MagicEraser.org (AI Image Enhancer & Upscaler) · Image upscaling
Standard resizing and AI enhancement are clearly distinguished, with login-free trial-credit exports; this output was 1888px and tiny text was partly redrawn.
US-West (California, AS2914)
JPEG · Output attached · 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
Web subscriptions, displayed top-ups and enterprise quote; USD · 2026-10-20
11 plan / billing-period records, all listed; — means that period is not recorded.
Verified 10,000 credits for $120 and 30,000 for $330, both one-time payments with non-expiring credits. The pricing page quantifies upscaling, restoration, editing and video use but not background removal; general credits are not asserted removal counts.
Verification has gaps; unknown prices and quotas are not inferred.
Cancellation, sources & verification record
Cancel anytime in account dashboard
Official plans and usage terms ↗ · 2026-10-20 Capture 1 ↗
Large top-up package selections ↗ · 2026-10-20 Capture 1 ↗ · Capture 2 ↗
Pricing recorded on the test date · 2026-10-16
10 Free Credits for guests without login; paid plans from ~$6.9/mo
Freemium + credit subscription
Historical context only; not current purchase guidance.
Delivery & access
- Free delivery record
- 1888×1888 (3.69x upscale, unwatermarked JPG)
- Watermark
- No watermark
- Free allowance
- Guests receive 10 Free Credits without login; Enhance 2K costs 2 Credits per run (5 free high-res upscales)
- Account requirements
- No login required (uses the shared 10 Guest Free Credits pool)
- Maximum specification record
- 1888×1888 (Guest no-login 2K tier exports 3.69x unwatermarked JPG; supports up to 8K/16K)
May include advertised capabilities; this does not establish a tested paid tier.
- Batch & API
- Available via Batch Image Editor
- Manual refinement
- Manual erase / adjustment recorded in the historical flow
Tested result & sample
Achieves 27.34 dB full-image PSNR, 0.8878 SSIM, and a high 0.7543 edge gradient correlation (hf_ratio=0.615); carved wood grain on "REALTEST 2026" and brass knuckles are crisp, while the tiny 3rd-row micro-text on the calibration plate shows mild generative stroke reshaping.
- Fixture
- anchor_upscale_v1
- Recorded environment
- US-West (California, AS2914)
- Network-observed server wait
- 6.8 s
- End-to-end reference
- 8.19 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
- Clearly shows 2 Credits cost and target resolution tier before clicking Apply Changes
- Interaction
- Provides a clean 512px -> 1888px before/after comparison slider and direct browser download with zero login walls.
- Entity record
- MagicEraser.org / ImgUpscaler.ai Matrix
- Origin
- US / APAC Independent AI Team
- Domain history record
- 2023–2026
- Third-party tracking record
- Google Analytics · Aliyun SLS Web Tracking
- External information
- Rare transparency in explicitly warning users that standard upscaling does not add detail and pointing them to the AI Enhance tool.
- Processing
- Cloud async AI super-resolution pipeline (api-v2.imgupscaler.ai / image_enhance_v1)
- Retention statement
- Signed OSS URLs expire after ~4 hours
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_upscale_v1 · Inspect fixture & file inspector →
Captures entry rules, result view and export disclosure.
Official plans and usage terms (snapshot ↗) · Large top-up package selections (snapshot ↗)
Screenshots, native output and official claims support different conclusions. Evidence grades are not product safety certifications.
Evaluation basis & full scores · Overall 88.9/100
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_upscale_v1
Honesty & reputation · 92/100 · 30%
| Criterion | Score | Rationale | Basis |
|---|---|---|---|
| Pricing transparency | 33 / 35 | Exemplary upfront honesty: clicking the "Upscaler" tab displays a prominent notice stating "This feature performs standard upscaling and does not enhance image details. If you want to enhance image quality, please use our Enhance tool", guiding users to the true AI super-resolution Enhance module (2 Free Credits, no login required, 1888×1888 unwatermarked JPG export). | Evidence A → |
| Interaction restraint | 28 / 30 | Provides a clean 512px -> 1888px before/after comparison slider and direct browser download with zero login walls. | Evidence A → |
| Background & reputation | 18 / 20 | Powered directly by its sibling ImgUpscaler.ai v2 super-resolution pipeline (api-v2.imgupscaler.ai / image_enhance_v1). | Evidence C → |
| Engineering & privacy | 13 / 15 | Completes 3.69x cloud super-resolution in 6.80s over HTTPS without disguising basic interpolation as AI upscaling. | Evidence B → |
Capability · 87.6/100 · 70%
| Criterion | Score | Rationale | Basis |
|---|---|---|---|
| Task quality | 57.2 / 65 | PSNR vs 2048px GT: 27.34 dB · SSIM: 0.8878 · Edge Gradient Correlation: 0.7543 · High-Freq Sharpness Ratio: 0.615 · Calibration Plate PSNR: 15.94 dB | Evidence A → |
| Deliverable specifications | 18.4 / 20 | Clean unwatermarked export (8/8); Free scale factor: 3.69x (1888×1888 (Guest no-login 2K tier exports 3.69x unwatermarked JPG; supports up to 8K/16K)) | Evidence A → |
| Processing speed | 12 / 15 | Network-observed server wait: 6.8s (API mode: async_polling; end-to-end 8.19s 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 benchmark using anchor_upscale_v1 (512px -> 2048px GT).