From pixels to compressed data
A typical JPEG encoder converts image colors into a representation that separates brightness from color information, groups pixels into blocks, transforms those blocks into frequency components, then quantizes and entropy-codes the values. Quantization discards information that is less noticeable at a chosen setting.
The discarded information cannot be recovered by saving the JPEG again at a higher quality. Each lossy re-encode can introduce additional changes, so keep a high-quality source for editing and export a delivery JPEG only when needed.
Recognizing compression artifacts
Strong compression can produce block boundaries, ringing around high-contrast edges, smeared texture or color bleeding. Smooth gradients may show banding, especially after repeated edits or aggressive settings.
The visible result depends on content, viewing size and display. A useful practical test is to inspect detailed areas at actual pixel size, then judge the complete image at its intended display size.
Frequently asked questions
Is JPEG compression always lossy?
Common JPEG photo encoding is lossy. JPEG has lossless coding modes in the standard, but they are not what most browser export controls produce.