Use a one-megabyte allowance where the detail matters
A 1 MB image budget can leave useful room for a photographic sharing copy, a larger illustration or an attachment that needs more detail than a thumbnail. This preset sets the target to 1,000,000 bytes and measures the generated result. It is not a claim that every camera image, scan or graphic can retain its original resolution within that ceiling.
Before processing anything, compare the source with the destination's requirement. An image already below one megabyte may be ready to send. Re-encoding it can change pixels and embedded information without solving a problem. Use this workflow when you need a measured reduction or another deliberate output choice, and keep the original as your reference.
Start with viewing needs rather than the largest dimensions
For a photograph intended to fill a presentation slide, a sensible delivery grid may preserve the visible detail without carrying every camera pixel. For a technical illustration that the recipient will enlarge, the smallest labels may dictate a larger image. The same byte allowance has a different meaning in those two situations.
Consider whether the source contains unnecessary surroundings. A wide desktop screenshot with one small error panel wastes much of its available area on irrelevant interface content. A permitted crop can preserve the needed lettering more effectively than compressing the entire desktop. A photograph whose context is important should not be cropped merely to make a number easier to meet.
Noise is another practical factor. A dimly lit scene with fine grain can require more encoded information than a clean image of similar dimensions. A target that worked for a bright product photograph may be much harder for that source. The compressor measures the actual candidates because dimensions alone cannot predict the final bytes.
Keep quality and resolution as separate decisions
Preserve dimensions prevents the search from changing the pixel grid. Use it when the recipient requires a minimum or exact size, or when you have already prepared appropriate dimensions. The search then stays within the allowed encoding settings. Minimum quality is a boundary on that search, not a guarantee of an acceptable appearance.
If the target cannot be met, compare two explicit alternatives: permit a smaller grid, or permit stronger lossy compression. A smaller image at a comfortable quality can be clearer in its intended display than a huge image with severe artifacts. Conversely, shrinking a dense diagram can make it unusable even if the remaining pixels are encoded cleanly.
Allow resizing only when that tradeoff is acceptable. The tool does not silently lower dimensions because the current constraints are difficult. It reports the closest measured result and a missed target when appropriate. That distinction protects the meaning of your settings and keeps a failed size requirement visible.
Choose a format for the delivery, not a label of quality
JPEG is widely useful for opaque photographs, while WebP offers another option for supported recipients. Inspect smooth skies, textured shadows and small high-contrast details after compression. A quality percentage is specific to an encoding choice; it is not a common scale on which every JPEG and WebP can be ranked.
PNG is useful for transparency and graphic edges. Its lossless encoder does not use a lossy quality control here. A one-megabyte budget may be ample for a small graphic yet insufficient for a full-resolution photographic PNG. Lossless storage of decoded pixels also does not mean that source metadata, profiles or prior compression damage are preserved or repaired.
JPEG flattens transparency onto the selected background. Check the resulting edge and the destination design before choosing it for a logo or cutout. If alpha matters, use an appropriate format and reassess the permitted dimensions or target. A successful byte measurement is not worth losing a required visual property.
One megabyte is defined explicitly here
This interface uses decimal units: 1 MB is 1,000,000 bytes. Some applications show binary megabytes or round their display, which can make the same file appear to have a different size. Compare actual bytes with the destination's stated limit rather than relying on a rounded filename or thumbnail property.
The target is a maximum, not an exact final length. A result comfortably below it can be useful. The search does not add padding to consume the unused budget. If the image is already under the limit, consider whether another export is needed at all before generating a new version.
When a destination still rejects a measured result, check file type and dimensions as well as bytes. It may not accept WebP, or it may enforce a minimum resolution. Do not change only the extension to satisfy a filter; the actual image representation remains unchanged and may become harder for another person to identify.
Review the copy that leaves your device
Configured input admission is still PNG, JPEG or WebP within 20 MiB, 12 megapixels and 6,000 pixels per edge. Browser resources remain relevant beyond those ceilings. A heavily compressed source can occupy little disk space while requiring substantial memory after decoding.
The workflow does not copy source identifying metadata or preserve the original ICC profile. It uses browser color processing, so retain a separate master for archival, high-depth or tightly color-managed work. Visible personal information remains visible after metadata changes; a smaller file is not automatically an anonymous one.
Download the result and open it in the intended viewing application. Confirm size, orientation, format and the detail the recipient needs. A save request does not establish completion. For broader preparation, Prepare for Web combines presets and responsive copies; for another ceiling, return to Compress Image. Keep alternatives clearly named and work from the original so that each comparison has a dependable baseline.