Polygons and early 3D

3D graphics build objects from polygons, usually triangles. Early hardware could draw only a limited number per frame, so designers built cars, buildings and characters from as few shapes as possible.

Sega's arcade game Virtua Racing (1992) is often cited as a landmark: a racing game built from flat-shaded polygons, running smoothly at a time when most racing games still used sprites. On home consoles, Nintendo's Star Fox (1993) used the Super FX chip in the cartridge to render simple polygon graphics on the Super Nintendo.

The push toward realism

As hardware improved, games added textures, lighting and millions of polygons. Realism became the goal for many big releases, and flat-shaded graphics looked dated.

The return of low-poly

In the 2010s, low-poly returned as a choice rather than a constraint. Independent developers found it:

• Cheap to produce: small teams can model and animate quickly.
• Timeless: stylised graphics age better than attempts at realism.
• Readable: clean shapes and flat colours make objects easy to identify.
• Light to run: fewer polygons mean better performance on modest hardware.

Illustrators and designers also adopted the look in posters, logos and animation.

Why it suits racing

In a fast game, the player must read the road instantly. Low-poly tracks have clear edges, simple surfaces and strong colour contrast. Turns, ramps and walls stand out without visual noise. Performance matters too: a high frame rate makes steering feel responsive.

Browser games and low-poly

Browser games run on a wide range of devices, from gaming PCs to school laptops. Low-poly graphics help them load quickly and run smoothly via web graphics technologies such as WebGL.

PolyTrack, a browser racing game by developer Kodub, uses this approach: clean, low-poly tracks with height changes and jumps in three themes. The PolyTrack, an independent fan site, offers a 0.5.0 player and a track directory with 3D overview previews of the 15 official courses (7 Summer, 4 Winter and 4 Desert) and 15 community tracks.

Design lessons from low-poly

• Silhouette first: a shape should be recognisable without detail.
• Limited palettes: a few colours create coherence.
• Lighting carries depth: flat shading with a clear light direction gives form.
• Function over decoration: in games, readability beats ornament.

Try making your own

Low-poly modelling is a good way to learn 3D. Free tools like Blender let beginners build simple objects and learn about vertices, edges and faces. In track-building games, the same principle applies: clear, readable sections are more fun to drive. The how-to-play guide on The PolyTrack covers the controls if you want to test drive a few layouts first.

Summary

• Low-poly began as a hardware limitation in early 1990s 3D.
• Games like Virtua Racing and Star Fox defined the look.
• Indie developers revived it for clarity, speed and style.
• It suits racing games that need instant readability.

ASCII in brief

ASCII, the American Standard Code for Information Interchange, was first published in 1963. It uses 7 bits, giving 128 codes from 0 to 127. Codes 0–31 and 127 are control characters (like newline and tab); 32–126 are printable characters, including space, digits, letters and punctuation.

Some useful landmarks:

• Space = 32
• "0" = 48, "9" = 57
• "A" = 65, "Z" = 90
• "a" = 97, "z" = 122

The 32 gap between cases

Uppercase A is 65; lowercase a is 97. The difference is exactly 32 for every letter. In binary, 32 is a single bit (00100000), so switching case means flipping one bit. This tidy design is why case conversion is so cheap in code.

Hexadecimal

Hexadecimal is base 16. It uses digits 0–9 and letters A–F for values 10–15. Each hex digit represents exactly 4 bits, so one byte (8 bits) is always two hex digits. That is why hex is popular: it maps neatly onto bytes.

Converting is straightforward:

• 65 in decimal = 4 × 16 + 1 = 0x41, so "A" is 41 in hex.
• 97 = 6 × 16 + 1 = 0x61, so "a" is 61.
• The case difference, 32, is 0x20.

Reading a hex string

Given 48 65 6C 6C 6F, convert each pair:

• 48 = 72 = H
• 65 = 101 = e
• 6C = 108 = l
• 6C = l
• 6F = 111 = o

The message is "Hello". Puzzle hunts love hiding text this way, and it often appears alongside a cipher layer.

Beyond ASCII

ASCII covers only English letters and basic symbols. Modern text uses Unicode, usually encoded as UTF-8. Helpfully, UTF-8 is designed so that the first 128 characters are identical to ASCII and take one byte each, while other characters, such as é or emoji, take two to four bytes. If a hex string contains bytes above 7F, you are probably looking at UTF-8 beyond ASCII.

Why this matters for ciphers

When you write a Caesar cipher in code, you subtract 65 from an uppercase letter's code to get a position from 0 to 25, shift it, then add 65 back. Understanding ASCII makes that arithmetic obvious.

Tools to explore

The ASCII table on caesarcipher.org lists codes in decimal, hex and binary, and the hex to ASCII converter turns hex strings into text and back. For a related encoding, the Base64 converter shows another way bytes are written as text.

Summary

• ASCII maps 128 codes to characters; A = 65, a = 97.
• Upper and lower case differ by 32, a single bit.
• Two hex digits make one byte.
• UTF-8 extends ASCII without changing its first 128 codes.

Where the term came from

Psychologist Edward Thorndike described the halo effect in 1920. He noticed that military officers rating their soldiers tended to rate them high or low across all qualities at once, as if one overall impression coloured every individual rating.

"What is beautiful is good"

In 1972, Karen Dion, Ellen Berscheid and Elaine Walster published a study titled "What Is Beautiful Is Good". Participants looked at photos of people and guessed their personalities and futures. People in more attractive photos were more often assumed to have desirable traits and better life outcomes. The study became a classic of social psychology.

What later research found

Later reviews found that the attractiveness halo is real but more limited than the slogan suggests. Attractive people are often judged as more sociable and socially skilled, but the effect is weaker for traits such as integrity or concern for others. Context, culture and the specific trait all matter.

Where it shows up

The halo effect has been studied in many settings:

• Hiring: appearance can influence impressions in interviews.
• Education: teachers' expectations can be subtly affected by appearance.
• Online profiles: photos shape first impressions before any words are read.

These are tendencies across many people, not certainties about any one interaction.

The reverse effect

A negative impression can also spread: one unfavourable feature colours other judgements. This is sometimes called the horn effect.

Why this matters for face rating tools

Because looks influence judgements, people are understandably curious about how a photo "reads". A face rating tool can satisfy that curiosity, but it also risks reinforcing the idea that a number defines a person. A responsible tool should frame its result as a description of one photo, not a verdict on someone's worth.

The Free Attractiveness Test describes its score as how one photo reads to a large group of people. Its scale runs from 60 to 99 with no failing grade, and the how it works page notes that lighting, angle and expression move the score, so three photos of the same person can score differently.

Reducing halo bias in your own decisions

• Use structured criteria when evaluating people, such as interview questions decided in advance.
• Separate first impressions from evidence. Note your impression, then check it against facts.
• Delay judgement until you have more information.
• Be aware of photos. If you review applications, consider whether photos are needed at all.

A healthy perspective

Knowing about the halo effect can be freeing. It shows that much of how appearance is judged comes from the observer's biases, not from anything fixed about the person being judged. Curiosity about a photo score is fine; letting it define self-worth is not. The test's privacy explanation also shows that the photo stays on your device.

Summary

• Thorndike named the halo effect in 1920.
• Dion, Berscheid and Walster's 1972 study linked beauty with assumed goodness.
• The effect is real but narrower than often claimed.
• Treat any face score as a description of a photo, not a person.

Before the boom

Daily puzzles are not new. Newspapers have printed crosswords for more than a century, and the New York Times crossword has been a daily ritual for generations. Sudoku spread widely in newspapers in the mid-2000s. What changed recently is the format: a free web page, one shared puzzle per day, and an easy way to share your result.

Wordle

Wordle, created by software engineer Josh Wardle, became public in October 2021. Players have six tries to guess a five-letter word, with coloured tiles showing whether each letter is correct, present elsewhere or absent. It spread rapidly in late 2021 and early 2022, helped by a share feature that produced a grid of coloured squares without revealing the answer. The New York Times bought Wordle in January 2022.

The design pattern

Many successful daily games share a few traits:

• One puzzle per day for everyone. Shared puzzles create conversation: everyone is working on the same thing.
• Limited attempts. A fixed number of guesses turns a search into a challenge.
• Colour-coded feedback. Green, yellow and grey tiles tell you what is right, close or wrong.
• Spoiler-free sharing. Results can be posted without revealing the answer.
• Short sessions. A few minutes, then you are done until tomorrow.

The "-dle" family

After Wordle, many games adapted the format to other subjects: geography, maths, films, sports and music. Some focused on guessing a word; others turned the tiles into attributes. Instead of letters, you compare features of a hidden item, such as a country's location or an artist's genre.

Spotle

Spotle applies the attribute approach to music artists. You have ten tries to identify a hidden artist. Each guess shows tiles comparing your guess with the answer on debut album year, solo or group, genre, country, gender and the game's popularity rank. Green means match, yellow means close, and grey means no match. The how-to-play guide on Spotle Answer Today explains each tile in more detail, including that the popularity rank is the game's own, not a live Spotify chart.

Why daily limits help

One puzzle per day sounds restrictive, but it is part of the appeal. It prevents burnout, keeps the puzzle a small ritual, and makes streaks meaningful. It also means everyone can discuss the same puzzle at lunch.

Time zones

A shared daily puzzle needs a shared clock. Games pick a reference time zone or use each player's local midnight. Spotle's current client checks its puzzle date against New York time, which is why the changeover may not happen at your own midnight.

Companion sites

Popular daily games often attract companion sites with hints, archives and explanations. Spotle Answer Today is an independent companion for Spotle, offering step-by-step hints, and the archive lists past answers for anyone catching up. It is not affiliated with the official game.

Summary

• Daily web puzzles grew from newspaper traditions.
• Wordle (2021) set the modern template.
• Attribute games like Spotle adapted it to new subjects.
• One shared puzzle per day is a feature, not a limitation.

The Fitzpatrick scale

The Fitzpatrick scale was developed in 1975 by Thomas B. Fitzpatrick, a dermatologist, originally to estimate how skin responds to ultraviolet light. It has six types, I to VI, defined mainly by how easily skin burns and how readily it tans:

• Type I: always burns, never tans.
• Type II: usually burns, tans minimally.
• Type III: sometimes burns, tans gradually.
• Type IV: burns minimally, tans well.
• Type V: rarely burns, tans deeply.
• Type VI: very rarely burns.

It remains widely used in dermatology, for example when discussing sun sensitivity or planning certain treatments.

Its limits

Because it is based on sun reaction rather than appearance, the Fitzpatrick scale was not designed as a visual colour scale. Critics have noted that its six types compress a wide range of deeper skin tones into the top two categories, which makes it a poor tool for describing the full diversity of skin colour.

The Monk Skin Tone Scale

The Monk Skin Tone Scale is a ten-shade scale developed by Ellis Monk, a sociologist at Harvard. Google released it publicly in 2022 under a Creative Commons licence and has used it in work on fairness in products such as image search and camera features.

Its purpose is visual: to give a range of reference tones broad enough to represent many skin colours, so that products and datasets can be checked for whether they work well across that range.

Comparing the two

Neither describes undertone. Warm, cool, neutral and olive undertones are a separate dimension.

Depth and undertone are different questions

Skin tone tools in fashion and beauty usually ask two separate questions:

1. Depth: how light or deep is the skin?
2. Undertone: which colour direction, such as golden, pink or olive, appears beneath the surface?

The skin tone test on Style Look Lab follows this split: nine questions, no photos, with depth bands of Fair, Light, Medium, Tan and Deep, and undertone candidates of Warm, Cool, Neutral and Olive. Its depth swatches follow the Monk Skin Tone Scale under its CC BY 4.0 licence.

Which one should you use?

• Sunburn risk or medical treatment: Fitzpatrick, with a dermatologist's guidance.
• Describing skin colour, choosing foundation shades or checking representation: a visual scale such as Monk, combined with an undertone assessment.

For choosing clothing colours, depth and undertone feed into broader colour analysis; the 12 color seasons guide shows how temperature, value and chroma combine.

Summary

• Fitzpatrick (1975) measures sun reaction in six types.
• Monk (released 2022) is a ten-shade visual scale.
• Neither covers undertone, which is a separate question.
• Use each for the job it was designed for.

MP4 is a container

An .mp4 file is a container: a wrapper that holds a video stream, an audio stream and some metadata. The video inside is compressed with a codec. H.264 (also called AVC) is the most widely supported; newer codecs such as H.265 (HEVC) and AV1 compress more efficiently but are not supported everywhere.

Compression throws information away

Uncompressed video is enormous. Codecs make files small by discarding detail that viewers are unlikely to notice, and by storing only the changes between frames where possible. How much they discard depends on the bitrate: the amount of data per second. Lower bitrate means smaller files and more visible artefacts, such as blockiness in dark areas or smearing during fast motion.

Why apps recompress

Messaging and social apps handle huge volumes of video. To save storage and bandwidth, and to make playback smooth on slow connections, they usually re-encode uploads to their own settings. Each time a video is re-encoded, it loses a little more detail. Sending a video that has already been compressed through another app that compresses it again is a common cause of mushy results.

Practical ways to keep quality

• Share the original file. Download the video from the tool that made it and share that, rather than a screen recording or a copy saved from another app.
• Send as a file where possible. Some messaging apps, including WhatsApp, let you send a video as a document, which avoids their usual video compression.
• Use high-quality upload settings. Some social apps have an option to upload in higher quality, sometimes only on Wi-Fi. Check the settings.
• Avoid repeated round trips. Downloading a posted video and reposting it elsewhere compounds compression.
• Keep the native size. Upscaling a 720-pixel-wide video to 1080 does not add detail and can make compression worse.

Audio matters too

Apps also compress audio. For music-driven trend videos, keep the original file's audio rather than re-recording it through a phone speaker.

Fast motion and dark scenes suffer most

Codecs struggle with lots of movement and with subtle shading in dark areas. If your video has a dim background, expect some blockiness after sharing. Brighter, evenly lit source photos tend to produce videos that survive compression better.

What Bopreel delivers

Bopreel produces trend videos as 15-second clips, vertical at 720×1280 by default, with the original song, ready to download. You can see the formats on the Hotel Lobby AI page and the What You Want Me To Do AI page, and finished videos remain available in your videos for downloading.

Summary

• MP4 is a container; H.264 is the common codec inside.
• Apps recompress uploads, and each re-encode loses detail.
• Share original files, send as documents, and use high-quality upload options.
• Bright, evenly lit content survives compression best.

The law in words

The time it takes to move to a target depends on two things: how far away it is, and how big it is. Far, small targets take longer; near, large targets are quick.

The formula

A common form, often called the Shannon formulation, is:

MT = a + b · log₂(D / W + 1)

• MT is movement time.
• D is the distance to the target's centre.
• W is the target's width.
• a and b are constants measured for a particular person and device.

The log term is called the index of difficulty, measured in bits. Doubling the distance, or halving the width, adds roughly one bit of difficulty and a fixed amount of time.

What the log means in practice

Because the relationship is logarithmic, the cost of distance grows slowly. Moving twice as far does not take twice as long. Small targets, on the other hand, get hard fast once they are tiny relative to the distance.

Speed and accuracy trade off

Fitts originally studied people tapping between two targets as quickly as possible while staying accurate. When people go faster than the law suggests, they miss more. Every aim test sits on this trade-off: you can be fast, or precise, but pushing one costs the other.

Design lessons

Interface designers use Fitts's law all the time:

• Make important buttons big. A larger W lowers difficulty.
• Put them close to where the pointer already is. A smaller D helps.
• Use screen edges and corners. The pointer stops at the edge, so edge targets behave as if they were infinitely deep. That is one reason menus are often placed at screen edges.

Training your aim

An aim trainer measures how quickly and accurately you hit a sequence of targets. Improvement usually comes from:

• Smooth, single movements rather than several small corrections.
• Consistent sensitivity settings, so your hand learns a stable mapping.
• Accuracy first. Many coaches suggest getting accurate at a comfortable pace, then gradually speeding up, rather than practising fast misses.

Measuring yourself

The aim trainer on blockblast.co asks you to hit 30 targets as fast as you can and reports your average time per target and your accuracy, which maps neatly onto Fitts's speed–accuracy trade-off. If you want to separate pure reaction from movement, compare it with the reaction time test, which measures how fast you respond with no aiming at all.

Limits of the law

Fitts's law describes one-dimensional, rapid pointing quite well. Real tasks add complications: moving targets, two-dimensional shapes, fatigue, and the time it takes to decide where to move. It is a model, not a complete description of skill.

Summary

• Movement time grows with the log of distance over width.
• Bigger and closer targets are faster to hit.
• Speed and accuracy trade off.
• Aim training improves the constants, not the shape of the law.