What is actually happening
A signature is mostly thin, high-contrast strokes on a plain background. Lossy compression works by discarding detail that is hard to perceive across a whole image — and in a mostly empty frame, a one-pixel pen stroke is exactly the kind of detail it discards first. Around the stroke you then get ringing and blocking artefacts, which read as fuzz, and the line appears to thicken, break or double.
Resizing causes a different problem. Downscaling averages neighbouring pixels, so a stroke that was one or two pixels wide is blended into the background and loses contrast. Upscaling does the reverse: it smears the stroke over more pixels and it looks soft. Neither is corruption — both are the maths of resampling applied to a shape with almost no redundancy.
Settings that keep a signature legible
- Crop tight around the signature first, in Crop Image. Compression is a fixed cost spread across the whole frame; the less empty background there is, the more of the budget goes to the strokes.
- Keep it PNG when there is no byte limit. PNG is lossless, so a cropped signature stays exactly as sharp as the original scan. It is also the format that preserves transparency.
- If a byte limit forces lossy output, stay high — quality 90 or above — and reduce size by cropping and modest dimension changes rather than by dropping quality far.
- Avoid repeated save cycles. Every JPEG re-encode is another generation of loss. Resize and compress once, from the original, using Signature Resizer rather than several tools in a row.
- Check the result at 100% zoom before submitting — not at fit-to-screen, where every artefact disappears.
If the source itself is a photo of a signature taken in low light, no setting will recover detail that was never captured. A re-shot image on plain white paper in good light, or a drawn signature from Signature Maker, will beat any amount of post-processing.
A worked example
Suppose the scan starts at 2400 × 800 as a 1.8 MB PNG. Cropping to the signature itself might bring it to 1600 × 500 and 900 KB — still PNG, still sharp. Converting to JPEG at quality 90 could land around 120 KB. Halving the dimensions to 800 × 250, which is still ample for a form field, commonly reaches 35–50 KB. A final small quality step usually finishes the job. These figures are illustrative; the sequence is the part that transfers to any signature.
Where the signature will be used
- **Inside an image field on a form** — a small, sharp JPEG or PNG at the exact pixel size the form states. Signature Resizer fits it to a box or exact dimensions without stretching.
- **Placed onto a PDF** — keep PNG so any transparent background stays transparent, then position it with Sign a PDF.
- **Printed** — work at higher resolution. See DPI and the printing guides for how many pixels a printed signature actually needs.
Common questions
Should I compress a signature at all?
Only if a limit requires it. Without a byte limit, PNG at the original resolution is the safest option and avoids the artefacts entirely.
Why does my signature look fine in the preview but fuzzy after upload?
The form may be re-encoding your image on its side, or displaying it far smaller than your preview. Check the file you actually uploaded with Image Metadata — if your copy is clean, the blur is being introduced downstream.
Is a photograph of a signature acceptable?
It depends entirely on the receiving system. A clean, evenly lit photo cropped tightly is often fine; a skewed photo with shadows usually is not. When in doubt, use a drawn signature or a flatbed scan.