📢 Help wanted: test new image compress (v2.15.0-dev.0) — big speed/size wins on image-heavy captures

[tinchox5]

TL;DR — 2.15.0-dev.0 ships a new compress option (ON by default) that downsamples oversized embedded images to roughly the resolution they're actually shown at. On captures with large images it can cut SVG output by ~90% and rasterization time by ~40%, with no perceptible quality loss. I'd love real-world testing before it goes stable.

What's new
Until now, SnapDOM embedded every image at its full natural resolution — even a 3000px photo shown in a 200px box. Those invisible pixels just bloated the output and slowed down rasterization.

compress downsamples each oversized raster to (a touch below) its visible resolution (box × scale × dpr), preserving aspect ratio and never upscaling. It covers:

• <img> (including cloned <canvas>/<video>)
• non-repeating CSS background-image
• SVG <image href>

Images already shown at or below their natural size are left untouched, so typical captures pay ~nothing.

Why it matters

The benefit depends on the export:

• Raster (toPng / toJpg / toCanvas / toBlob): much faster — the browser decodes/composites small images instead of huge ones. Final pixel size is essentially unchanged.
• SVG (toSvg / raw): much smaller output — the downsampled images are embedded in the SVG.

In a heavy gallery test (1 hero + several large thumbnails, scale 2): −44% toCanvas time, −96% SVG output, mean per-pixel difference < 1/255 (imperceptible).

⚠️ Behavior change: compress is ON by default in this dev build. If you need images embedded verbatim (e.g. you'll upscale the SVG later), pass compress: false.

Try it

  npm i @zumer/snapdom@2.15.0-dev.0
  # or
  npm i @zumer/snapdom@dev

  import { snapdom } from '@zumer/snapdom'

  await snapdom.toPng(el)                    // compressed (default)
  await snapdom.toPng(el, { compress: false }) // verbatim, for comparison

What I'd love you to check

Especially on real pages with large images:

1. Fidelity — capture the same element with { compress: false } vs default and compare. Any visible softening / blur, especially on
   logos, diagrams, or sharp graphics shown near their natural size?
2. Speed & size — how much faster / smaller for your use case? (share rough before/after numbers)
3. Correctness — any blank / broken images, wrong colors, lost transparency, or issues with background-image, retina (dpr: 2),
   object-fit, or cross-origin images?
4. Edge cases — tiled/repeat backgrounds, SVG charts (Highcharts/D3), gradients mixed with images.

When reporting, please include: browser + OS, the export method (toPng/toSvg/…), scale/dpr, and ideally a before/after screenshot or repro.

Thanks a ton 🙏 — this one could make a real difference for image-heavy captures, and your testing is what lets it ship with confidence.

[tinchox5]

Here there is a live example https://codepen.io/zircle/pen/LExxxLM