, it’s literally the ONE TIME ever.
I can tell it's posterization and not gif color quantization because of the pattern dithering and decently preserved details on the bomb and bull penis cane. There would have had to have been no dithering and way fewer colors than the 32, most of which were allotted to the bomb and cane. You can't really selectively choose what gets dithered or more colors like this otherwise.
Thank you for reading if you've gotten this far. That all might have been a lot to take in at once, so if you're still unclear about something, please don't hesitate to leave a question! And as always, here are the PSDs used in this post that are free to peruse.
playing around with posterization
Viktor Vektor
You can always count on him to fight for your life.
Keep reading if you want to see this in GIF form. (Best viewed in it's original scale/mobile.)