FlowVidFlowVid.

October 6, 2026 · 7 min read

What Is Image Compression? A Plain-English Guide

By The FlowVid Team, Image tooling engineers·Last updated October 6, 2026

How we checked this: Explanations below are cross-referenced against MDN's image format reference and Google's web.dev image-performance guidance. Numbers reflect typical phone-photo workflows we see day-to-day across the FlowVid tool suite, verified October 2026.

Image compression is how a 5 MB phone photo turns into a 200 KB web upload without looking obviously different. Under the hood, two completely different techniques share the name: one throws away information you can't see, the other finds smarter ways to encode what's there. Which technique a tool uses — and which one *should* be used for your particular image — determines whether the result is a crisp half-size file or a blocky mess. Here's what's happening, in plain English.

What image compression actually does

An uncompressed image stores every pixel's red, green, and blue values directly — a 4000 × 3000 photo takes about 36 MB raw. Compression finds shortcuts. Two broad strategies exist:

  • Describe repeating patterns efficiently — if a whole area is the same shade of blue, store it once with "this colour × 500 pixels" instead of 500 separate values. This is lossless; the pixels come back exactly.
  • Approximate what the eye can't notice — human vision is far less sensitive to small changes in colour than to small changes in brightness, so group similar colours and round off tiny variations. This is lossy; the exact pixels change, but the image looks the same.

JPEG uses the lossy approach aggressively. PNG uses the lossless approach exclusively. WebP and AVIF can do either, which is why they beat both older formats on size.

Lossless vs lossy — which to pick

The decision is content-driven, not personal preference:

ContentPickFormatTypical compression
Photograph (phone, DSLR)LossyJPG or WebP5–10× smaller
Screenshot of UI or textLosslessPNG or WebP-lossless2–4× smaller
Logo or iconLosslessPNG or SVG2–3× smaller
Scanned documentDepends on contentJPG for photos, PNG for textvaries
Transparent image (profile avatar)Lossless (needs alpha)PNG or WebP2–4× smaller
Pick lossy for photos, lossless for graphics with hard edges.

Why file size differs between formats

The same source photo can produce files of wildly different sizes depending on the format. A typical 12 MP phone photo:

  • BMP / TIFF (uncompressed) — ~36 MB
  • PNG (lossless) — ~10–15 MB
  • JPG at quality 85 (lossy) — ~2–3 MB
  • WebP at quality 85 (lossy) — ~1.5–2 MB
  • AVIF at quality 85 (lossy) — ~1–1.5 MB

WebP and AVIF are newer and better, but their support is now universal on the modern web — browsers released after 2022 handle both. Most portals and messaging apps, however, still expect JPG, which is why JPG remains the default output of general-purpose compressors.

How JPEG does its magic (briefly)

JPEG's trick is a 1970s insight dressed up in modern hardware. The algorithm first splits the image into 8 × 8 pixel tiles, then re-expresses each tile not as 64 colours but as 64 waves — low-frequency waves that describe the broad gradients, high-frequency waves that describe sharp edges. The high-frequency waves are the ones your eye is worst at noticing, so JPEG rounds them off aggressively. That's why JPEG shines on photos (lots of gradients) and struggles on text (nothing but sharp edges). It's also why heavily compressed JPEGs show those characteristic 8 × 8 blocks around high-contrast lines — the tile boundaries are leaking through.

How much can you compress without losing quality?

For a typical phone photo, re-encoding at JPEG quality 75–85 cuts the file 5–10× with no change your eye can see. Below quality 60 you start to notice softness; below 40 you see JPEG blocks around high-contrast edges. The sweet spot moves with content — portraits tolerate aggressive compression better than landscapes with fine foliage, because the eye forgives soft skin but notices mushy leaves.

If you need a specific KB target, don't guess at the quality setting — hand the file to a target-size tool like Compress Image or Compress JPEG and let it binary-search for the highest quality that fits. For strict portal limits, pre-sized options like Compress Image to 50KB or Compress Image to 100KB land you on the number exactly.

Compression and image dimensions — not the same thing

People often use "compress" to mean two different operations. True compression leaves the pixel dimensions alone and only shrinks the encoded file. Resizing (also called downscaling) throws away pixels and ends up with a smaller image on screen as well as a smaller file on disk. For most portals and web uploads you want both — and in that order, because resizing first means the compression step has less work to do and can keep a higher quality setting. Our reduce image size guide walks through the full two-step workflow.

Compressing an image yourself

The whole workflow runs in a browser, no install required:

  1. Decide if the content is a photo (go lossy / JPG) or a graphic (go lossless / PNG).
  2. If the dimensions are far larger than needed, resize first with Reduce Image Size — a smaller image compresses to a smaller file at the same quality setting.
  3. Pick the compressor. Compress Image for a general reduction, a KB-target variant if the destination demands a specific number, Bulk Image Compress for a folder.
  4. Download. The original stays untouched; the compressed version is a new file.

Frequently asked questions

What is image compression in simple terms?
It's a way to make an image file smaller on disk without making the picture look obviously worse. Two techniques exist: lossless (re-encodes without discarding anything) and lossy (throws away detail your eye can't see). JPG uses lossy; PNG uses lossless.
Does compressing an image reduce its quality?
Lossless compression doesn't — the pixels come back exactly. Lossy compression does, but at sensible quality settings (75–85 for JPEG) the loss is below what the human eye can detect. The visible quality drop only starts around quality 60 and below.
Why does compressing make images smaller?
Compression finds shortcuts. Repeating patterns are stored once instead of many times (lossless), and tiny colour variations you can't see get rounded off (lossy). Both reduce the number of bytes needed to represent the image.
Which image format compresses the most?
AVIF is currently the best — about 50% smaller than JPG at matched quality. WebP is close behind. JPG remains the universal baseline because every portal, email client, and messaging app supports it.
Can I compress an image multiple times?
You can, but you shouldn't. Each lossy re-save drops a little more detail; the damage stacks. Always start from the best original you have, and if you need a different target later, re-compress from the original rather than from the earlier compressed version.

Sources

Keep reading

More guides

Next

More guides