Image tools

Because JPEG throws information away every single time it saves, not once when the photo was taken. Each save divides the image into blocks, converts them to frequencies and discards the ones the eye is least likely to miss, and it does that to whatever it is given, including an image that has already been through the process. Open, small edit, save, repeat, and the damage compounds visibly. The habit that avoids it: keep the original, do all your edits in one session, and save once at the end.

One changes the picture and the other changes a note attached to it. Resizing alters the number of pixels, so the file genuinely gains or loses detail. DPI is a number stored alongside the image saying how large it should be printed, and changing it alone does nothing whatsoever to the pixels. This is why "make it 300 DPI" is not an instruction a picture can obey: a 400 pixel wide image at 300 DPI is a very sharp inch and a third. What the printer needs is enough pixels, not a bigger number in the file.

From the saved file, yes. A raster crop writes out a new, smaller grid of pixels and the ones outside it are not stored anywhere, so there is nothing to recover. Two things do survive and catch people. Editors that offer a non-destructive crop keep the full image and record the crop as an instruction, so the original is still in the file until you export a flat copy. And cropping a PDF is not a crop at all: it moves the visible box and leaves everything outside it in the file.

Because your phone wrote it in. Photographs carry EXIF data: the camera, the settings, the exact time, and on almost every phone by default the GPS coordinates where the shutter was pressed. That is fine in your own library and much less fine on a listing, a forum or a marketplace, where a picture of something for sale in your front room carries the front room. Most social platforms strip it on upload. Almost nothing else does, including email, messaging apps that send as a file, and your own website.

Match the format to the content rather than to habit. Photographs want JPEG, WebP or AVIF, because they are built for continuous tone and gradients. Flat colour, screenshots of text, logos and anything with hard edges want PNG, because JPEG smears fuzzy halos around every sharp boundary and PNG stores them exactly. Anything needing transparency rules JPEG out entirely. And a logo that has to work at several sizes wants SVG, which is not an image at all but a set of drawing instructions.

Source: PNG specification, W3C

Because it is twenty five years newer and can guess better. Both work by predicting what a block of pixels will look like from its neighbours and then storing only the difference, but WebP has far more prediction modes to choose from and a more efficient way of encoding what is left. AVIF goes further again. The practical catch is not quality, it is where the file has to open: an old system, a print bureau or a client on ancient software may simply refuse it, which is why converting back to JPEG is still a tool people need.

No. Detail that was never captured is not hiding in the file waiting to be recovered, and no amount of processing invents it. What an upscaler does is guess plausible pixels between the ones you have, and the honest measure of a good one is that the guess is not distracting. Modern model-based upscalers guess very convincingly, which creates its own problem: they are inventing texture rather than restoring it, so a face may come back sharp and subtly not the same face. Fine for a background. Not fine for evidence.

Because a phone screen packs two or three physical pixels into every one the page thinks it has. A 200 pixel wide logo displayed at 200 points is being stretched across 400 or 600 real pixels, and the browser fills the gap by smearing. Two fixes. Supply a raster logo at two or three times the size it will display at. Or use SVG, which is drawing instructions rather than pixels and is therefore exactly as sharp at any size and usually smaller than the raster version too.

It depends entirely on what your machine saved it as, and the answer matters more for screenshots than for photographs. A screenshot is mostly flat colour and sharp text, which is the exact content JPEG handles worst, so a JPEG screenshot arrives with visible fuzz around every letter. Saved as PNG it is pixel-for-pixel exact. If your screenshots look soft, check what your operating system is defaulting to before blaming the screen, because most let you change it and several default to JPEG to save space.

Because a watermark is a business model rather than a technical necessity, and this site does not have that business model. An upload service pays per file for bandwidth and processing, so it needs you to either see an advert, hand over an email address, or pay, and a watermark is the lever that converts the third. Nothing here touches a server, so none of those costs exist and there is nothing to convert. The output is the output: no mark, no cap, no queue, and no account standing between you and it.