Design questions, answered
155 short answers from 32 guides. Each set links to the full guide for the worked examples and the reasoning behind the answer.
Calligraphy Fonts vs Fancy Unicode Text: Choosing and Using Script Fonts
How do fancy text generators work?
Fancy text generators swap each letter for a symbol from the Unicode Mathematical Alphanumeric Symbols block, U+1D400 to U+1D7FF, which exists so mathematicians can tell a bold script A from a plain A. Because the pasted characters are different characters, the style survives anywhere text goes, unlike a real font, which only travels as plain characters.
Is fancy Unicode text bad for accessibility?
Yes, fancy Unicode text causes real problems. Screen readers either skip the symbols or read them letter by letter as "mathematical bold script capital F". Searches and hashtags do not match the plain word, digits and most punctuation have no script form, and each symbol takes two UTF-16 code units, so a 30-character bio can count as 60.
What is the difference between pointed pen and brush script fonts?
Pointed pen script fonts such as Great Vibes, Parisienne and Allura come from Copperplate and Spencerian penmanship and have very thin hairlines, so they suit large wedding names and certificates. Brush script fonts such as Kaushan Script, Pacifico and Lobster come from brush lettering and sign painting, and their heavier strokes survive small sizes and busy backgrounds better.
Why does a script font look wrong in all caps?
Connected script fonts look broken in all caps because their capitals were never designed to join. Set names and headings in title case instead. Keep script text large as well: thin hairlines vanish below about 14 pt in print and 20 px on screens, and script works for a few words rather than paragraphs.
Color Harmony Explained: Building Palettes That Work on Screen
What is a triadic colour scheme?
A triadic colour scheme adds two colours at 120° and 240° from the base hue, giving three evenly spaced hues that feel balanced and colourful. Starting from Tailwind's #3b82f6 on the RGB wheel, the triad is #3b82f6, #f63b82 and #82f63b. On the OKLCH wheel the same triad becomes #3b82f6, #e24956 and #3ba01b, all at equal perceived lightness.
What is the difference between the RGB and RYB colour wheels?
The RGB wheel is the hue circle of HSL and HSV, modelling mixed light, with red at 0°, green at 120° and blue at 240°, so red's complement is cyan. The painter's RYB wheel, as in Johannes Itten's 12-part wheel, puts red, yellow and blue at the primaries, so red's complement is green and blue's is orange, matching how paint mixes.
What is the 60-30-10 colour rule?
The 60-30-10 rule turns a set of harmonious hues into a design: use one colour for most surfaces, a second for supporting areas and the rest as small accents. A colour harmony rule such as complementary or triadic only gives you hues, not a finished design, so a split like 60-30-10 decides how much of each to use.
What is an analogous colour scheme?
An analogous colour scheme uses hues close to the base, at -60°, -30°, +30° and +60° around the colour wheel, which gives a calm, natural-looking palette. From #3b82f6 on the RGB wheel the analogous palette is #3bf6af, #3bdff6, #3b82f6, #523bf6 and #af3bf6. Harmony rules say nothing about readability, so check text contrast separately.
CSS Gradients, Box Shadows and Transforms: The Parts People Get Wrong
Why does the order of CSS transform functions matter?
The browser multiplies the CSS transform list left to right into one matrix, so the rightmost function applies first and each one to its left works in the coordinates the previous ones produced. translate(100px, 0) rotate(45deg) moves the element straight right and spins it, while rotate(45deg) translate(100px, 0) moves it about 71px right and 71px down.
What does spread do in CSS box-shadow?
In CSS box-shadow, spread grows the shadow shape when positive or shrinks it when negative, before the blur is applied. The full syntax is an optional inset keyword, then x and y offsets, blur, spread and colour. A negative spread with a large y offset, such as 0px 12px 12px -8px, puts a shadow under the bottom edge only.
How do I add multiple box shadows to one element?
Separate the shadows with a comma in a single box-shadow declaration. Realistic elevation usually stacks two: a small, darker shadow for the contact shadow and a large, faint one for ambient light. The first shadow listed is painted on top. Setting box-shadow again in another rule replaces the whole list rather than adding to it.
Why doesn't a CSS transform move neighbouring elements?
CSS transforms are visual only and do not change layout. A transformed element keeps its original box in the document flow, so its neighbours stay where they were even if it looks twice the size. Because layout is not recalculated, browsers can animate transform and opacity on the compositor, which is why those two properties are recommended for smooth animation.
Why does box-shadow on a transparent PNG icon show a rectangle?
box-shadow shadows the element's rectangular box, not the visible pixels of a transparent PNG. To make a shadow follow the shape of an icon's visible pixels, use filter: drop-shadow() instead. Also note that an ancestor with overflow: hidden clips box shadows, which is why cards in a carousel often lose their bottom shadow.
Depth of Field and Hyperfocal Distance, Worked Out
What is the depth of field of a 50mm lens at f/8?
A full-frame 50 mm lens at f/8 focused at 3 m, with a circle of confusion of 0.03 mm, is sharp from about 2.34 m to 4.19 m, a total depth of field of about 1.85 m. Only 0.66 m of that lies in front of the subject and 1.19 m behind it. The hyperfocal distance for those settings is about 10.5 m.
How does aperture change depth of field?
Stopping down to a higher f-number narrows the cone of light and shrinks every blur disc, so depth of field grows. With a 50 mm lens focused at 3 m, the sharp zone is 0.30 m deep at f/1.4, 1.85 m at f/8 and 5.00 m at f/16. Doubling the f-number roughly doubles the depth while the zone is small compared with the distance.
Why do portrait photographers focus on the nearer eye?
Portrait lenses at wide apertures have very little depth of field. An 85 mm lens at f/1.8 focused at 2 m is sharp only from 1.972 m to 2.029 m, about 5.7 cm in total. That keeps both eyes sharp when the face is square to the camera, but not when the head is turned, so photographers focus on the nearer eye.
Should I use f/22 for landscape photos?
Usually not. Small apertures such as f/22 increase diffraction blur, which softens the whole frame even as depth of field grows. On full frame, f/8 to f/11 is usually the best compromise, and on smaller sensors diffraction appears sooner. Focusing at the hyperfocal distance gets the most depth of field from a moderate aperture instead.
What is the circle of confusion in photography?
The circle of confusion, written c, is the largest blur disc that still looks sharp in the final print or screen. It is a viewing convention, not a property of the lens: the classic standard of an 8 × 10 inch print viewed from about 25 cm gives c as about the sensor diagonal divided by 1,500, or by 1,730 under the stricter Zeiss standard.
Exposure Value Explained: Stops, Sunny 16 and Equivalent Exposures
What is one stop of light in photography?
One stop is a doubling or halving of light. Going from 1/250 s to 1/125 s is one stop more light, because the shutter stays open twice as long. ISO 200 is one stop more sensitive than ISO 100, and an aperture change of one stop divides or multiplies the f-number by √2, about 1.41.
Why are f-stops numbered 1.4, 2, 2.8, 4 and 5.6?
The f-number is the focal length divided by the opening's diameter, and light goes with the area, 1/N². Doubling the light multiplies the diameter by √2, so each full stop divides N by about 1.41. That gives the series 1, 1.4, 2, 2.8, 4, 5.6, 8, 11, 16, 22, and it is why f/4 to f/8 is two stops, not one.
What does EV100 mean?
EV100 is the exposure value at ISO 100, used to describe how bright a scene is whatever ISO you shoot at. It equals log₂(N² ÷ t) − log₂(ISO ÷ 100). At ISO 400, f/8 and 1/125 s give a setting EV of 12.97 but suit a scene of EV100 10.97, two stops darker.
What are third-stop shutter speeds?
Cameras step in thirds of a stop, so between 1/125 s and 1/250 s come 1/160 s and 1/200 s. Apertures go f/5.6, f/6.3, f/7.1, f/8, and ISO goes 100, 125, 160, 200. The marked numbers are rounded, so two settings that look a third of a stop apart might really be 0.3 or 0.35 stops apart.
What exposure do I need to photograph the Milky Way?
The Milky Way core is about EV100 −11 to −9. At ISO 3200 and f/2.8, an EV of −10 needs a shutter time of about 251 seconds, far longer than stars allow before they trail on a fixed tripod. That is why night-sky images are often shot at higher ISO or wider apertures, stacked, or tracked.
Field of View and Crop Factor: What a Lens Really Sees
Is a lens's angle of view quoted horizontally or diagonally?
Lens makers almost always publish the diagonal angle of view, because it is the largest of the three and does not depend on the camera's aspect ratio. That is why a 24 mm lens is sold as an 84° lens, although on full frame it covers 73.7° horizontally and 53.1° vertically. Action cameras and webcams vary, so compare figures of the same kind.
What is the Micro Four Thirds equivalent focal length?
Micro Four Thirds has a crop factor of 2.00, so multiply the real focal length by two for the full-frame equivalent: a 25 mm lens frames like 50 mm on full frame, with the same 46.8° diagonal angle. Because the sensor is 4:3, its horizontal angle is slightly narrower, 38.2° against 39.6°, and its vertical angle wider.
What focal length do I need to fill the frame with a given width?
Use f = D × d ÷ W, where D is the distance, d is the sensor width and W is the scene width to fill. To fill a 12 m wide stage from 30 m away on full frame, f = 30 × 36 ÷ 12 = 90 mm, so an 85 mm lens leaves a small margin. On a 23.5 mm wide APS-C sensor the same shot needs about 58.8 mm.
Does crop factor change the f-number?
No. Crop factor describes framing only, so f/2.8 is f/2.8 for exposure on every sensor. A 35 mm lens on a 23.5 × 15.6 mm APS-C sensor is still a 35 mm lens with the same f-number; it simply frames like a 53.7 mm lens on full frame. Depth of field at the same framing does change with sensor size.
Does focusing closer change the field of view?
Yes, slightly. When a lens focuses closer it moves away from the sensor, the image distance becomes about f × (1 + m), where m is the magnification, and the angle of view narrows. A 50 mm lens focused at 0.45 m on full frame drops from 39.6° to about 35.5° horizontally. Beyond a few metres the effect is negligible.
HEIC Photos Explained: Why iPhone Images Will Not Open and What to Do
What is the difference between HEIF and HEIC?
HEIF (High Efficiency Image File Format) is a container standard, ISO/IEC 23008-12, published in 2015 and built on the same box structure as MP4. HEIC is what people call a HEIF file whose images are compressed with HEVC, also called H.265. The ftyp box usually carries the brand heic, and the file extension is .heic.
Why can't Chrome or Windows open HEIC files?
HEVC, the codec inside HEIC, is covered by many patents licensed through several separate pools, so many vendors decline to ship a decoder in free products. Chrome, Edge and Firefox cannot open HEIC, and Windows 10 and 11 need the HEIF Image Extensions and an HEVC codec from the Microsoft Store. Safari 17 and later, iPhone, iPad and Mac open it.
When did the iPhone start saving photos as HEIC?
Apple switched the iPhone camera to HEIC in iOS 11 in 2017. Its reason was storage: an HEVC-coded photo needs roughly half the space of a JPEG at similar visual quality. A typical iPhone HEIC photo is stored as a grid of 512 by 512 pixel tiles, each coded separately, which the decoder places side by side.
Is AVIF the same as HEIC?
Not quite. AVIF uses the same HEIF container as HEIC, but with AV1 compression inside instead of HEVC. That is why some tools detect both formats with the same code and then check the brand in the file's ftyp box. Some files downloaded as "HEIC" are actually AVIF.
Why do colours look duller after converting HEIC to JPG?
iPhones capture in the Display P3 colour space, which is wider than sRGB. A browser canvas works in sRGB, and a converter that does not embed a colour profile can make strongly saturated reds, oranges and greens look a little muted compared with the Photos app. For most snapshots the difference is hard to spot.
HEX, RGB, HSL and HSV: A Developer's Guide to Color Formats
How do HEX color codes work?
A HEX color code is RGB written in hexadecimal, with each pair of digits giving one channel from 00 (0) to FF (255). In #2563EB, 25 is 37 red, 63 is 99 green and EB is 235 blue, so it is identical to rgb(37, 99, 235). CSS also accepts three-digit shorthand, where #F80 means #FF8800.
Why does yellow look brighter than blue at the same HSL lightness?
HSL is a simple transform of RGB that ignores how human vision works, and our eyes are much more sensitive to green than blue. So hsl(60 100% 50%) yellow looks far brighter than hsl(240 100% 50%) blue, though both have 50% lightness. WCAG contrast uses relative luminance, weighting green at roughly 0.7152 and blue at 0.0722.
What is OKLCH in CSS?
OKLCH is a perceptual color space added in CSS Color Level 4 alongside oklab(), based on Björn Ottosson's Oklab model from 2020. In OKLCH, equal steps in lightness look equal, unlike HSL, so it produces perceptually even palettes. It is supported in current versions of all major browsers and is increasingly used for design tokens.
Why do image editors show HSB instead of HSL?
HSB, also called HSV, maps neatly onto the familiar square-and-slider color picker: the slider chooses hue, and the square sets saturation horizontally and value, or brightness, vertically. That is why Photoshop, Figma and many image editors show HSB. CSS does not support HSV directly, so convert to HSL, RGB or HEX for the web.
How Image to ASCII Art Conversion Works
What luminance formula is used to convert an image to ASCII art?
The Image to ASCII Art converter uses the luma weights from ITU-R BT.709: Y = 0.2126 R + 0.7152 G + 0.0722 B. Averaging red, green and blue equally gives poor results because the eye is far more sensitive to green than to blue. With these weights pure green gets a luma of 182, pure red 54 and pure blue only 18.
What is a character ramp in ASCII art?
A character ramp is a string of characters ordered from lightest, with the least ink, to darkest. The 0 to 255 brightness range is split into equal bins, one per character, so with the 10-character ramp " .:-=+*#%@" brightness 230 to 255 becomes a space and 0 to 25 becomes @. These ramps were popularised by Paul Bourke's notes.
Why does my ASCII art of a photo look muddy?
Photos rarely use the full brightness range, so a converted face often comes out as a muddy middle grey. A small contrast boost stretches values away from middle grey (128) and makes edges and features stand out. For photos, the standard ramp at 80 to 100 columns with contrast around 20 works well, on subjects with clear shapes against plain backgrounds.
Why does ASCII art look like a negative on a dark background?
ASCII art designed for white paper uses dense characters for dark areas. On a dark terminal the characters are the bright part of the picture, so the mapping must be reversed: bright areas need dense characters and dark areas need spaces. The light text on a dark background option in Image to ASCII Art does that inversion for login banners and dark-mode pages.
Why does ASCII conversion average each cell instead of sampling one pixel?
Sampling one pixel per character cell, usually the centre one, is fast but misses thin lines that fall between samples and turns fine texture into random noise. Averaging the luma of every pixel the cell covers, called area sampling, means a one-pixel line inside a cell darkens that cell slightly instead of disappearing.
How to Make a Meme: The Impact Font, Outlines and Caption Layout
Why is meme text white with a black outline?
White meme text with a black outline reads on any picture: the white fill shows on dark areas, the black stroke shows on light areas, and together they stay legible over the busy middle tones of a photo. Subtitles use the same principle. Impact's condensed letters and thick, even strokes also survive JPEG compression and downscaling to a phone feed.
Who designed the Impact font?
Impact was designed by Geoffrey Lee in 1965 for the Stephenson Blake foundry. Microsoft licensed it for its Core Fonts for the Web collection in the 1990s, and it shipped with Windows 98 and with Mac OS, which made it one of the few heavy display fonts on almost every computer and the natural choice for early meme sites.
When should I use a caption bar instead of top and bottom text?
A caption bar, a white band above the picture with black text in a plain sans-serif, suits memes where the joke is in the image's details, because the picture is not covered. It also handles longer captions, since the band grows to fit rather than shrinking the text, and it reads like a sentence, which suits observational humour.
How do I write a good meme caption?
A good meme caption is short: two to six words per line keeps the text big, because every extra word costs font size. Put the setup in the top text and the payoff at the bottom, and if there is only one line, put it at the bottom. Explain the expression or situation rather than describing what is visible, and check it at phone size.
Why does my meme font look different on a phone?
Impact is not installed on many Android phones or on iOS, so the Meme Generator falls back to a heavy font such as Arial Black or the system's bold sans-serif. The fallback is measured with the browser's actual font metrics, so the caption still fits the image, though the letter shapes differ a little from the classic look.
How to Trim a Video Without Re-encoding (and When You Must)
What is a keyframe in video?
A keyframe, or I-frame, is a complete picture stored now and then in a compressed video. Codecs such as H.264, HEVC, VP9 and AV1 describe the frames in between as changes: P-frames refer to earlier frames and B-frames to frames on both sides. A keyframe plus the frames that depend on it is called a group of pictures, or GOP.
What does -ss before -i do in ffmpeg?
Placed before -i, ffmpeg's -ss option seeks in the input, jumping straight to the closest seek point before the requested time instead of reading from the start, which makes long files quick. When transcoding, ffmpeg then decodes and discards frames up to the exact time. With -c copy it keeps them, so the clip starts at the keyframe at or before the requested start.
How often do phone videos have keyframes?
Phone and consumer cameras typically place a keyframe about every 0.5 to 2 seconds, while streaming and web encodes often use 2 seconds and professional intra-frame codecs such as ProRes store every frame as one. Screen recorders and some webcams can space keyframes several seconds apart, sometimes 10 or more. These are common ranges; inspecting the file is the only way to know.
What CRF should I use when re-encoding a trimmed video?
CRF is a constant-quality setting for H.264 where lower numbers mean higher quality, and 18 to 23 is the usual range. The Video Trimmer's frame-accurate mode uses CRF 20 with AAC audio at 192 kbps, which is visually close to the source for most footage. It is still a new lossy encode, so repeating it several times compounds the loss.
Can I trim an iPhone video without converting it?
Yes. A stream-copy trim keeps the container and codecs, so a MOV from an iPhone stays a MOV with its HEVC video, every frame untouched, and finishes in seconds. The trade-off is that the clip starts at the keyframe at or before the chosen time. For a file that plays everywhere, convert the trimmed clip to H.264 MP4 afterwards.
How to Trim Audio Cleanly: Cuts, Fades and Formats
Should I save trimmed audio as WAV or MP3?
Save as WAV if the clip will be edited again, dropped into a video editor or sent to someone who will, because WAV is uncompressed 16-bit PCM and adds no further lossy step. The cost is about 10.6 MB per minute for 44.1 kHz stereo. Saving MP3 from an MP3 source is a second generation of lossy compression that cannot be undone.
How do I make a ringtone from a song?
Select a hook of about 30 seconds, set a fade in of 0.5 s and a fade out of 2 s with the quadratic shape, and save as MP3 at 192 kbps, which is about 720 KB. For an iPhone, save as M4A instead, since iOS ringtones are AAC files with an .m4r extension and a 40-second maximum.
What is the difference between a linear and a quadratic fade?
A linear fade lowers the gain in equal steps, but hearing is roughly logarithmic, so a long linear fade-out seems to hang at nearly full volume and then drop away. A quadratic fade, gain = x², falls faster early on and eases into silence, which sounds more natural over several seconds. For a 5 ms anti-click fade the shape makes no audible difference.
What does normalising audio do?
Normalising raises the whole clip so its loudest peak sits just under full scale, at −1 dBFS in the Audio Trimmer's WAV export, without clipping. It sets the peak, not perceived loudness, so a clip with one sharp laugh may still sound quieter than another. Matching loudness between clips needs an LUFS meter in an audio editor.
What is a zero crossing cut in audio?
A zero crossing cut is an edit placed exactly where the sound wave crosses zero, so the output does not jump from silence to a high level in one sample, which is heard as a click or pop. Some editors do this automatically. The alternative, which always works, is a short fade of 5 to 10 ms at each end of the clip.
How to Watermark Photos: Placement, Opacity and What It Really Protects
What size should a watermark be on a photo?
Size a watermark as a percentage of the image so a mixed batch looks consistent. Good starting points are text at 3 to 6% of the shorter side for a corner credit, logos at 12 to 20% of the image width, and a margin of 2 to 4%. At 5%, a 4000 x 3000 photo gets 150 px text and a 1000 x 750 photo gets 38 px.
What is copyright management information under 17 U.S.C. 1202?
In the US, a name, copyright notice or other identifying information conveyed with a work, such as a watermark, counts as copyright management information under 17 U.S.C. 1202. Removing or altering it knowing this will help conceal infringement is a separate violation, with statutory damages under section 1203 of $2,500 to $25,000 per violation, on top of any infringement claim.
How do I make a tiled watermark pattern?
A tiled watermark lays out copies of the mark in rows offset by half a step, like brickwork, and rotates the whole pattern by a chosen angle such as -30°. Keep the gap between repeats at 5 to 15% of the image's shorter side so no clean region large enough to reuse remains, and the opacity at 20 to 35% so the photo stays viewable.
What colour should a watermark be?
White reads on most photos, because subjects are usually mid-tone or darker, while black or a dark grey suits product photos on white backgrounds. A thin contrast outline, dark around light text or light around dark text, keeps letters legible where the background matches the fill. In Watermark Image the stroke is one tenth of the text height, at least 1.5 px.
Should I resize photos before adding a watermark?
Yes, when posting to the web. Marking a 6000-pixel file and letting a platform shrink it to 1080 pixels also shrinks the watermark, sometimes to a smudge. Resize to the final size first, then add the mark, and check it at the size people will view it, which is often a phone screen. Keep unmarked originals somewhere safe.
JPEG, PNG, WebP, AVIF and JPEG XL: What Each Image Format Is For
Should I use PNG or JPEG?
Use PNG for screenshots, flat graphics and anything with sharp edges or text, because it is lossless and compresses flat areas and repeated patterns to almost nothing. Use JPEG for photos, where its lossy compression of 8 x 8 blocks throws away detail you will not miss. Sharp edges suffer in JPEG, which is where the ringing around text comes from.
What is the difference between AVIF and WebP?
WebP (2010) and AVIF (2019) both offer lossy and lossless modes, full alpha and animation in one format. AVIF is a still-image wrapper around the AV1 video codec and supports 10- and 12-bit colour for HDR. WebP displays in all current browsers, including Safari since version 14, while AVIF displays in Chrome, Edge, Firefox and Safari 16 and later.
How do I serve WebP or AVIF images with a fallback?
Wrap the image in a <picture> element with a <source> for AVIF and another for WebP, each with its type, followed by an <img> pointing at a JPEG. The browser takes the first source whose type it supports and falls back to the img. Keep width and height on the img so the layout does not jump while the image loads.
Does converting a JPEG to PNG improve its quality?
No. Converting a JPEG to PNG stores the lossy image losslessly, which keeps every compression artefact and makes the file bigger. The same applies to re-encoding a heavily compressed JPEG as WebP: it keeps all the existing artefacts and adds a second round of loss. Start from the original, highest-quality source whenever possible.
When is converting images to WebP not worth it?
Converting to WebP can backfire on small, already optimised PNGs of flat graphics, where WebP can come out bigger. Lossy WebP always halves colour resolution (4:2:0), which smears thin red and blue text strokes, and a heavily compressed JPEG source keeps its artefacts and gains a second round of loss. Measure the real size change on your own file.
Modular Type Scales: Picking a Ratio and Making It Fluid
Why are type scale ratios named after musical intervals?
Type scale presets borrow ratios from music, where an interval between two notes is a ratio of their frequencies. In just intonation a major third is 5 : 4, so the major third scale uses 1.25, and a perfect fourth is 4 : 3, giving 1.333. The names are a convenient vocabulary, not a claim that harmonious sounds make harmonious type.
What sizes does a perfect fourth type scale give from 16 px?
A perfect fourth scale with a ratio of 1.333 and a 16 px base gives 9, 12, 16, 21.33, 28.43, 37.9, 50.52 and 67.34 px from two steps down to five up. Five steps up reaches 67 px, compared with 49 px for a major third, which shows how quickly a larger ratio compounds.
Who popularised the modular type scale for the web?
Tim Brown popularised modular type scales for the web with his 2011 A List Apart article "More Meaningful Typography", which built scales outward from a 16 px body size, and with his Modular Scale tool. The idea is older in spirit: metal type was cast in a traditional series of sizes that grows roughly geometrically.
Is fluid typography with clamp() accessible?
Fluid typography needs testing, because zooming a desktop browser to 200% halves the viewport width in CSS pixels and shrinks the vw term. A body size of 20 px on a 1280 px window can end up only 172% larger at 200% zoom. WCAG 2.1 Success Criterion 1.4.4 Resize Text requires text to reach 200%, so test fluid scales at that zoom.
What line height should headings in a type scale use?
Line height should fall as the type scale step increases. Body text at 1.5 is comfortable, but a 49 px heading at 1.5 looks loose, so display sizes work better at roughly 1.1 to 1.2. Applying one line height to every step is a common mistake when building a modular type scale.
ND Filters for Long Exposure: Stops, ND Numbers and Density
What does an ND filter do?
A neutral density (ND) filter is grey glass or resin that cuts the light reaching the lens without changing its colour. That lets you use a slower shutter speed or a wider aperture than the scene would otherwise allow, such as thirty seconds at midday to turn a busy sea into mist, or f/1.8 in sunshine for video.
What shutter speed do I get with a 10-stop ND filter?
Multiply the metered shutter time by the filter factor, which is 1,000 for an ND1000 sold as 10 stops. A metered 1/125 s becomes 8 s, 1/60 s becomes about 17 s, 1/15 s becomes 1 minute 7 seconds, and 1/2 s becomes 8 minutes 20 seconds. Above 30 seconds most cameras need Bulb mode and a remote release.
What ND filter do I need for video?
For video, an ND filter lets the shutter stay at about half the frame interval, the 180-degree rule, such as 1/50 s at 25 fps. If the camera wants 1/400 s at f/2.8 and you need 1/50 s, you need 3 stops, an ND8. Variable ND filters are popular because the strength can be dialled while recording.
How do I focus and meter with a dark ND filter?
Compose and focus without the ND filter, then switch the lens to manual focus and meter the scene. Screw the filter on, cover the optical viewfinder eyepiece if the camera has one, and set the computed time. Meters and autofocus often struggle through a 10-stop filter because the viewfinder or screen goes almost black.
Why do ND filter photos have a colour cast?
Inexpensive strong ND filters often add a warm or magenta tint, and many pass infrared, which some sensors record as a red cast on foliage. Stacking cheap filters compounds the cast. Shooting raw and correcting white balance helps, or you can buy filters with IR coatings.
NPF Rule vs 500 Rule: Sharp Stars Without a Tracker
What is the 500 rule in astrophotography?
The 500 rule says the longest exposure in seconds before stars trail is 500 divided by the full-frame equivalent focal length. A 20 mm lens on full frame gives 25 s, and a 16 mm lens on APS-C with a 1.53 crop factor gives 20.4 s. It dates from film and normal-size prints, and photographers viewing at 100% often use 400 or 300 instead.
What shutter speed should I use for the Milky Way with a 20 mm lens?
On a 24 MP full-frame camera with a 20 mm lens at f/2.8, the NPF rule gives 13.9 s for stars on the celestial equator, while the 500 rule allows 25 s. Aiming at the Milky Way core in Sagittarius, around −29° declination, stretches the NPF time to about 15.9 s. A 45 MP body cuts it to about 11.5 s.
Who created the NPF rule?
The NPF rule is credited to Frédéric Michaud of the Société Astronomique du Havre. It adds aperture (N) and pixel pitch (P) to focal length (F), and the simplified form widely published from that work is t = (35 × N + 30 × p) ÷ f, with p in micrometres and f the actual focal length in millimetres.
How fast do stars move on a camera sensor?
A star on the celestial equator drifts across the sensor at focal length times 7.29 × 10⁻⁵ radians per second, multiplied by cos δ for other declinations. For a 20 mm lens that is 1.458 µm per second, so on a 24 MP full-frame sensor with 6 µm pixels a star crosses a quarter of a pixel each second.
How do I collect more light without star trails?
To collect more light without star trails, widen the aperture, raise ISO to around 3200 to 6400, use a wider lens, or stack frames. Averaging ten 13-second frames cuts random noise by about √10, roughly 3.3 times, close to what one 130-second exposure would collect. A star tracker removes the limit for stars but blurs the foreground.
Photo Collage Layouts: Grids, Gaps and Print Sizes That Work
What size in pixels is a 6x4 photo collage?
A 6 x 4 inch print at 300 ppi needs 1800 x 1200 pixels, which is the usual target for photo prints viewed at reading distance. In the Photo Collage Maker that means the 3:2 print shape at a width of 1800. Each photo also needs enough pixels for its own cell, or it will be enlarged and look soft.
What is the best collage layout for 5 photos?
Five photos suit the two over three feature layout, with two larger cells on top and three smaller cells below. Feature layouts give some photos more space, and the large spaces are filled first in list order, so put the strongest picture at the top of the list because viewers look at the largest image first.
How big should the gap between collage photos be?
A gap between 0.5% and 2% of the collage width usually looks right, which is 10 to 40 px on a 2048 px collage. Zero gap suits before-and-after comparisons where the halves should touch, while very wide gaps make photos read as separate items. An outer border equal to the gap gives an even frame.
When should I use a uniform grid instead of a feature layout?
Use a uniform grid, such as 2 x 2, 3 x 2 or 3 x 3, when the photos deserve equal weight, like product angles, a sequence of steps or a week of outfits. Use a feature layout, such as one large plus two, when the set has a clear hero shot, like a couple at a wedding followed by details.
Pixel Density Explained: PPI, Dot Pitch and the Retina Threshold
What is the PPI of a 27 inch 1440p monitor?
A 27 inch monitor at 2560 x 1440 has a pixel density of 108.79 PPI. Its diagonal is √(2560² + 1440²) = 2937.2 pixels, and dividing by 27 inches gives the PPI. That works out to a 0.2335 mm dot pitch, and pixels become invisible to a 20/20 eye beyond about 31.6 inches, so text looks fine but not crisp.
What is a Retina display?
A Retina display has enough pixel density that each pixel subtends less than one arcminute at the device's usual viewing distance, so a 20/20 eye cannot separate pixels. Apple introduced the name with the iPhone 4 in 2010 at 326 PPI, which puts the threshold at 10.55 inches, roughly where people hold a phone.
How many PPI does a phone or laptop need to look sharp?
Divide 3437.75 by the viewing distance in inches to get the PPI where pixels fall below one arcminute. A phone held at 12 inches needs about 286 PPI, a laptop at 20 inches about 172 PPI, and a monitor at 28 inches about 123 PPI. Beyond that, extra density mainly smooths text rendering rather than adding visible detail.
Why does my calculated phone PPI not match the manufacturer's figure?
Marketing diagonals are rounded and sometimes include rounded corners. A 2532 x 1170 phone sold as 6.1 inches works out at 457.25 PPI, but Apple quotes 460 because its footnote gives 6.06 inches measured as a plain rectangle, which gives 460.3 PPI. Expect a computed PPI within about a percent of the published one.
Placeholder Text and Images: Lorem Ipsum, Dummy Images and Their Pitfalls
Where does lorem ipsum come from?
Lorem ipsum is mangled Latin from Cicero's De finibus bonorum et malorum, written in 45 BC. The word "dolorem" was split across a line, leaving "lorem" as the first word. Richard McClintock, a Latin scholar at Hampden-Sydney College, is credited with tracing the source, and the standard passage spread through Letraset sheets in the 1960s and desktop publishing software.
How do I make a placeholder image without an external service?
Draw the placeholder on an HTML canvas in the browser, as the Placeholder Image Generator does, then download it as a file or copy it as a data URI. No server is involved, so there is no URL to break. Hotlinked services such as placehold.it and lorempixel.com went away and broke every page that pointed at them.
When should I use a data URI for an image?
A data URI carries the image inside HTML or CSS as Base64 text, which makes it about 33% larger than the file. It suits tests, Storybook stories and tiny images because it needs no extra request, no network and no CORS. It is a poor choice for large images on real pages, because it cannot be cached separately.
Should a placeholder image be PNG or JPEG?
PNG is almost always the better choice for a placeholder image. A placeholder is a flat background with a few letters, which PNG compresses well while keeping text edges sharp. JPEG is built for photographs and on flat graphics often produces a larger file with faint ringing around the letters, so pick it only when something downstream insists.
How do I stop lorem ipsum from reaching production?
Add a CI step or pre-commit hook that fails when the word "lorem" appears in templates and content files, for example with grep -ri "lorem ipsum" src/ templates/. It is crude but works. Placeholder content is supposed to be temporary but rarely looks temporary enough, so an automated check stops filler outliving the sprint it was meant for.
Print Resolution: How Many Pixels You Need at 300 DPI
What is the difference between DPI and PPI?
PPI is pixels per inch of the printed image, while DPI is the printer's dot density. An inkjet advertising 1440 or 2880 DPI places many tiny ink dots to reproduce each image pixel. Print shops and camera software usually say DPI when they mean image PPI, so a lab asking for "300 DPI files" wants 300 image pixels per printed inch.
Why is 300 DPI the standard for photo prints?
300 DPI matches the limit of a 20/20 eye, which resolves about one arcminute. The density at which one pixel subtends one arcminute is about 3438 divided by the viewing distance in inches, so 300 PPI suits a distance of 11.5 inches, roughly where people hold a photo or a magazine. Farther away, 300 PPI is more than the eye can use.
How does viewing distance affect the resolution a print needs?
The required pixel density falls quickly as viewing distance grows, following roughly 3438 divided by the distance in inches. A print viewed at 12 inches needs about 286 PPI, at 24 inches 143 PPI and at 36 inches 95.5 PPI. That is why posters seen from a metre or more commonly target 150 PPI, and billboards use a few tens of pixels per inch.
Does upscaling an image make it good enough to print at 300 DPI?
No. Resampling a 1500 pixel image to 3000 pixels satisfies a print lab's file check but adds no detail. Modern upscalers can soften the result convincingly, but they invent detail rather than recover it, so the sharpness of the print still depends on the pixels the original image captured.
How much bleed should a print file have?
Printed pieces that run to the edge are usually supplied with a bleed margin of about 3 mm (1/8 inch) per side, which is trimmed off after printing. Because of that margin, the pixels needed for an edge-to-edge print are slightly more than the trim size alone would require at the chosen DPI.
Pulling Colours and Fonts Out of an Image: What Works and What Does Not
Can you identify a font from an image?
Partly. No font name is stored in a picture, only shapes, so identifying a typeface means comparing those shapes with known fonts. The RedBison Font Finder from Image compares a selected line against 33 fonts it can render locally, 26 Google Fonts families plus common system fonts, and ranks them, so treat the answer as a shortlist to confirm by eye.
Why does a font finder suggest a free font instead of the real one?
A browser-based font finder can only compare against fonts it can render, so if an image uses a commercial typeface that is not in its list, the top result is the closest free Google Fonts family or system font. Server services such as WhatTheFont use databases of hundreds of thousands of fonts. For a mockup, a similar free font is often enough.
How can you tell Arial from Helvetica?
Arial was drawn in 1982 to match Helvetica's widths, so on short words a font matcher often ties them. The tells are the leg of the R, the spur of the G, and the ends of the t and a. When identifying a font from an image, pick a sample word that includes those distinctive letters if you can.
When does font matching from an image not work?
Font matching from an image struggles when the real font is not in the candidate list, when text is italic, condensed, outlined or letter-spaced, when the photo is in perspective, when the sample is only one or two letters, or when it is hand-drawn lettering rather than a font. The ranking can still narrow the search to a style, such as geometric sans or slab.
How does an image colour picker find the right pixel?
An image colour picker measures the box the image is drawn in, works out the scale that fits the image without distortion, and subtracts any empty bars. A 400 by 400 image in an 800 by 400 box has 200 px bars on each side, so a click 210 px from the left reads image pixel 10, not 105. Clicks on the bars are ignored.
PX vs REM vs EM: Choosing CSS Units for Accessible Sizing
Is a CSS pixel the same as a screen pixel?
No. A CSS pixel is an absolute unit defined as 1/96 of a CSS inch, not a device pixel. The W3C specification ties it to a reference pixel, the visual angle of one pixel on a 96 DPI screen viewed at about 28 inches. A phone with three device pixels per CSS pixel still draws a 16 px font at about the same apparent size as a desktop monitor.
Why do nested em font sizes keep getting bigger?
Em in the font-size property refers to the parent's computed font size, so nesting multiplies. A list item set to 1.2em and nested three levels deep reaches 16 x 1.2³ = 27.648 px, 73% larger than the root. The same rule written as 1.2rem gives 19.2 px at every depth, which is why font sizes are usually set in rem.
Should borders use px or rem?
Borders and hairlines should usually stay in px. A 1 px border should remain 1 px, but written as 0.0625rem it becomes 1.5 px when the root font size is 24 px and may render blurry or uneven. Box shadows and small decorative offsets that are visual detail rather than reading size are also usually left in px.
Should media queries use em or px?
Many teams use em for media query breakpoints. In media queries, em and rem resolve against the browser's initial font size, not the html rule, so a min-width of 40em means 640 px at default settings and moves when the reader enlarges text. That way a reader with large text gets the narrower layout sooner.
What is the difference between page zoom and the browser's default font size?
Page zoom scales everything on the page, px included, while the default font size setting changes only the root font size. Text set in rem or em follows the default font size, but text set in px does not move. A reader who sets 20 or 24 px and leaves zoom at 100% therefore only sees larger text where sizes use rem or em.
Raster to Vector: How Image Tracing Works and When to Redraw Instead
What is the difference between raster and vector images?
A raster image, such as a PNG, JPEG or WebP, is a grid of pixels, so enlarging it means inventing pixels and it looks soft or blocky. A vector image, such as SVG, is a list of drawing instructions that the renderer draws fresh at any size, so a vector logo stays sharp from favicon to billboard and each shape stays editable.
How does converting a PNG to SVG work?
Converting a PNG to SVG means tracing it. The tracer optionally blurs away noise, reduces the image to a small colour palette, splits each colour into its own layer, follows the edges of every area into outlines, fits those outlines with straight lines and Bezier curves, and writes each outline as an SVG path with a fill colour.
How many colours should I use when tracing a logo?
Use the fewest colours that still look right. A logo with two brand colours plus white should be traced with three colours, not sixteen. Palette size is the biggest single control over a trace, and extra colours add noise and file size. The Image to SVG Converter's Logo preset uses 4 colours, high detail and no smoothing.
What does Inkscape Trace Bitmap use to trace images?
Inkscape's Trace Bitmap uses Potrace, the best-known monochrome tracer, written by Peter Selinger in 2001. Potrace follows the same idea as colour tracers such as imagetracerjs, but for black and white images and with a more sophisticated curve optimiser, which is why desktop tracing gives results similar to a good browser tool.
How large should an image be before converting it to vector?
Start from the largest, cleanest source you can find, because tracing cannot recover detail that the pixels do not hold. For a simple logo, a source of 500 pixels or more across is ideal. A 64 pixel icon has too little information to trace well, and heavily compressed sources trace badly too.
Resize, Convert, Compress: A Practical Order for Web Images
Should I resize an image before or after compressing it?
Resize first, then compress. File size tracks pixel count, and pixel count goes with the square of the dimensions, so no quality slider comes close to the saving from resizing. Resize to the largest size the image will be shown at, pick the format, then compress at a quality you have checked by eye.
Should I save a screenshot as PNG or JPEG?
Save screenshots as PNG or lossless WebP. A screenshot of text saved as JPEG gets fuzzy halos around the letters and may not even be smaller. JPEG suits photos, while PNG is lossless, supports transparency and is the better fit for screenshots, UI, diagrams and logos.
How much bigger does Base64 make an image?
Base64 makes an image about 33% bigger, because every 3 bytes become 4 characters. That overhead is fine for a 500-byte icon but wasteful for a 200 KB photo, which also loses its own cache entry once inlined as a data URI. Vite, for example, inlines assets under 4 KB by default and links everything else.
Does converting a JPEG to PNG improve its quality?
No. Converting a JPEG to PNG never restores quality: it stores the existing JPEG artefacts losslessly, and the file grows. Conversion only changes the container and, for lossy targets, loses a little more. Keep a master copy of the original and always work from it.
What resolution does a photo need to print sharply from a PDF?
Print sharpness depends on how many image pixels land on each inch of paper, and a print shop expects about 300. A 4000 × 3000 photo across an A4 page with 20-point margins gives about 519 pixels per inch, while a 640 × 480 screenshot on the same page gives about 83, which looks soft.
Rotate, Flip and Straighten Photos Without Surprises
What is the Exif orientation tag?
The Exif Orientation tag, number 274 or 0x0112, tells a viewer how to turn a photo's stored pixels for display. Cameras keep the pixels in the sensor's landscape orientation and write the tag instead of rearranging them. It has eight values, and 6 (rotate 90° clockwise) and 8 (rotate 270° clockwise) are the common ones for portrait phone photos.
Why is the text in my selfie backwards?
Front-facing phone cameras and many webcams mirror the preview, and some save the mirrored version, so a T-shirt slogan or a whiteboard comes out reversed. The fix is a horizontal flip, not a rotation, because only a single flip produces or undoes a mirror image. A 180° rotation equals both flips together and never reverses text.
What does crop to fill mean when rotating an image?
Crop to fill keeps the largest upright rectangle that lies entirely inside the rotated photo, so no empty corners appear. A 400 x 300 image rotated 10° crops to 363 x 240, and at 30° to 300 x 173, so large angles leave little of the frame. For a square at 45°, the result is the side divided by √2, about 70.7% of the width.
Does rotating a photo remove its Exif metadata?
In the Rotate and Flip Image tool, yes. The rotated file is created from pixels on a canvas, so it does not carry the original Exif block: no GPS position, camera serial or timestamps, and no orientation tag. The saved file contains upright pixels, so every program shows it the same way.
What does keep size mean when rotating an image?
Keep size leaves the canvas at the original dimensions and centres the rotated photo in it, so the photo's corners are cut off and small empty triangles appear at the edges. It suits a set of images that must all keep the same dimensions, unlike show whole image, which grows the canvas, or crop to fill, which trims the frame.
Sketching and Diagramming in the Browser: Canvas, Excalidraw and When to Use Each
Should I export diagrams as PNG or SVG?
Export diagrams as SVG for documentation, because SVG is vector, stays sharp at any zoom and is usually smaller than PNG for line diagrams. PNG is a bitmap that blurs when scaled up but is universally supported, so it suits screenshots, chat and slides. JPEG makes line art and text smudgy, so use it only when a file size limit forces you to.
Why can't I edit a drawing after exporting it as a PNG?
Scene-graph drawing tools store a list of shape objects and redraw the canvas from that list, which is what makes each shape selectable. Exporting a PNG flattens that list into pixels, so the image is no longer editable. Keep an editable copy as well, such as an .excalidraw file or an SVG exported with the scene embedded.
How do I make arrows stay attached to shapes in Excalidraw?
Start and end the arrow on the edge of a shape so both ends bind to it. Bound arrows follow their shapes when you move a box, which is what makes reorganising an Excalidraw diagram quick. A loose arrow that does not start or end on a shape's edge stays behind when the box moves.
What is an .excalidraw file?
An .excalidraw file is the editable save format of Excalidraw. It is JSON: an array of elements with ids, types, coordinates, styles and, for arrows, the ids of the shapes they bind to. Because it is text, it diffs reasonably well in Git, and it keeps a diagram safe when the copy in the browser's localStorage is cleared or replaced.
How do I annotate a screenshot for a bug report?
Upload the screenshot to a browser sketchpad such as the RedBison Drawing Tool, add a red arrow pointing at the problem and a text label describing it, then download a PNG or copy it to the clipboard. The Drawing Tool exports PNG at 1600 by 1200 pixels, twice its 800 by 600 canvas, ready to paste into a ticket.
The Golden Ratio in Design: The Maths, the Myths and Real Uses
Who discovered the golden ratio?
Euclid gave the construction of the golden ratio around 300 BC in the Elements, Book VI, Definition 3, under the name "extreme and mean ratio". The names "golden section" and "golden ratio" are much later, from the nineteenth century, and the symbol φ for 1.6180339887 was adopted in the early twentieth century.
Is the nautilus shell a golden spiral?
No. The chambered nautilus is a genuine logarithmic spiral, but not a golden one. When Clement Falbo measured shells in the California Academy of Sciences collection, their spirals fitted rectangles with ratios around 1.33, ranging from about 1.24 to 1.43, not 1.618. A nautilus spiral roughly triples in radius per turn, against 6.85 for a golden spiral.
How do I construct a golden rectangle with a compass?
Draw a square, mark the midpoint of its base, and swing an arc centred on that midpoint through the top corner on the far side, down to the base line extended. The radius is 1.118 sides, so the arc lands 0.5 + 1.118 = 1.618 sides from the start of the base, which is φ times the square's side and gives the long edge of the golden rectangle.
Is a credit card a golden rectangle?
Not quite. An ISO/IEC 7810 bank card is 85.60 x 53.98 mm, a ratio of 1.586 rather than the golden ratio of 1.618. Many everyday rectangles that are called golden are not, and ISO paper sizes use √2 = 1.414 instead. A 34 x 21 rectangle, built from Fibonacci numbers, comes within 0.1% of golden.
What is a golden ratio type scale?
A golden ratio type scale makes each heading φ, about 1.618, times the size below it. From a 16 px body that gives headings of 25.89, 41.89 and 67.77 px. Those steps are large enough for a poster or a landing page but usually too coarse for an interface, where gentler ratios tend to work better.
Video Bitrate and File Size: The Simple Maths Behind Storage
What is the difference between Mbps and MB per second?
Mbps means megabits per second and is used for bitrates, while MB means megabytes and is used for file sizes. There are eight bits in a byte, so a video at 8 Mbps uses 1 MB per second, not 8. Dividing by 8 twice, or not at all, when moving between the two is one of the most common estimating mistakes.
How many minutes of video fit on a 128 GB card?
Recording time is capacity × 8 ÷ total bitrate. A 128 GB card holds about 2 h 50 min at 100 Mbps and about 114 minutes at 150 Mbps. Real cards hold a little less, because formatting and camera files take space and a quoted bitrate is often a maximum rather than an average, so budget 5 to 10% headroom.
What bitrate does YouTube recommend for 1080p video?
YouTube Help's recommended upload settings for SDR video are 8 Mbps for 1080p at 24 to 30 fps and 12 Mbps for 1080p at 48 to 60 fps. At 8 Mbps an hour of video is about 3.6 GB. For 4K the upper recommended values are 45 Mbps at 24 to 30 fps and 68 Mbps at 48 to 60 fps.
What is bits per pixel in video encoding?
Bits per pixel is a rough quality yardstick: bitrate ÷ (width × height × frame rate). 1080p at 30 fps and 8 Mbps works out to 0.13 bits per pixel, 4K30 at 45 Mbps to 0.18, and 1080p60 at 12 Mbps to 0.10. The number is only comparable within one codec, but it quickly shows whether a setting is generous or starved.
Why is my video file bigger than the calculated size?
Real video files differ from the bitrate formula because most codecs use variable bitrate, containers add overhead, and files can carry extra streams. MP4 and MOV overhead is usually well under 1%, but transport streams (.ts, .mts) often carry a few percent. If a file is much larger than predicted, check for uncompressed PCM audio or a different bitrate mode.
WCAG Color Contrast: How the Ratio Works and How to Pass AA
What contrast ratio is needed for WCAG AAA?
WCAG success criterion 1.4.6, Contrast (Enhanced), requires 7:1 for normal text and 4.5:1 for large text to reach level AAA. The 7:1 figure is meant to compensate for roughly 20/80 vision. AAA is a stretch goal: WCAG itself says it is not recommended as a blanket requirement for whole sites, and laws and procurement rules point at level AA.
What contrast do input borders and icons need under WCAG?
Under success criterion 1.4.11, Non-text Contrast, user interface parts and graphics such as input borders, focus rings, toggle tracks and chart lines need at least 3:1 against their surroundings. A pale #e5e7eb input border on white is only about 1.2:1, so the edge of the field is effectively invisible to many users.
What is relative luminance in WCAG?
Relative luminance is a colour's perceived brightness on a scale from 0 for black to 1 for white. WCAG computes it by linearising each sRGB channel and weighting them 0.2126 for red, 0.7152 for green and 0.0722 for blue, because eyes are most sensitive to green. That is why pure green on black reaches 15.3:1 while pure blue on black manages only 2.44:1.
What is the lightest grey text that passes WCAG AA on white?
#767676 is the lightest neutral grey that passes WCAG AA for normal text on a white background, at a contrast ratio of 4.54:1. One step lighter, #777777, falls just below the 4.5:1 threshold and fails, even though nobody can see the difference between the two greys. WCAG thresholds are exact, so ratios must not be rounded up.
Why is the WCAG minimum contrast 4.5:1?
The 4.5:1 figure for WCAG AA was chosen to compensate for the contrast loss that comes with roughly 20/40 vision, the acuity typical of people around age 80. The 7:1 AAA level does the same for about 20/80 vision. The thresholds, unchanged since WCAG 2.0 in December 2008, estimate what people with common uncorrected vision loss need to read comfortably.