Initializing... drag & drop files here
Supports: JPG, JPEG, JFIF
To compress a JPEG, upload your photo, lower the Image Quality to about 75–80% (visually transparent for most photos), optionally resize down to your display dimensions, and click Compress. Everything runs on our servers — no install — and files auto-delete after a few hours.
Real result: in our production data the median JPEG drops ~74% (a 2 MB photo → ~0.5 MB) at quality 75–80%, which is near-indistinguishable from the original.
.jpg, .jpeg, and .jfif all work, and files are uploaded over an encrypted connection and auto-deleted after a few hours.A modern phone camera produces 3–10 MB JPEGs at 12–48 MP — overkill for a website, an email, or a chat attachment. Compressing routinely cuts file size by 60–90% with no perceptible quality loss for the intended use, especially when the display target is smaller than the camera's native resolution. Typical scenarios:
Working with a different source first? Convert HEIC to JPG from iPhone photos or PNG to JPG for screenshots before compressing, and use Resize JPG if you need precise pixel dimensions rather than auto-scaling.
Quality is the JPEG quantization factor exposed on a 0–100 scale. Higher means larger files and more retained detail; the perceptual curve is steep above ~90 and shallow below ~70.
| Quality | Typical size vs Q100 | Best for | Visible artefacts? |
|---|---|---|---|
| Q95–100 | 80–100% (huge) | Archival masters, print, photo editing source | None visible; mostly wasted bits |
| Q90 | ~50% | High-end product photography, fine textures (jewelry, fabric) | None on standard monitors at normal viewing distance |
| Q85 | ~30% | E-commerce, hero images, portfolio sites — the documented web "sweet spot" | None for almost all viewers |
| Q80 | ~22% | General web photos, blog images, social media uploads | None perceptible on natural photos |
| Q70 | ~16% | Thumbnails, list views, low-priority web assets | Faint blocking on flat gradients (skies) |
| Q60 | ~12% | Email previews, OG share cards | Visible blocking on close inspection |
| Q50 or below | <10% | Aggressive size-first targets, throwaway previews | Obvious 8×8 block artefacts and colour bleed |
The 85 vs 95 gap is the most lopsided trade in the format: roughly 50–60% smaller files for changes most viewers can't see at arm's length.
JPEG isn't the densest format anymore, but it's the most universally supported. Compare what you'd save in modern formats — but only if your audience's browsers support them.
| Property | JPEG | WebP | AVIF |
|---|---|---|---|
| Year released | 1992 | 2010 | 2019 |
| Typical size vs JPEG Q85 | baseline | ~25–35% smaller | ~50% smaller |
| Lossless mode | No | Yes | Yes |
| Alpha transparency | No | Yes | Yes |
| Animation | No | Yes | Yes |
| Max dimensions | 65,535 × 65,535 px | 16,383 × 16,383 px | 65,536 × 65,536 px |
| Bit depth | 8-bit | 8-bit | 8/10/12-bit |
| Browser support (per caniuse) | Universal | Chrome/Edge/Firefox/Safari 16+ (~96%) | Chrome 85+/Firefox 93+/Safari 16.4+ |
| Encode speed | Fastest | Fast | Slow |
In practice, ship JPEG as the <picture> fallback and serve WebP or AVIF to browsers that support them — that's the highest-impact image optimization you can do without breaking older clients.
Lower the Image Quality slider to 75–80% — that range is visually transparent on natural photos while cutting most of the bytes — and resize the image down to the dimensions it's actually displayed at. Re-saving an already-low-quality JPEG won't help much, since that detail is already gone.
Quality 80–85 is the documented sweet spot. The perceptual gap between Q85 and Q95 is invisible to most viewers at normal monitor distance, but the Q85 file is roughly 50–60% smaller. Above Q90 you're paying for bits no one sees; below Q70 you start to see 8×8 block artefacts on flat areas like skies and skin. If your tool only offers presets, "High" usually maps to Q80–85 — that's the right default.
Three usual suspects. (1) Your source is enormous — a 48 MP photo at Q80 is still ~2 MB. Turn on Auto Scale or compress with the file-size target instead of pure quality. (2) Your source is already compressed — re-compressing a Q85 JPEG to Q85 saves almost nothing because the data is already gone. (3) The image is full of fine detail (foliage, textured fabric, grainy night shots), which legitimately needs more bits than a smooth portrait at the same quality.
Yes. JPEG is lossy, so every save discards more high-frequency detail and the 8×8 block grid drifts slightly each time, compounding artefacts. One re-encode at a slightly lower quality is usually fine. Three or four passes show visible degradation. If you have the original RAW, HEIC, or PNG, compress from that source rather than from an existing JPEG.
Nothing — they're the same format and the same file. The .jpg extension exists because older Windows versions limited extensions to three characters; Unix-family systems never had that limit so .jpeg was common there. Both .jpg and .jpeg use the JPEG/JFIF compression standardized in ITU-T T.81 (1992). XConvert accepts .jpg, .jpeg, and .jfif interchangeably.
Progressive renders the whole image at low quality first, then refines in 3–5 scans; baseline draws top-to-bottom in a single pass. For images above ~10 KB, progressive files are typically 2–10% smaller AND give users the "instant blurry preview" loading experience. Below 10 KB the multi-scan overhead can make progressive slightly larger, so baseline wins. Modern browsers handle both fine — there's no compatibility reason to prefer baseline.
Yes. Pick Specific file size, enter your target in KB or MB, and the encoder iterates on quality (and, if Auto Scale is on, dimensions) until it lands at or just under your target. This is the right mode when a CMS or job board enforces a hard cap — "upload must be under 200 KB" needs a target, not a quality slider you have to guess at.
For natural photos at web sizes, 4:2:0 (default for JPEG) cuts file size by roughly 15–20% with no visible quality loss because the human eye resolves luma far better than chroma. 4:4:4 only matters for synthetic content with saturated edges — red text on blue background, UI screenshots, or charts — where 4:2:0 produces faint colour fringing. For camera photos, 4:2:0 is always the right choice; for screenshots-as-JPEG, consider PNG or WebP lossless instead.
Most online compressors strip non-essential EXIF to save bytes, but XConvert preserves what it can during JPEG re-encoding. If privacy matters (GPS coordinates on photos you share publicly), strip EXIF deliberately with an EXIF tool before sharing rather than relying on a compressor's behaviour. Note that orientation is preserved either way — your portrait-mode shots won't flip sideways.
processing happens on our servers and files are deleted automatically after a few hours. There's no account requirement, no watermark, and no file-count limit gating the compressor. Free users can compress JPEGs up to 300 MB each — more than enough for any camera JPEG (typical 48 MP raw-quality JPEG is 10–25 MB).