Large images are one of the most common reasons websites load slowly. A single unoptimised photo from a modern smartphone can be 5–10 MB — and if you have three of them on a page, you've already loaded more data than most entire websites should serve.
The good news: you can typically cut image file size by 60–80% with no visible quality loss. Here's how.
Every byte a browser downloads delays page render. Google's Core Web Vitals — which directly affect your search ranking — measure how fast your page loads visual content (LCP). Oversized images are the single most common LCP failure.
Beyond SEO:
A properly optimised image is indistinguishable from the original to a human eye, but loads in a fraction of the time.
Format choice is the highest-leverage decision. The same image can be 3× smaller simply by choosing a better format.
| Format | Best for | Typical size vs JPEG |
|---|---|---|
| JPEG | Photos, complex images | Baseline |
| WebP | Photos and graphics | 25–35% smaller than JPEG |
| AVIF | Photos and graphics | 50%+ smaller than JPEG |
| PNG | Graphics with transparency | Larger than JPEG for photos |
Recommendation: Use WebP for broad compatibility, AVIF for the best compression. Both are supported by all modern browsers (Chrome, Firefox, Safari, Edge).
To convert to WebP or AVIF: imgshrnk Compressor → select output format → download.
Quality settings between 75–85% are indistinguishable from 100% for most photos in a browser context. Here's the practical guide:
Going below 70% starts to introduce visible artefacts — particularly in gradients, sky, and skin tones. Going above 90% wastes file size with no visible benefit.
A 4000×3000px image displayed at 800×600px still forces the browser to download and decode 4K worth of pixels. Resize to the actual display size before uploading.
For responsive images, a good rule of thumb is 2× the CSS display size (for high-DPI screens). A card that shows at 400px wide should have an image of 800px wide maximum.
Modern cameras embed a lot of data in image files: GPS coordinates, camera model, lens information, timestamps, even a full thumbnail copy of the image. This EXIF data adds kilobytes to every file and leaks location data.
Stripping metadata from a JPEG typically saves 5–30 KB per image. Over a whole product catalogue, this adds up.
Strip metadata automatically: imgshrnk Metadata Remover — re-encodes images via Canvas API, removing all EXIF in the process.
Here are real-world examples:
| Image type | Original | WebP 80% | Savings |
|---|---|---|---|
| Smartphone photo (JPEG) | 4.2 MB | 520 KB | 88% |
| Product photo | 1.1 MB | 210 KB | 81% |
| Website hero image | 890 KB | 195 KB | 78% |
| Blog thumbnail | 340 KB | 68 KB | 80% |
These numbers are achievable without any visible quality degradation in a browser at typical display sizes.
imgshrnk's batch compressor lets you drag and drop multiple images, choose a format and quality, and download everything compressed — all in your browser, with no uploads and no file size limits beyond 50 MB per image.
The tool shows you the file size reduction for each image before you download, so you can adjust settings if the savings aren't enough.