About Imagyc

The idea

Almost every online image tool asks you to upload your file to someone else's server. For anyone working with client material, that is a problem nobody likes to think about. Imagyc started from the question: can it all be done without uploading anything?

It can. A modern browser has everything needed. Your image is opened, processed and returned inside your own device.

The tools

• Compress, convert and resize: JPEG, PNG, WebP, AVIF and JXL, with resolution reduction and no metadata.
• Convert to SVG: a logo, icon or photo as a clean one-color SVG for Elementor and WordPress. Opens PNG, JPG, PDF, AI, PSD, TIFF, HEIC and SVG.
• Vectorize: color SVG in flat layers, with color swapping and a poster mode for photos.
• Background remover: a photo, product or logo as a transparent PNG or a JPG on white, with AI running right in your browser.
• Compress PDF: shrinks PDFs full of photos and scans while keeping the text selectable, password-protected ones included.
• PDF to JPG: every page of the PDF becomes a JPG or PNG image — a one-page receipt takes a single click.

What that changes in practice

• Original resolution, always. No reason to limit what costs us nothing.
• No watermark and no sign-up.
• No usage limit: up to 10 images at a time, as often as you like. A hundred images cost the same as one: nothing.
• Works offline once the page has loaded.

How we test

The numbers on this site come from measurement, not guesswork. Here is what we measured for each tool, on which files and against what.

• Compressor. Six real images (a 7-megapixel photo, two product shots, a gradient, a screenshot with text and a flat illustration) went through a quality sweep in every format. Each result gets two scores: MS-SSIM on luminance, for structure, and SSIM on the color channels, for the colored fringe around text. The lower one counts. The reference was TinyPNG, on the same image at the same score, and each format's final setting was checked by eye on a 100% crop.
• Convert to SVG and Vectorize. The engine was written from scratch, with no third-party library, and is measured against reference drawings: the traced outline matches 99.6% to 100% of the original's area. Every change to the engine is compared with the previous version before it ships.
• Background remover. The automatic choice between person, object and person + object was checked on 61 photos (ten of our own and 51 from the public samples of the rembg and U²-Net projects) and on six composites with gaps, such as a racket, a chair and a fence. The edge was compared with Canva's cutout of the same photo, hair against foliage: the shape matches (IoU 0.97), and the green-tinted edge dropped from 51% to about 4% of edge pixels. Canva's sits at 1.6%.
• Compress PDF. Eight PDFs of different kinds: a Word report, a scanned document, a LaTeX paper, a CMYK file, a slide deck, a 33-page manual, a bank receipt and a protected PDF. Every output passes qpdf's check, and its text is compared with the original's. The Word report, for example, went from 20.2 MB to 923 KB with identical text.
• Before every release. An automated test checks the two worst bugs the compressor ever had (a color disappearing from a three-color logo, and soft transparency turning on-or-off), and every page is opened on desktop and on a phone.

Metrics rank the candidates; the eye decides what is invisible. When the measurement contradicted the hunch, the measurement won.

Guides

The articles explain image formats, resolution, filenames, metadata and SVG, with a source for every claim.

Contact

Suggestions, bugs or proposals: contact@imagyc.com.