Charlie Loyd 🕳️🐋💨

@vruba.bsky.social

867 Followers 402 Following

A pixel/geography/other person in Oakland. With @rahawahaile.bsky.social. Marginally more active at @vruba@everything.happens.horse on mastodon. He/him.

Another smooth interpolation between a world map and its constituent dots arranged in a disk, this time color-coded (as described in the post).
An animation that interpolates between a population map of the world and a map in which all the points are (nearly) evenly spaced in a disk.
A bag of Laser Rice
A photo of the thing literally described in the main text of the post.
A screenshot of a Pitchfork story headed “Bob Dylan Launched a Patreon for Some Reason: It includes letters and essays that appear to have utilized AI”, by Nina Corcoran, dated March 29, 2026.
A side by side image of part of a marina in two versions. The one on the left is slightly grainy and lacks color detail. The one on the right is imperfect but noticeably better.
A side-by-side comparison of a beach, a breakwater, and colorful boats tied to the breakwater.
A screenshot of a CSV showing satellite image catalog IDs with various numbers and notes.
A screenshot showing the violet roof in Google’s imagery contrasting with an on-the-ground photo of a blue roof, more like Potato’s rendering.
A side-by-side comparison of two images of an industrial part of Durban.
A screenshot of a graph with a y axis labeled 0.3441, 0.344, 0.3442, 0.344, 0.3443, 0.344, 0.3444, 0.344.
A photograph of a small helium tank abandoned by the side of a street
A photo from the mid-’90s by Steve Mann of himself and five other MIT students(?) in early wearable computing gear, looking ridiculous but at least as if they are in on the joke
A satellite image of lightly chaparral-covered hills, well lace by roads, and a lake. The perspective is from more than 45° off nadir – very unusual for satellite imagery. A large wildfire is burning, covering nearly half the image in smoke, and we see firefighting trucks parked near the fire line as well as the helicopters.
A satellite image showing the aftermath of a wildfire in a rugged, brushy landscape. The top and left quadrants are mostly untouched as far as can be seen, but the southeast part of the image is colored like ash and charcoal. Fire roads run along several ridgelines, and on close inspection, trucks (perhaps including fire trucks) are parked on them. There is a hint of smoke in the air but no obviously active burning.
A screenshot of a folder called “difficult-colors” with 22 color swatch images, mostly fairly muted red-browns and blue-greens.
Another passage from the book, at the end of p. 149 and the beginning of page 150, describing the criminalization of burning.
The first paragraph of the conclusion on page 165, reading: “Over the past century, peasants and the state struggled over fire to assert both ideological claims of how landscapes should be managed and physical claims to resource access. The chief tool of the state in implementing its vision (rooted in an anti-fire received wisdom) was criminalization through repression and rhetoric. Yet the state was not monolithic. Some bureaucrats and politicians spoke for moderation, while rural state agents (gendarmes, Forest Service field staff, local leaders) necessarily adapted to local contexts. The peasants who relied upon fire to manage their landscape did not perceive burning as a crime, and they resisted state control. They took advantage of internal state diversity, of strategic village solidarity vis-à-vis the state, and of the biophysical character of fire itself in order to effectively resist state control. The result is an impasse; fires continue to burn illegally and inconsistencies and corruption are the norm.”
Side-by-side images of a plane flying over dense Yucatán forest. On the left is Maxar’s version, which shows a bright and rainbowed image of the plane in the lead, then a very faint and translucent but sharp image of the plane. On the right is Potato, which shows a fainter rainbowed image, then a bright and opaque off-white image of the plane, then an even fainter trailing rainbowed image.
A GitHub comment in which I outline that the model emits about 160 mg of carbon dioxide equivalent per 250 megapixel image, or less than you would emit by drinking coffee (or breathing) while running it.
Flipping through a notebook
A screenshot of the section in the Wikipedia article “Announcement of polio vaccine success” describing the national celebration. Link: https://en.wikipedia.org/wiki/Announcement_of_polio_vaccine_success#Test_results_announced
Text spam with a photo of someone wearing a shirt with nonsense writing on it, clearly made by a lousy image generation model.
Bugs Bunny wishing all Syrian people a very pleasant morning. (This is an image macro meme; see https://knowyourmeme.com/memes/i-wish-all-lesbians-a-very-pleasant-evening for details.)
A screenshot of a research vessel’s cruise report. It shows an illustration of a seamount that’s roughly ring-shaped and named Bagel Seamount.
A satellite image of a mostly bare, sere landscape. It looks hot and dusty, and the hills are low and weathered, all suggesting a desert climate. No plants other than a few stray shrubs are visible, but some faint furrows suggest seasonal farming. There are a few dry streambeds. The only structures are three poles for power lines. A road clips through the bottom left of the image, and another one runs vertically through the center. They are unpaved single-lane tracks but seem well maintained. In the center of the image, on the road, is a cluster of four or five dark dots that must be people walking.
A satellite image of a quiet scene on a coastline. Compositionally, the frame can be roughly divided into strips each a quarter the width of the image. From the left, we see a strip of thick rainforest, then relatively open brush or similar coastal plain. Along the halfway line, a beach runs vertically, with two house-sized buildings visible, each with a small clearing. The northern one has a more obvious garden; the southern one has a small boat docked at it. Then the third strip of the image is shallow seawater over submerged dunes, with occasional breaking waves; and finally, along the right edge of the image, we see relatively flat open water.
A screenshot of the first (undistorted) frame of the map, showing that Estonia is relatively dark, with Tallinn very bright; Latvia is very green, suggesting widespread Foursquare use; and Lithuania is purple, suggesting more consistently dense population and spread-out income.
An animated cartogram that starts with a world map and ends up wildly spatially distorted to show things in a roughly equal-density way in a disk. We see the expected pattern of wealth concentrated in the US and Western Europe, and in cities globally; we see that Turkey and Indonesia are relatively over-served by Fourquare; we see that Africa (with population but relatively low GDP and 4sq POI density) is literally marginalized by this visualization.
An apparently abstract image of swirling, feathery textures that fit together to roughly fill a circle. Close investigation reveals some resemblance to a world map.
Another animation of Foursquare POI (strongly correlated with population) spreading out to be more evenly spaced, making a rough cartogram. This one is most of North America and is monochrome.
Altogether very similar to the one up-thread.
An animation of a map of Europe made of points. The points are colored by country. At the beginning, it looks roughly like a standard population map, with dense cities and mostly empty water. As it plays, the points spread out from each other to fill in empty spaces and expand cities. We see that Turkey, for example, has many Foursquare POI. At the end of the animation, the cartogram has not wholly converged, so some texture (density variation) remains.
Clearly misaligned TIFFs.
A colorful, swirly, motion-blurred abstract shape in a blank gray space. (It’s a long exposure of samples from the sRGB gamut rotating in Oklab space.)
A colorful, abstract figure. (It’s a visualization of the sRGB gamut in Oklab space.)
A screenshot of Google Translate suggesting that Manambavana translates literally as Condescending
A satellite image of a rural landscape. The land is mainly covered by small, irregular fields, some divided by hedges, berms, or canals. There are also fragmented woodlands, in which scattered buildings and small roads are visible. A calm-looking, moderately turbid river runs through the frame. On a sand bank in the river we can barely see about half a dozen white rectangles just a few meters across, which may be sheets or other large cloths. Indistinct, nearly single-pixel figures are next to them, at the edge of the water.
A screenshot of a python notebook showing a result of 27.5 mm/year west and 14.4 mm/year north.
A column of three satellite images, each showing the same scene. The first two are about 10–20% cloudy. The last is cloudless except for a sort of unpleasant fill-in where the clouds in the first two overlapped. It’s also not quite as richly colored.
A map-like monochrome synthetic aperture radar image showing the trace of the hurricane as an enormous spiral on the Gulf of Mexico.
A wildly, some would say unpleasantly, colorful timeseries-based autoregressed landcover map of Amereny Massif and environs.
An image of a segment of Madagascar’s eastern coast in wildly unrealistic colors. This is a surreal map without obvious physical meaning, although some general landforms (like the water and the bare soil of the highlands) appear as abstract patterns.
A clearly fake screenshot of pansharpening podcast rankings, consisting of ten podcasts that each sound absolutely awful in a slightly different way. For example, the first is: “The Guided Superresolution Minute (With Matty F. and Matt R.). This mainstay distills the pansharpening news every Monday and Thursday. Monthly giveaways, dramatic ad reads, and the famous Checking the Answering Machine segment have won TGSM seventeen” and then fades out. The seventh is: “Pansharpening: A Podcast. In the tradition of arts and design podcasts where the narrator’s voice is so subdued that it sounds like they’re making an hour-long apology, Sackler School of Ornament professor Strobe Brzezinski presents an intimate and historically informed look” and then fades out. Overall the impression is that there is a whole niche genre of podcasts about pansharpening but they are all terrible.
A blank image.
A word cloud with the three words Mike said, and a small caption reading “Create free word clouds at wourdclod.io” at the bottom.
A screenshot of a report from the Rhodium Group that summarizes projections of which sectors might draw increasing electrical power over the coming decade. A graph shows that they believe 46% will be from transportation, 22% will be from data centers (which includes “AI” in some proportion not shown), 16% from buildings, and 15% industrial.
A sort of Venn-ish diagram with two overlapping circles, one for AI maximalists and the other for AI skeptics. These are presented as thesis and antithesis. Each contains various qualities attributed to the shallow booster or the shallow skeptic, such as “Considers bad arguments against AI maximalism equivalent to good arguments for it” and “Advocates an IP law framework so harsh and intrusive that it makes Metallica seem like Phish”. Then, off to the side, there’s a third bubble that clearly represents a self-regarding author, and lightly satirizes my own sort of annoying “enlightened centrist” views, with entries such as “Knows that AI is just fancy curve-fitting and also knows that curve-fitting is crucial to addressing certain problems”. The overall tone is a sort of overly sincere look at different views in such a way as to make everyone as angry as possible.
A photo of the same pot, now simmered down to about a quarter of the volume. The onions are now soupy and lightly browned.
A photo of a cooking pot full of chopped onions.
A satellite image showing a rural village on Java. Perhaps 300 houses and a few roadside businesses, most of them red-roofed, are surrounded by agricultural fields and fragmentary forests.
A satellite image showing a semi-covered street market with green and red foodstuffs visible in the backs of trucks and in vendors’ stalls.
A photo of somewhat amateurishly constructed papercraft cube, 5 cm (2 in) on each edge, its faces labeled with X, Y, and Z arrows.
An image from the film “American Beauty” (1999) showing a plastic bag floating on the breeze, with the sub “It’s hard to stay mad when there’s so much beauty in the world”.