BgSub (In-Browser AI) · Background removal
Free, login-free on-device processing suits users who want to keep photos local; complex curls lose edge detail.
US-West (California, AS2914)
2048 × 2048 · PNG · Open native file ↗
Subsequent pricing and service checks are shown separately without altering historical scores.
One run does not establish all complex scenes, untested paid maxima or long-term uptime.
Plans & historical pricing
BgSub free web remover; excludes custom/API services · 2026-10-20
No usage-count limit published; API/batch services outside this scope
1 plan / billing-period records, all listed; — means that period is not recorded.
4096×4096 is an official supported specification; actual delivery remains tied to the historical test and attached file.
Cancellation, sources & verification record
No payment method required
Official plans and usage terms ↗ · 2026-10-20 Capture 1 ↗ · Capture 2 ↗
Pricing recorded on the test date · 2026-10-16
$0 (100% Free)
Completely free web application (no subscriptions or per-image fees)
Historical context only; not current purchase guidance.
Delivery & access
- Free delivery record
- 2048×2048 (100% pixel retention, 5.25MB watermark-free true-Alpha PNG)
- Attached file dimensions
- 2048 × 2048 · PNG
Read directly from the attached file; this does not establish free download availability at that size.
- Watermark
- None
- Free allowance
- Completely free with unlimited usage (computation runs on user's local device via WebGL/WASM)
- Account requirements
- No login required; exports full 2048×2048 original resolution directly
- Maximum specification record
- 2048×2048 (no-login local in-browser inference exports full 2048×2048 PNG, supporting up to 4096×4096)
May include advertised capabilities; this does not establish a tested paid tier.
- Batch & API
- Single-image interactive workspace
- Manual refinement
- Left sidebar includes built-in Edit (erase/restore brush), Color, Image, Filter, and Enhance tools
Tested result & sample
Constrained by lightweight client-side WebGL/WASM model capacity, complex curly hair outer edges show softer semi-transparent halos (mad_band=0.1481, SAD(k)=65.96) compared to heavy cloud models, but foreground recall reaches 99.38% at full 2048×2048 resolution.
- Edge error MAD
- 0.1481 · Lower is better
- Global error SAD(k)
- 65.955 · Lower is better
- Foreground retention
- 99.380% · Higher is better
- Background residue
- 0.106% · Lower is better
These historical measurements describe alpha error at edges (MAD), total alpha error / 1000 (SAD), foreground retention and background residue. Low-resolution outputs are aligned to the reference under the original protocol; this does not imply a full-resolution export.
- Fixture
- anchor_bgremove_v1
- Recorded environment
- US-West (California, AS2914)
- Local processing record
- 1.77 s
- End-to-end reference
- 1.77 s
Network wait includes round trips, queuing and polling error, not pure inference time. Local timings depend on the device; repetition counts are not standardized in legacy records.
Experience, rules & trust
- Disclosure
- Discloses local browser execution upfront and shows exact 2048×2048 (5.53 MB) size in the save modal
- Interaction
- Zero login requirement, zero marketing popups, zero cross-site redirects; processes locally in the browser and saves a 2048×2048 watermark-free PNG in one click.
- Entity record
- BgSub Independent Team (bgsub.com)
- Origin
- Independent Developer Project
- Domain history record
- Launched 2022
- Third-party tracking record
- Not recorded or not verified
- External information
- Praised on Product Hunt and privacy tool directories for keeping images 100% inside the local browser.
- Processing
- On-Device In-Browser Inference (WebGL / WASM, 0 bytes uploaded to cloud, 1.77s compute time)
- Retention statement
- Images are processed 100% in local browser memory and never transmitted to any server (verified via CDP network capture)
A recorded statement does not verify deletion or all data-processing behavior.
Verifiable audit trail & evidence
Four SHA-256-verified stages link the benchmark fixture, same-run interface screenshots, native exported artifact and archived pricing sources.
anchor_bgremove_v1 · Inspect fixture & file inspector →
Captures entry rules, result view and export disclosure.
Screenshots, native output and official claims support different conclusions. Evidence grades are not product safety certifications.
Evaluation basis & full scores · Overall 74.9/100
Overall = honesty & reputation × 30% + capability × 70%. Component maxima follow this report’s rubric. List scores are rounded for display; original precision is retained here.
Scoring version 2.3.1 · Test fixture anchor_bgremove_v1
Honesty & reputation · 96/100 · 30%
| Criterion | Score | Rationale | Basis |
|---|---|---|---|
| Pricing transparency | 40 / 40 | 100% free with no paywalls or hidden credit limits; clicking Save Image clearly displays exact output dimensions '2048 × 2048 (5.53 MB)' before saving. | Evidence A → |
| Interaction restraint | 20 / 20 | Zero login requirement, zero marketing popups, zero cross-site redirects; processes locally in the browser and saves a 2048×2048 watermark-free PNG in one click. | Evidence A → |
| Background & reputation | 21 / 25 | Built by an independent developer team (launched 2022) and well-regarded in privacy-focused and indie web tool communities. | Evidence C → |
| Engineering & privacy | 15 / 15 | CDP network packet inspection proves 0 bytes of image data are uploaded to any remote server—100% of segmentation inference runs locally in the browser via WebGL/WASM (15/15 privacy score). | Evidence A → |
Capability · 65.9/100 · 70%
| Criterion | Score | Rationale | Basis |
|---|---|---|---|
| Task quality | 32.9 / 65 | Transition band error mad_band=0.1481 · Global SAD(k)=65.955 · Foreground recall 99.38% / Background leak 0.106% | Evidence A → |
| Deliverable specifications | 20 / 20 | True Alpha channel (8/8); Max deliverable spec: 2048×2048 (no-login local in-browser inference exports full 2048×2048 PNG, supporting up to 4096×4096) | Evidence A → |
| Processing speed | 13 / 15 | Network-observed server wait: 1.77s (API mode: on_device_wasm; network upload 0s / end-to-end 1.77s shown for reference only, never penalized) | Evidence A → |
Legacy scores may include advertised maximum capabilities. Read tested output separately from paid capabilities. Records without matching tiers and scope are not ranked together.
Revisions & corrections
- 2026-10-16
Initial real-machine evaluation using synthetic ground-truth anchor_bgremove_v1 and CDP network timing isolation.