What resolution do I actually need?

Take the diagonal in pixels — the square root of width squared plus height squared — and divide by the diagonal in inches. A 1920×1080 panel has a 2202-pixel diagonal, so at 24 inches it works out to 92 PPI and at 15 inches to 147 PPI. The resolution alone tells you nothing without the physical size.

There is no single number, because it depends on viewing distance. The commonly used threshold is 300 PPI for something held about a foot away, which is roughly where a person with normal vision stops resolving individual pixels. A laptop viewed from two feet needs only around 150 PPI to look equally seamless, and a television across a room needs far less again.

Not on a desktop. Past about 220 PPI the operating system has to scale the interface, and any application that handles scaling badly ends up blurry or too small to use. The practical sweet spots are around 110 PPI, where nothing needs scaling, and around 220 PPI, where a clean 2× scale factor applies. The range in between is the frustrating one.

7.2 megapixels. At the standard 300 DPI for handheld prints, an 8×10 inch print needs 2400×3000 pixels. A 16×20 poster at the same DPI would need 28.8 megapixels — but posters are viewed from further away, so 150 DPI and 7.2 megapixels is usually plenty for one.

Only for things you hold. The required resolution falls with viewing distance: 300 DPI for a photo print or brochure, 150 DPI for a poster, around 72 DPI for a trade-show banner and as little as 20 DPI for a billboard. Printing a billboard at 300 DPI would produce a file of tens of gigabytes to display detail no viewer could ever resolve.

PPI is pixels per inch and describes an image or a screen. DPI is dots per inch and describes a printing device. They are used interchangeably in casual conversation, but a printer quoting 4800 DPI is describing how finely it can place ink droplets to simulate a colour, not how much image data it can use — feeding it 4800 PPI of image achieves nothing.

You can, but you cannot create detail that was never captured. Upscaling interpolates new pixels from existing ones, and modern machine-learning upscalers do it convincingly enough for many uses, but the fine texture is invented rather than recovered. For anything examined closely, reshooting at higher resolution beats any amount of upscaling.

80 to 85 for photographs. That range typically achieves a 15:1 compression ratio with no difference visible on a screen. Going to 95 can multiply the file size several times for no perceptible gain, and dropping below 70 introduces blocking and halos around edges that no later processing can remove.

Around 2.3 MB at quality 85 for a typical photograph, roughly 3.4 MB at quality 90 and about 7 MB at quality 95. Content matters a great deal though: a picture of a clear sky may come in at half those figures while a detailed forest scene comes in above them, at exactly the same setting.

Yes, alongside it. WebP typically saves 25–35% over JPEG at matched visual quality and AVIF often more again, and both are supported by every current browser. Serve them through the picture element with a JPEG fallback so older clients still get an image, and let the browser pick.

Because JPEG is lossy and generation loss compounds. Each save re-quantises the image and discards a little more detail, permanently. Keep the master in a lossless format — the camera raw file, a PNG or a layered document — and export a fresh JPEG each time rather than editing and re-saving the same one.

Roughly the distance at which a pixel subtends less than one arcminute, the limit of normal visual acuity. In practice that works out to about 3,438 divided by the PPI, in inches — so a 92 PPI monitor stops showing pixels beyond about 37 inches, while a 460 PPI phone screen has already crossed the threshold at 7 inches.