A compact copy for forms and small previews
This preset asks the compressor for an image at or below 100,000 bytes. That can be a useful delivery budget for a small photograph, an application attachment or a modest preview. It is not a guarantee that a large camera image will remain attractive at its original dimensions. PRVKIT measures the generated output and reports whether the target was met under the settings you allowed.
Choose a still PNG, JPEG or WebP from your device. Before changing controls, check what the recipient needs: accepted format, minimum dimensions, framing and maximum bytes. An employer's document upload, a profile image and a thumbnail in a content system may all mention “100 KB” while requiring quite different pictures.
Give the encoder a sensible pixel count
Suppose a photograph is much wider than the slot where it will appear. Preserving every source pixel can force the quality search to make aggressive compromises that the viewer never benefits from. A smaller image can look cleaner in its actual location because its encoding budget is concentrated on fewer pixels.
That does not mean you should always resize. A service might require a specific width or expect a document image to remain readable when enlarged. Keep Preserve dimensions when the pixel grid is part of the requirement. If the target is missed, treat that as useful information about the current constraints instead of assuming the tool should secretly change them.
A screenshot with a short error message is a good example of a different strategy. Crop irrelevant interface areas first, if permitted, rather than shrinking an entire desktop until the message becomes illegible. The compressor cannot know which words the support team needs. Establish that useful region before spending the byte budget on compression.
Compare photographic and graphic choices
JPEG is often a practical option for an opaque small photograph. Begin with the existing minimum-quality constraint and inspect facial details, fine edges and smooth backgrounds. WebP can provide an alternative where the receiving application accepts it, but compatibility is a real requirement, not a cosmetic preference.
PNG is appropriate when transparency or crisp interface text needs preserving. Its lossless encoding does not respond to a lossy quality slider in this workflow. If a photographic PNG is too large, choosing JPEG or WebP may be reasonable only when the alpha and compatibility decisions allow it. A tiny byte target does not override the purpose of the image.
For transparent artwork converted to JPEG, choose the background color and inspect the resulting edge. The flattened color becomes part of the pixels. It will not automatically adapt to the background of a website or document later. Keep an alpha-supporting format when that flexibility is essential.
What target search actually promises
The preset is 100 KB in decimal units: 100,000 bytes. The search evaluates actual encoded results within finite quality settings. If resizing is allowed, it can also reduce dimensions as part of that bounded process. It does not pad a successful file until it has exactly the target length.
Minimum quality is your lower boundary for lossy encoding, not a promise of visual quality. A detailed or noisy photograph can require more bytes than a clean portrait at the same dimensions. If the constraints prevent success, the reported closest result remains a missed target, even if it looks acceptable to you.
Do not derive a universal recipe from one successful image. A small group photograph has many faces, while a headshot spends more pixels on one face. The same quality and dimensions may have very different usefulness. Choose representative content when comparing strategies, and make the final decision with the actual source you intend to send.
Readability is a stronger check than the percentage
For a profile photograph, check eyes, hair and the edge against the background. For a form attachment containing text, read the required words and numbers in the downloaded copy. For a small product image, inspect the identifying features that distinguish it from similar products. A quality slider value cannot perform those checks for you.
View the file near the size it will be used and inspect any critical detail more closely. A fitted browser preview can make both a sharp image and a damaged one look reassuringly similar. The saved file is what the recipient receives, so reopen that file rather than relying only on the working preview.
If the destination rejects it, compare its exact maximum and supported format with the actual output properties. Some systems display binary units or enforce other rules. Do not change the extension to bypass validation: that leaves the encoded format unchanged and can create a confusing or unusable attachment.
Preserve a clean way back
The tool creates a separate local copy and does not overwrite the selected original. Re-encoding does not copy source identifying metadata or preserve the original ICC profile; browser color processing determines the result. Retain a master when capture information or color-managed editing remains important.
Configured source limits are 20 MiB, 12 megapixels and 6,000 pixels per edge for supported still PNG, JPEG and WebP. A 100 KB output target does not make every source admissible. File bytes, decoded pixels and browser memory are different resource boundaries.
Use Compare Images for a separate visual comparison after saving, or return to Compress Image for a different target. When dimensions are the main issue, choose them explicitly in Resize Image. Work from the original for each alternative, keep the copy that actually serves the task and save it before leaving the current page.