Product blog

Product blog

Product updates, field guides, and field notes—each with links to docs and the hub where relevant. High-level milestones also appear on the changelog.

  1. Two paths from one photo — a real ML model (background-removal segmentation, super-resolution upscaling) on capable hardware and a labelled classical fallback on modest hardware — both running on-device with no upload

    Product highlight · Jun 15, 2026

    When the fallback becomes the real thing: real AI models vs classical baselines in the browser

    The v1.7.0 "real models everywhere" shift explained: what a real model (a segmentation network for background removal, a super-resolution model for upscaling) actually buys you over a classical baseline like a bicubic resize, why the fallback still has a job, how the app decides which to run, and why telling you which one ran is the whole point of "honest everywhere".

    • Real models where hardware allows
    • Honest classical fallback elsewhere
    • It tells you which one ran

    NSS Background Remover

    Read
  2. Amazon listing image grid showing a compliant pure-white main image and lifestyle secondary slots with spec callouts

    Field guide · Jun 12, 2026

    Amazon product photo requirements (2026): the complete spec, explained

    The pure-white main image, the 85% fill rule, the resolution that unlocks zoom, and what belongs in your secondary slots — the whole Amazon photo spec without the guesswork.

    • Pure-white (255,255,255) main image
    • 85% frame fill + 1600px for zoom
    • Main vs lifestyle slots explained

    NSS Background Remover

    Read
  3. A product photo cost comparison: studio invoice and freelancer day-rate on one side, a phone and a free browser editor on the other

    Field guide · Jun 11, 2026

    The real cost of product photography — and how a browser tool replaces most of a studio

    A cost-accounting look at product imagery for small sellers: what studios, freelancers, and DIY setups really cost per finished image, why marketplaces keep raising the visual bar, and a repeatable phone-plus-browser workflow built on NSS Background Remover that produces catalog-grade images at near-zero marginal cost.

    • Cost per finished image is the metric
    • The visual bar keeps rising
    • A studio in a browser tab

    NSS Background Remover

    Read
  4. The all-in-one NSS Background Remover editor: layers, 3D scene, video editor, and on-device AI on one pipeline

    Product highlight · Jun 9, 2026

    Inside the all-in-one NSS Background Remover editor: layers, 3D, video, and on-device AI

    A guided tour of the all-in-one NSS Background Remover editor: how layers, selection, adjustments, 3D, the video editor, and the on-device cutout and upscaling models fit on one on-device pipeline that never uploads your files.

    • Layers, 3D & video editor
    • On-device cutout & upscaling
    • .nss-project saves, no upload

    NSS Background Remover

    Read
  5. The full NSS Background Remover story: on-device models, straight-alpha export, and the worker-isolation rebuild

    Product highlight · Jun 7, 2026

    How we built a background remover that runs entirely in your browser — and what it taught us about client-side AI

    The complete build story — privacy stance, models in the browser, straight-alpha export, the worker rebuild — and the lessons that came out of shipping real client-side AI.

    • Fully on-device AI
    • Straight-alpha, professional output
    • The rebuild that reshaped how we build

    NSS Background Remover

    Read
  6. Three model tiers — Lite, Standard, Pro — with their real download sizes and hardware requirements

    Field guide · Jun 5, 2026

    Honest AI tiers: Lite, Standard, Pro — sized in gigabytes, not hype

    Why naming real model sizes and recommending a tier your device can run beats promising everyone the heaviest model.

    • Lite 0 / Standard ~400MB / Pro ~2GB
    • recommendTier() probes WebGPU + memory
    • Delete downloaded weights any time

    NSS Background Remover

    Read
  7. An Etsy listing grid showing strong square thumbnails and a converting photo set

    Field guide · Jun 5, 2026

    Etsy product photos that convert

    What Etsy buyers respond to, the image specs and square-thumbnail rule, the mix of lifestyle and detail shots that converts, and how to prep it free and on-device.

    • The thumbnail earns the click
    • Lifestyle + scale + detail mix
    • Consistent, free, on-device

    NSS Background Remover

    Read
  8. A garment shown with the ghost-mannequin effect — shape intact, mannequin removed, inner neck visible

    Field guide · Jun 2, 2026

    Ghost-mannequin product photos without a studio

    What the ghost-mannequin effect is, why it converts for apparel, the front-plus-inner-neck two-shot method, and how to composite it without a studio.

    • The invisible-mannequin look
    • Front + inner-neck two-shot
    • Composited free, on-device

    NSS Background Remover

    Read
  9. Tracing a silent failure from symptom to undisposed model session to the worker-isolation fix

    Field guide · May 26, 2026

    Diagnosing a silent failure: the ONNX worker-session bug that broke tool execution

    A debugging story a developer hitting the same wall can learn from: the symptom, the trace, the root cause, and the structural fix.

    • Silent failure, no stack trace
    • WASM heap corruption from an undisposed session
    • Fixed by per-job worker isolation

    NSS Background Remover

    Read
  10. Capability detection choosing WebGPU or WebAssembly, with measured inference timings and a fallback path

    Field guide · May 25, 2026

    WebGPU vs WASM for client-side ML: what actually changed our inference speed

    WebGPU is faster; WASM is universal. Here is how a real tool uses both, and the measured difference it makes.

    • 2–5s WebGPU vs 8–15s WASM (RMBG-1.4)
    • Automatic capability detection
    • Fall back, then retry on GPU failure

    NSS Background Remover

    Read