Compress Image Under a Target Size in KB

How does Target Size mode fit an image under a KB limit?

In Target Size mode, the compression engine uses a two-phase binary search algorithm rather than a single guessing pass. For lossy formats (JPEG and WebP), the algorithm first tests quality settings across up to 16 binary search iterations while keeping full native resolution intact.

If a photo is visually intricate and cannot reach your requested limit at full resolution even at the minimum quality floor (0.01), the search engine initiates a second binary search phase. This second phase scales dimensions down, searching across 14 scale steps down to 1% of the original width and height, until the file size constraint is met.

For PNG outputs, browser canvas encoders treat PNG as strictly lossless. Because quality parameters are ignored by canvas PNG encoders, the tool achieves target kilobyte thresholds on PNGs exclusively by scaling down dimensions.

Measured results: compressing a photographic image

We tested the compression engine in Chrome using a real 2048 x 1536 px smartphone photograph (sample-phone-photo.jpg, 2615.4 KB, captured on a Samsung phone at Prospect Park and released CC0). Running Target Size mode reliably produced outputs within budget across common application thresholds.

Targeting 100 KB yielded a 97.7 KB file (100049 bytes) at full 2048 x 1536 px resolution with quality calculated at 7.2%, while targeting 50 KB yielded a 43.5 KB file (44572 bytes) at quality 2.5%. At those low quality levels on a full-resolution photo, the JPEG encoder produces heavy block artifacts and visible pixelation. Targeting an aggressive 20 KB limit automatically scaled the canvas to 361 x 271 px with quality at 60.0%, producing a 19.1 KB file (19600 bytes) that fit securely beneath the 20 KB cap. Running Quality mode at the default 75% setting produced a 629.5 KB file (644623 bytes), representing a 75.9% reduction.

To preserve sharp details when aiming for a 50 KB cap, resize the photo before compressing: resizing this sample to 1200 px wide first (1200 x 900 px) and then targeting 50 KB produced a 48.8 KB file (49989 bytes, sample-compressed-1200px-50kb.jpg) at 10.3% quality, keeping text and edges noticeably cleaner than forcing full resolution into the same byte budget. For a 20 KB exam cap on a detailed photo, expect the image to scale down to about 361 px wide (361 x 271 px). All sample files and test outputs are preserved at reports/seo-batches/samples/b04/.

What happens to transparency during compression?

JPEG images do not have an alpha channel for transparency. When you compress a graphic with a transparent background to JPEG, the engine paints a solid white (#ffffff) background behind your artwork before encoding.

This prevents dark artifacts or black bounding boxes around cutouts and icons. If your project requires a transparent backdrop, select PNG or WebP as the export container. To isolate a subject from its background before compressing, use Remove Background.

Why does an aggressively compressed image look soft?

When you request a small kilobyte limit like 20 KB on a detailed camera photo, the engine must scale down the canvas once JPEG quality reaches its lower threshold. Scaling down discards high-frequency pixel details, making fine text or facial features look soft.

To preserve sharpness when meeting strict portal caps, crop out background clutter using Resize Image or crop the photo tightly around the subject first. If your file contains printed or handwritten notices, you can also extract the text using Image to Text (OCR), or generate an uncompressed static code with QR Code Generator.

Frequently asked questions