The hidden cost of saving JPG three times
Not marketing — math. Each hop through a different online compressor throws away data you cannot get back.
JPG stores photos by approximating detail. The first save decides what to discard. A second save approximates the approximation. A third makes blocky artifacts visible in faces and text — especially when a portal forces you down to 50 KB.
This article is not arguing you should never use multiple tools. It explains when stacking them hurts and why FilezyAI runs enabled steps once.
A concrete chain
iPhone HEIC → convert to JPG on site A → compress on site B → resize on site C for a 600×600 application photo. Each step that outputs JPG re-quantizes DCT coefficients. Edges soften; skin texture clumps; small text on ID scans fuzzes.
One pass on convert + compress + resize + rename HEIC applies resize before a single lossy encode targeted at your KB band — not three separate encodes.
When multiple tools are fine
- PNG → PNG resize in a lossless editor, then one JPG export at the end.
- PDF text exports where JPG is not in the chain until the final step.
- Manual tuning in Squoosh when you inspect every preview pixel.
When to avoid stacking
- Tight KB caps on faces (passport, visa, exam photos).
- Text-heavy JPG scans of documents.
- Any workflow where you already compressed “good enough” on site B and site C compresses again.
Try it with your own eyes
Take the same source photo through three random JPG compressors, then through FilezyAI once with the same final KB target. Zoom to 200% on an eye or a date stamp. The difference is often visible — that is what exam reviewers see when they open your upload.