On this page

What is BgSub?

BgSub is a specialist AI tool from BgSub project for Product images. BgSub is a browser image tool for removing backgrounds and preparing subject cutouts. An independent seller can isolate an authorized product image for a listing, while a creator can prepare a subject for a thumbnail or composite and inspect the edges before use. An official blog article dated 2022-08-21 calls it an independently developed project, and its public author uses the alias DeepKnight. Current real name, legal operator, team size and controlling ownership remain unknown. The reviewed official site and native source describe automatic removal, original/processed review, image adjustments and PNG saving. Its FAQ advertises free processing, rights to processed images and support up to 4096 x 4096, while native source also contains account, credits and full-size routes; current full-size and account entitlements require verification. Ordinary in-browser processing and optional Contribute image upload are separate paths, and website logs/cookies have their own Privacy terms. On 25 October 2026, both original 512 x 512 synthetic cases were executed once after explicit Terms approval. Both failed, each passing three of six conditions: the bottle lost most of its cream label, while the mug missed body-opacity and handle-hole-transparency thresholds despite passing overall silhouette IoU. First native automatic results, PNGs and every failed condition are retained. This profile covers a supervised specialist tool rather than autonomous store management or publishing.

Best suited for

  • Independent sellers, creators and individuals who need a reviewable cutout of an authorized product or subject in a browser. It is a practical option to evaluate when quick background separation matters and a person can inspect outline, fine structures, transparent holes and original colors before reuse.
  • A pilot focused on first native browser cutout, original subject/color retention and low-contrast handle-hole boundary, using an authorized local source image with a clearly defined subject; desired dimensions, transparency and intended listing or creative use. Ordinary import can invoke automatic removal. Optional image URL, sample, edit, FullSize/account and contribution paths have separate behavior and are excluded from the frozen two-case pilot.

Not suited for

  • Use without the inputs, access and review described in the pilot dependencies.
  • Both original fixed 512x512 cases were executed on 25 October 2026 and failed, each passing 3/6 conditions. Bottle IoU is 0.838686 and cream-label opaque retention is 0.433129%. Mug IoU 0.958243 passes, but body opacity 92.0616%, hole transparency 89.2258% and source-color retention fail. The first PNGs were not repaired or replaced.
  • The finite geometric bottle and low-contrast mug cases do not establish natural-photo quality, hair/fur performance, complex scenes, broad accuracy, processing speed or a product ranking.

Capabilities, with sources

  • 01The official site describes automatic image-background removal and background replacement for subject cutouts.Official vendor statement · checked 2026-10-03Source ↗
  • 02A dated official blog post on 2022-08-21 describes BgSub as an independently developed project.Official vendor statement · checked 2026-10-03Source ↗
  • 03The official-home-linked blog and its linked Blogger profile publish the author alias DeepKnight.Official vendor statement · checked 2026-10-03Source ↗
  • 04The official FAQ states that processed-image rights belong to the user and advertises free processing; current route entitlements still require verification.Official vendor statement · checked 2026-10-03Source ↗
  • 05The official FAQ advertises support for images up to 4096 x 4096; this is a documented ceiling rather than a verified current export or free allowance.Official vendor statement · checked 2026-10-03Source ↗
  • 06The retained native source reads an ordinary local image through FileReader and sends image data into a browser-worker removal path.Official vendor statement · checked 2026-10-03Source ↗
  • 07The native source includes Original/Processed review, erase/restore, background, transform and filter controls; effects of the optional editing controls were not tested in the frozen native pilot.Official vendor statement · checked 2026-10-03Source ↗
  • 08The native source contains Save controls and a Save image dialog whose Save link downloads a PNG named from the source image.Official vendor statement · checked 2026-10-03Source ↗
  • 09The official historical blog states that routine background processing occurs in the browser without uploading the source image.Official vendor statement · checked 2026-10-03Source ↗
  • 10The native source includes a distinct optional Contribute route that uploads a chosen image for algorithm improvement.Official vendor statement · checked 2026-10-03Source ↗
  • 11The English Privacy text separately describes website logs, cookies, advertising and possible account/contact information.Official vendor statement · checked 2026-10-03Source ↗
  • 12The Terms require a trademark license agreement for use of the BgSub logo or artwork in linking.Official vendor statement · checked 2026-10-03Source ↗

Inputs and outputs

Inputs

An authorized local source image with a clearly defined subject; desired dimensions, transparency and intended listing or creative use. Ordinary import can invoke automatic removal. Optional image URL, sample, edit, FullSize/account and contribution paths have separate behavior and are excluded from the frozen two-case pilot.

Outputs

Two first native 512x512 RGBA PNGs were obtained through Save Image > Save on 25 October 2026. Both contain stored transparency but fail their fixed quality conditions: bottle label/body preservation and mug body/hole preservation are insufficient. Native PNGs and unchanged source images are retained for review. A downloaded cutout is not evidence of publication or listing updates.

Ecommerce & Retail fields

Ecommerce & Retail evidence fields for BgSub
Workflow stagesecommerce product imagesSource 1
Fields awaiting verification · 7

These fields have no verified value in this profile. Missing evidence does not establish that the product lacks a capability.

Ecommerce & Retail evidence fields for BgSub
Commerce platformsNot verifiedNot verified in the reviewed official material.
Integration by platformNot verifiedNot verified in the reviewed official material.
Store data & permissionsNot verifiedNot verified in the reviewed official material.
Output formatsNot verifiedNot verified in the reviewed official material.
Batch processingNot verifiedNot verified in the reviewed official material.
Localization languagesNot verifiedNot verified in the reviewed official material.
Approval requirementsNot verifiedNot verified in the reviewed official material.

Content Creators & Social Media fields

Content Creators & Social Media evidence fields for BgSub
InputsAn authorized local source image with a clearly defined subject; desired dimensions, transparency and intended listing or creative use. Ordinary import can invoke automatic removal. Optional image URL, sample, edit, FullSize/account and contribution paths have separate behavior and are excluded from the frozen two-case pilot.Source 1
Fields awaiting verification · 9

These fields have no verified value in this profile. Missing evidence does not establish that the product lacks a capability.

Content Creators & Social Media evidence fields for BgSub
Creator platformsNot verifiedNot verified in the reviewed official material.
Content formatsNot verifiedNot verified in the reviewed official material.
OutputsNot verifiedNot verified in the reviewed official material.
Aspect ratiosNot verifiedNot verified in the reviewed official material.
Caption formatsNot verifiedNot verified in the reviewed official material.
Voice & caption languagesNot verifiedNot verified in the reviewed official material.
Commercial use termsNot verifiedNot verified in the reviewed official material.
Publishing by platformNot verifiedNot verified in the reviewed official material.
Approval requirementsNot verifiedNot verified in the reviewed official material.

A practical BgSub workflow

  1. Prepare the first native browser cutout, original subject/color retention and low-contrast handle-hole boundary fixture: One existing authorized synthetic 512x512 RGB bottle image: teal body, black cap, cream label with three dark marks, light-gray background. This is a finite geometric fixture, not a natural product photograph. Independent full/core/background reference masks remain evaluator-only.
  2. Check BgSub access through Web app and confirm the selected feature’s actual permissions.
  3. In the actual official BgSub browser UI, use Open Image to import only the frozen bottle RGB image once. Preserve the first native automatic background-removal result and download it through native Save/Save Image > Save as the first unedited PNG at original 512x512. Retain the complete teal bottle, black cap, cream label and three dark label marks, and make the exterior background truly transparent. Do not change a background, draw or supply a mask, repair/edit pixels, resize/move the subject, use FullSize/autoFullSize/login/credits/contribution/batch or retry.
  4. Inspect a real first unedited native BgSub PNG download, exactly 512x512 with stored transparency, complete original bottle/cap/label/detail colors and all six fixed checks satisfied. Actual failures and unknowns remain visible. Compare it against the source input and retain the output/action log.
  5. Run the boundary case: In a fresh source case in the same official BgSub task session, use Open Image to import only the frozen low-contrast mug RGB image once. Preserve its first native automatic background-removal result and first unedited native PNG download through Save/Save Image > Save at original 512x512. Retain the near-white body, dark stripe and thin handle, while removing background visible through the real handle hole and outside the mug. Do not redraw/repair the handle, fill the hole, change contrast/background, supply a mask, resize/move/filter/enhance, use FullSize/autoFullSize/login/credits/contribution/batch or retry. Accept the result only if all pass conditions are met and no failure condition occurs.

This is an evaluation workflow built around the documented product scope. Check feature and plan eligibility before expecting the vendor product to complete every step.

Setup and integrations

Official hosted browser webapp. Reviewed source routes an ordinary local file through FileReader into a browser worker; page initialization may load processing resources. Self-hosting, complete offline operation, current model identity and network isolation have not been established.. Documented access methods: Web app. Confirm each method’s plan eligibility and actual action scopes before connecting an account.

Access and setup steps

  1. Start at the official webapp, https://bgsub.com/webapp/. Read the current Terms and image-processing conditions before opening an image, because the entry states that opening an image or URL means agreement.
  2. Choose one image you are entitled to process and preserve its original file. For a useful first review, pick a subject whose outline, narrow parts, label and colors can be compared with the original.
  3. Inspect the actual image-size, FullSize/autoFullSize, account and credit conditions before import. The official free and 4096 x 4096 statements do not establish the entitlement of every current route.
  4. Use the native Open Image control for the selected local file. Static code indicates that import can trigger automatic removal immediately; do not treat a first import as an unrecorded dry run.
  5. Wait for the same automatic operation and record any initialization, gate or error separately. A pending operation is not a completed cutout, and a wait timeout alone does not establish a failed image.
  6. Compare Original and Processed where those controls are available. Inspect the complete subject, cap or label, thin handles, internal holes and low-contrast edges. Preserve the first automatic result before making an optional edit.
  7. Use the current native Save control, or the desktop Save Image control if present, then the Save link in its Save image dialog. Check the downloaded PNG itself for dimensions, real alpha transparency and retained subject color; a checkerboard screenshot is insufficient.
  8. Treat background replacement, erase/restore, filters, transforms, AI Coloring and full-size processing as separate choices. They are not part of the frozen first-result cases and can change the subject or output conditions.
  9. Before putting a cutout in a listing or social post, review product appearance, source rights and destination requirements. Keep the optional Contribute image-upload flow separate from ordinary cutout work; review website logging and cookies separately from the image-processing claim.

Test access: public tool. On 25 October 2026, after the user explicitly approved the by-use Terms, two unchanged synthetic 512x512 inputs were each imported once using native Open Image. Automatic first results and the first Save Image > Save PNGs were retained; both cases failed 3/6 conditions. No URL input, manual repair, retry, FullSize, login, credits, purchase or optional contribution was used. Backend identity, network isolation and total cost remain unknown. Open the official access or installation page ↗

Pilot dependencies and unexecuted setup checks

Pilot dependencies

  • The frozen BgSub contract, exact two 512x512 RGB inputs, evaluator-only masks and measurement script are retained unchanged. The native pilot completed on 25 October 2026 after explicit user Terms approval: one local Open Image import, automatic first result and first Save Image > Save PNG per source in the same task session. Original case steps retain preparation-stage wording; current authorization, outputs and failed conditions are recorded separately. No mask was supplied to the product and no repair, alternate engine, FullSize/account/credits, contribution, batch or quality rerun occurred.
  • On 25 October 2026, after the user explicitly approved the by-use Terms, two unchanged synthetic 512x512 inputs were each imported once using native Open Image. Automatic first results and the first Save Image > Save PNGs were retained; both cases failed 3/6 conditions. No URL input, manual repair, retry, FullSize, login, credits, purchase or optional contribution was used. Backend identity, network isolation and total cost remain unknown.
  • Confirm free processing advertised · current route entitlements unverified against the current vendor terms; usage and connected-service costs can affect the pilot.
  • Create a test workspace or use public/authorized material. Keep an input baseline, output artifact and action log for comparison.

Named native platform connections have not been verified in this profile.

Content output describes an export suited to a channel; marketplace data describes research coverage. Exact data scopes and permissions need a setup review.

Integration fields awaiting verification

API: Not verifiedNot verified in the reviewed official material.

Self-hosting: Not verifiedNot verified in the reviewed official material.

Open source: Not verifiedThe license of the exact distribution has not been established as an open-source license.

Pricing and additional costs

Free processing advertised · current route entitlements unverified

The retained official FAQ states that processed images belong to the user and processing is free. It also advertises maximum 4096 x 4096 support. The native source separately includes verified-account/token handling, free/pay-as-you-go/subscription credit fields and FullSize/autoFullSize. Current prices, quotas and full-size entitlement have not been established. A small-image worker branch clears Authorization at no more than 300000 pixels; that remains a source observation. In the 25 October 2026 pilot, two ordinary 512x512 images were processed and exported without login, purchase or FullSize. Per-run charge/credit debit and total operating cost were not measured.

Price details awaiting verification

Do not equate the free FAQ statement with every feature or resolution, or infer a mandatory charge solely from account/credit code. Two ordinary 512x512 outputs were obtained without signing in or checking out; no FullSize, account, credit, purchase or broader quota entitlement was tested. Actual per-run total cost remains unknown.

Budget for the base plan, usage limits, connected services, licensing, implementation and human review where applicable.

Recorded tests and results

Product-specific plan: The original cases were written for this product. Results cover only the recorded cases and configuration.

Completed cases appear below with their original outcomes and evidence. Unexecuted plans and access attempts are collapsed.

See the testing method and all product plans →

Actual public vendor UI product execution

BgSub · Product version: UI-displayed label: 220907 · Native Open Image automatic background removal and first Save Image > Save PNG export · 2026-10-25T05:54:50.437Z

Scope: Two original frozen 512x512 synthetic BgSub cases, one native Open Image automatic removal and first Save Image > Save PNG export per case, with all six unchanged acceptance conditions.

Observed conclusion: Both original cases failed, each passing 3/6 conditions. Bottle IoU 0.838686 and label opaque retention 0.433129% fail. Mug IoU 0.958243 passes, while body opaque retention 92.0616%, hole transparency 89.2258% and source-color retention fail. First native outputs are retained unchanged.

Execution metadata, usage and audit scope

Model: Not disclosed; model tag: Not disclosed; digest: Not disclosed; inference runtime: Not disclosed; runtime version: Not disclosed.

Official entry: https://bgsub.com/webapp/

Tested pages: https://bgsub.com/webapp/

Observed UI settings: inputTransport: Native Open Image local file chooser; sourceDimensions: 512x512 RGB; importsPerCase: 1; automaticResultsPerCase: 1; sameSession: true; export: Save Image > Save; exportsPerCase: 1; processedDisplay: 100% 512 × 512; manualRepair: false; fullSize: false; login: false; contribution: false; qualityRetry: false. These describe visible controls, not inferred model configuration.

Capture method and scope: Root agent retained native UI screenshots, visible DOM, first-result observations, export dialog and download receipts. A separate agent checked exact first PNG bytes and recomputed frozen metrics offline. No replacement image model or private product API was used.

Authorized test scope: Before the first import, the user explicitly said 同意条款并继续 BgSub 测评. Approval covers the two existing public synthetic inputs, ordinary native processing and export; no private images, reference-mask input, account, purchase or expanded feature scope.

Backend calls and hidden server actions were not observed. Browser metadata does not identify the vendor inference runtime.

Reported tokens: input not recorded, output not recorded. No usage source was recorded.

Measured cost: Not measured. Two ordinary native 512x512 imports and PNG exports, without login, purchase or FullSize. Per-run charge/credit debit and total operating cost were not measured.

Audit: Separate-agent review of retained native UI evidence and exact frozen-script pixel recomputation; native product actions were performed once per source by the root agent.. Recorded read-access entries: 0; blocked-action entries: 0. Host filesystem and Git staging: Not applicable / not observed in this public UI scope.

Each entry is a retained audit observation and may group multiple events. Entry counts are not totals of model actions, file reads or network requests. The downloadable execution record retains the complete entries.

Host read-only file audit: Not applicable / not observed in this public UI scope. Any retained artifact hashes establish captured bytes only.

  • Two fixed 512x512 synthetic geometric images and one first native automatic removal per source. These results do not establish natural-photo quality, overall accuracy, repeatability, a product ranking or the full editing feature set.
  • The user explicitly approved the by-use Terms before the first image import. The original frozen contract retains historical pending-authorization fields; later approval is stored separately rather than editing that pre-inference record.
  • Both inputs used the native Open Image local file chooser, followed by automatic processing and the first Save Image > Save PNG download in the same task session. No URL import, manual repair, mask submission, FullSize/autoFullSize, login, credit purchase, contribution, sharing, batch operation or quality rerun was used.
  • The native Web App initial import screen displays version:220907, retained only as a UI-displayed label. This does not identify a model, inference runtime or backend release. Backend model/provider/build, token usage and actual network processing route were not observed. Browser use does not establish wholly local/private or offline processing.
  • No checkout or account upgrade was performed. Per-run charges, credit debits, hardware, electricity and total operating cost were not measured; total cost remains unknown.
  • startedAt records the approved session boundary and completedAt the final observed download receipt. Inference start/completion was not timed; durationMs remains unknown for both cases.
  • A separate Codex agent reviewed saved evidence and independently recomputed pixel measurements. No independent human review, hidden-server instrumentation or full network/host-filesystem audit was performed.
  • Public receipts omit only private absolute paths and explicitly record their projection. Native PNGs, input/reference images, frozen contract, measurement script and visible UI captures remain unmodified.
First native BgSub bottle cutout and original cap/label preservation Executed · failed

Actual input

One existing authorized synthetic 512x512 RGB bottle image: teal body, black cap, cream label with three dark marks, light-gray background. This is a finite geometric fixture, not a natural product photograph. Independent full/core/background reference masks remain evaluator-only.

Expected behavior

A real first unedited native BgSub PNG download, exactly 512x512 with stored transparency, complete original bottle/cap/label/detail colors and all six fixed checks satisfied. Actual failures and unknowns remain visible.

Observed result

Failed 3/6 conditions: silhouette IoU, bottle-core opacity and source-color retention. The cream label was largely removed; only 0.433129% of its frozen group remains opaque. Native provenance, PNG/alpha and exterior transparency passed.

Original response and native screenshots

Native BgSub test screenshot for bgsub-primary: bgsub-primary-first-jpg
Actual vendor UI capture · Open full screenshot

Original expected-behavior review

failed: Failed 3/6 conditions: silhouette IoU, bottle-core opacity and source-color retention. The cream label was largely removed; only 0.433129% of its frozen group remains opaque. Native provenance, PNG/alpha and exterior transparency passed.

Recorded duration: Not recorded

Acceptance conditions

  • passed: The exact original input hash is unchanged, only that RGB image was imported once through the actual official BgSub Open Image control, and the retained evidence proves its first native automatic background-removal result followed by its first unedited PNG downloaded through the actual native Save control (or available Save Image control) and download-dialog Save link. Record current labels; Save PNG is not a literal label established by the static source. No reference mask, previous product output, manual repair, FullSize/autoFullSize, login/credit path, contribution/upload/share action, batch processing or unrelated external action was supplied/performed. Explicit Terms approval preceded the native Open Image local-file import. Operator record retains one import, automatic first result, zero edits/retries and one Save Image > Save download. Input and output SHA-256 match retained bytes. Visible UI scope only; hidden backend actions were not observed.
  • passed: The actual first downloaded native artifact is decodable PNG, exactly 512x512, with a stored alpha channel or PNG transparency data; decoded alpha has minimum <=10 and maximum >=245. Checkerboard visibility alone does not pass. The first native export is a decodable 512x512 RGBA PNG (36356 bytes), with stored alpha ranging from 0 to 255.
  • failed: Using all 262144 pixels without alignment, cropped exclusions or ignored edges, alpha>=128 foreground has IoU>=0.90 against the full frozen bottle reference mask. Foreground intersection 44260 / union 52773 gives IoU 0.8386864495101661, below the fixed 0.90 minimum.
  • failed: At least 95% of all frozen bottle-core pixels retain alpha>=245. Bottle-core opaque retention is 38289 / 48769 = 78.5109393262%, below 95%.
  • passed: At least 95% of the frozen exterior-background pixels have alpha<=10. Exterior-background transparency is 205299 / 205299 = 100%, meeting 95%.
  • failed: In every pre-frozen source-RGB group within each foreground reference core, alpha>=245 coverage meets that region's fixed minimum fraction (95% bottle/body, 90% mug handle). Of those retained opaque pixels, at least 99% differ from the source RGB by at most 2 in every channel. All per-group numerators and denominators are reported; zero opaque pixels cannot pass. Transparent-pixel hidden RGB outside these cores is not scored. Cream-label group retains only 41 / 9466 = 0.4331290936% opaque pixels, and dark detail marks retain 1178 / 1757 = 67.0461013090%, both below 95%. Teal body retains 34414 / 34890 = 98.6357122385%; cap retains 2656 / 2656 = 100%. RGB matching among retained opaque pixels is 100% for each group, which cannot rescue the failed source-color opaque-coverage requirement.
First native BgSub low-contrast mug, narrow handle and real hole boundary Executed · failed

Actual input

The unchanged distinct synthetic 512x512 RGB near-white mug on light-gray background, with a narrow handle, real background-colored handle hole and dark body stripe. The full/body/handle/hole/exterior masks existed before any Photopea call and remain evaluator-only.

Expected behavior

A genuine first unedited native BgSub PNG download, exactly 512x512 with real hole/exterior transparency, retained low-contrast body/dark stripe/narrow handle and all six fixed checks satisfied. Lost handle, opaque hole or changed source color remains a separately reported failure.

Observed result

Failed 3/6 conditions: body opacity, handle-hole transparency and source-color retention. IoU 0.958243 passes, but body retention 92.0616% and hole transparency 89.2258% miss their independent 95% thresholds. Native provenance, PNG/alpha and silhouette passed.

Original response and native screenshots

Native BgSub test screenshot for bgsub-boundary: bgsub-boundary-first-jpg
Actual vendor UI capture · Open full screenshot

Original expected-behavior review

failed: Failed 3/6 conditions: body opacity, handle-hole transparency and source-color retention. IoU 0.958243 passes, but body retention 92.0616% and hole transparency 89.2258% miss their independent 95% thresholds. Native provenance, PNG/alpha and silhouette passed.

Recorded duration: Not recorded

Acceptance conditions

  • passed: The exact original input hash is unchanged, only that RGB image was imported once through the actual official BgSub Open Image control, and the retained evidence proves its first native automatic background-removal result followed by its first unedited PNG downloaded through the actual native Save control (or available Save Image control) and download-dialog Save link. Record current labels; Save PNG is not a literal label established by the static source. No reference mask, previous product output, manual repair, FullSize/autoFullSize, login/credit path, contribution/upload/share action, batch processing or unrelated external action was supplied/performed. The second frozen source used the native Open Image local-file chooser once in the same task session. Operator action/observation and download records retain automatic first processing, zero edits/retries and one Save Image > Save download. Input and output SHA-256 match retained bytes; no reference mask was supplied.
  • passed: The actual first downloaded native artifact is decodable PNG, exactly 512x512, with a stored alpha channel or PNG transparency data; decoded alpha has minimum <=10 and maximum >=245. Checkerboard visibility alone does not pass. The first native export is a decodable 512x512 RGBA PNG (33801 bytes), with stored alpha ranging from 0 to 255.
  • passed: Using every image pixel, alpha>=128 foreground has IoU>=0.90 against the full frozen mug reference including its handle and actual hole. Foreground intersection 42385 / union 44232 gives IoU 0.9582429010671007, meeting the fixed 0.90 minimum. This does not override separate body, hole and source-color failures.
  • failed: At least 95% of mug-body-core and at least 90% of mug-handle-core have alpha>=245; both regional fractions are separately reported. Mug-body opaque retention is 32425 / 35221 = 92.0615541864%, below 95%. Handle retention is 4069 / 4069 = 100%, meeting 90%; the combined condition fails because both regions are required.
  • failed: At least 95% of the inset handle-hole and at least 95% of exterior-background have alpha<=10; both regional fractions are separately reported. Handle-hole transparency is 2766 / 3100 = 89.2258064516%, below 95%. Exterior transparency is 213267 / 213601 = 99.8436336909%, meeting 95%; the combined condition fails because both regions are required.
  • failed: In every pre-frozen source-RGB group within each foreground reference core, alpha>=245 coverage meets that region's fixed minimum fraction (95% bottle/body, 90% mug handle). Of those retained opaque pixels, at least 99% differ from the source RGB by at most 2 in every channel. All per-group numerators and denominators are reported; zero opaque pixels cannot pass. Transparent-pixel hidden RGB outside these cores is not scored. Dark body stripe retains 1368 / 1708 = 80.0936768150% opaque pixels; near-white body retains 31057 / 33513 = 92.6715006117%, both below 95%. Near-white handle retains 4069 / 4069 = 100%, meeting 90%. RGB matching among retained opaque pixels is 100% for each group, but inadequate retained source-color coverage fails this condition.

Limits of this execution

  • Two fixed 512x512 synthetic geometric images and one first native automatic removal per source. These results do not establish natural-photo quality, overall accuracy, repeatability, a product ranking or the full editing feature set.
  • The user explicitly approved the by-use Terms before the first image import. The original frozen contract retains historical pending-authorization fields; later approval is stored separately rather than editing that pre-inference record.
  • Both inputs used the native Open Image local file chooser, followed by automatic processing and the first Save Image > Save PNG download in the same task session. No URL import, manual repair, mask submission, FullSize/autoFullSize, login, credit purchase, contribution, sharing, batch operation or quality rerun was used.
  • The native Web App initial import screen displays version:220907, retained only as a UI-displayed label. This does not identify a model, inference runtime or backend release. Backend model/provider/build, token usage and actual network processing route were not observed. Browser use does not establish wholly local/private or offline processing.
  • No checkout or account upgrade was performed. Per-run charges, credit debits, hardware, electricity and total operating cost were not measured; total cost remains unknown.
  • startedAt records the approved session boundary and completedAt the final observed download receipt. Inference start/completion was not timed; durationMs remains unknown for both cases.
  • A separate Codex agent reviewed saved evidence and independently recomputed pixel measurements. No independent human review, hidden-server instrumentation or full network/host-filesystem audit was performed.
  • Public receipts omit only private absolute paths and explicitly record their projection. Native PNGs, input/reference images, frozen contract, measurement script and visible UI captures remain unmodified.

Download the product execution record (JSON) →

Test plans and preparation · 0 original cases unexecuted

Dependencies before a product pilot

  • The frozen BgSub contract, exact two 512x512 RGB inputs, evaluator-only masks and measurement script are retained unchanged. The native pilot completed on 25 October 2026 after explicit user Terms approval: one local Open Image import, automatic first result and first Save Image > Save PNG per source in the same task session. Original case steps retain preparation-stage wording; current authorization, outputs and failed conditions are recorded separately. No mask was supplied to the product and no repair, alternate engine, FullSize/account/credits, contribution, batch or quality rerun occurred.
  • On 25 October 2026, after the user explicitly approved the by-use Terms, two unchanged synthetic 512x512 inputs were each imported once using native Open Image. Automatic first results and the first Save Image > Save PNGs were retained; both cases failed 3/6 conditions. No URL input, manual repair, retry, FullSize, login, credits, purchase or optional contribution was used. Backend identity, network isolation and total cost remain unknown.
  • Confirm free processing advertised · current route entitlements unverified against the current vendor terms; usage and connected-service costs can affect the pilot.
  • Create a test workspace or use public/authorized material. Keep an input baseline, output artifact and action log for comparison.
First native BgSub bottle cutout and original cap/label preservation Product case · executed (failed)

Controlled input

One existing authorized synthetic 512x512 RGB bottle image: teal body, black cap, cream label with three dark marks, light-gray background. This is a finite geometric fixture, not a natural product photograph. Independent full/core/background reference masks remain evaluator-only.

Request

In the actual official BgSub browser UI, use Open Image to import only the frozen bottle RGB image once. Preserve the first native automatic background-removal result and download it through native Save/Save Image > Save as the first unedited PNG at original 512x512. Retain the complete teal bottle, black cap, cream label and three dark label marks, and make the exterior background truly transparent. Do not change a background, draw or supply a mask, repair/edit pixels, resize/move the subject, use FullSize/autoFullSize/login/credits/contribution/batch or retry.

Steps

  1. Before any native case, Root records the actual current UI/defaults and action-time approval for importing the two specified public synthetic images and accepting the by-use Terms. Verify this complete frozen contract, all input/reference hashes and measurement-script hash. Keep evaluator references, previous product outputs and metrics outside the product; do not open a sample image or run an unrecorded trial.
  2. Use a fresh task-owned official BgSub browser session/state at https://bgsub.com/webapp/. Record visible UI/version/build or mark undisclosed values unknown. Keep ordinary small-image/default transparent processing; do not enable FullSize/autoFullSize, log in, use credits, choose URL/clipboard/sample input, alter settings or enter any optional contribution/share path. Record page initialization/model-parameter loading separately; an initialization or gate does not establish a case result.
  3. Choose the native Open Image control and import only this exact frozen 512x512 RGB source once. Local image import invokes the native automatic removal path; there is no Photopea Select > Remove BG command here. Retain the first native automatic response, status/error/gate and visible Original/Processed state. Do not import again, restart the processing for quality, switch engine, manually erase/restore/refine/colour/transform/filter/enhance or supply a mask.
  4. After that same original operation completes, use the actual native Save control (the desktop DownloaderBox may also offer Save Image), then the Save link in its Save image download dialog. Download the first unedited native PNG at the requested original 512x512 dimensions/default transparent background, and verify the actual file rather than assuming source dimensions are retained. Record exact current control labels, actual layout/platform, download identity, screenshots, export dimensions/hash and any failure. Width-based layout and Quasar mobile export detection are different conditions; do not switch environment between cases. Do not use FullSize, a second export, screenshot-only substitution, re-encoding or output repair.
  5. Keep the first downloaded bytes immutable at the case fixed result path. Only after native provenance/authorization is retained may the frozen pixel script measure those bytes against the unchanged evaluator-only masks and source-RGB groups. Report every individual condition and failed/unknown region; no alignment, resizing, mask recomputation, threshold tuning or reading/comparing earlier product output/metrics is allowed.

Expected output

A real first unedited native BgSub PNG download, exactly 512x512 with stored transparency, complete original bottle/cap/label/detail colors and all six fixed checks satisfied. Actual failures and unknowns remain visible.

Observable pass conditions

  • The exact original input hash is unchanged, only that RGB image was imported once through the actual official BgSub Open Image control, and the retained evidence proves its first native automatic background-removal result followed by its first unedited PNG downloaded through the actual native Save control (or available Save Image control) and download-dialog Save link. Record current labels; Save PNG is not a literal label established by the static source. No reference mask, previous product output, manual repair, FullSize/autoFullSize, login/credit path, contribution/upload/share action, batch processing or unrelated external action was supplied/performed.
  • The actual first downloaded native artifact is decodable PNG, exactly 512x512, with a stored alpha channel or PNG transparency data; decoded alpha has minimum <=10 and maximum >=245. Checkerboard visibility alone does not pass.
  • Using all 262144 pixels without alignment, cropped exclusions or ignored edges, alpha>=128 foreground has IoU>=0.90 against the full frozen bottle reference mask.
  • At least 95% of all frozen bottle-core pixels retain alpha>=245.
  • At least 95% of the frozen exterior-background pixels have alpha<=10.
  • In every pre-frozen source-RGB group within each foreground reference core, alpha>=245 coverage meets that region's fixed minimum fraction (95% bottle/body, 90% mug handle). Of those retained opaque pixels, at least 99% differ from the source RGB by at most 2 in every channel. All per-group numerators and denominators are reported; zero opaque pixels cannot pass. Transparent-pixel hidden RGB outside these cores is not scored.

Failure conditions

  • The source, reference masks, thresholds or measurement code change after the first product call; the operator supplies a mask, repairs/edits output, uses another engine/product, re-imports/restarts/removes again for quality, enters an optional FullSize/login/credit/contribution/batch path, or substitutes an old output. This makes the run nonconforming, not passed.
  • The first native result or PNG download fails a fixed dimension/alpha/IoU/region/color condition. Preserve the actual original artifact/error and each failed check; an overall average cannot hide a region failure or replace the first output.
  • Missing native output, pending service work, unverified current UI/provenance or uncertain action-time authorization remains explicitly unexecuted/incomplete/unverified. A wait timeout alone is not terminal failure and must not trigger another import/removal. Do not invent an image, successful download or pass score.
First native BgSub low-contrast mug, narrow handle and real hole boundary Product case · executed (failed)

Controlled input

The unchanged distinct synthetic 512x512 RGB near-white mug on light-gray background, with a narrow handle, real background-colored handle hole and dark body stripe. The full/body/handle/hole/exterior masks existed before any Photopea call and remain evaluator-only.

Request

In a fresh source case in the same official BgSub task session, use Open Image to import only the frozen low-contrast mug RGB image once. Preserve its first native automatic background-removal result and first unedited native PNG download through Save/Save Image > Save at original 512x512. Retain the near-white body, dark stripe and thin handle, while removing background visible through the real handle hole and outside the mug. Do not redraw/repair the handle, fill the hole, change contrast/background, supply a mask, resize/move/filter/enhance, use FullSize/autoFullSize/login/credits/contribution/batch or retry.

Steps

  1. Before any native case, Root records the actual current UI/defaults and action-time approval for importing the two specified public synthetic images and accepting the by-use Terms. Verify this complete frozen contract, all input/reference hashes and measurement-script hash. Keep evaluator references, previous product outputs and metrics outside the product; do not open a sample image or run an unrecorded trial.
  2. Use a fresh task-owned official BgSub browser session/state at https://bgsub.com/webapp/. Record visible UI/version/build or mark undisclosed values unknown. Keep ordinary small-image/default transparent processing; do not enable FullSize/autoFullSize, log in, use credits, choose URL/clipboard/sample input, alter settings or enter any optional contribution/share path. Record page initialization/model-parameter loading separately; an initialization or gate does not establish a case result.
  3. Choose the native Open Image control and import only this exact frozen 512x512 RGB source once. Local image import invokes the native automatic removal path; there is no Photopea Select > Remove BG command here. Retain the first native automatic response, status/error/gate and visible Original/Processed state. Do not import again, restart the processing for quality, switch engine, manually erase/restore/refine/colour/transform/filter/enhance or supply a mask.
  4. After that same original operation completes, use the actual native Save control (the desktop DownloaderBox may also offer Save Image), then the Save link in its Save image download dialog. Download the first unedited native PNG at the requested original 512x512 dimensions/default transparent background, and verify the actual file rather than assuming source dimensions are retained. Record exact current control labels, actual layout/platform, download identity, screenshots, export dimensions/hash and any failure. Width-based layout and Quasar mobile export detection are different conditions; do not switch environment between cases. Do not use FullSize, a second export, screenshot-only substitution, re-encoding or output repair.
  5. Keep the first downloaded bytes immutable at the case fixed result path. Only after native provenance/authorization is retained may the frozen pixel script measure those bytes against the unchanged evaluator-only masks and source-RGB groups. Report every individual condition and failed/unknown region; no alignment, resizing, mask recomputation, threshold tuning or reading/comparing earlier product output/metrics is allowed.

Expected output

A genuine first unedited native BgSub PNG download, exactly 512x512 with real hole/exterior transparency, retained low-contrast body/dark stripe/narrow handle and all six fixed checks satisfied. Lost handle, opaque hole or changed source color remains a separately reported failure.

Observable pass conditions

  • The exact original input hash is unchanged, only that RGB image was imported once through the actual official BgSub Open Image control, and the retained evidence proves its first native automatic background-removal result followed by its first unedited PNG downloaded through the actual native Save control (or available Save Image control) and download-dialog Save link. Record current labels; Save PNG is not a literal label established by the static source. No reference mask, previous product output, manual repair, FullSize/autoFullSize, login/credit path, contribution/upload/share action, batch processing or unrelated external action was supplied/performed.
  • The actual first downloaded native artifact is decodable PNG, exactly 512x512, with a stored alpha channel or PNG transparency data; decoded alpha has minimum <=10 and maximum >=245. Checkerboard visibility alone does not pass.
  • Using every image pixel, alpha>=128 foreground has IoU>=0.90 against the full frozen mug reference including its handle and actual hole.
  • At least 95% of mug-body-core and at least 90% of mug-handle-core have alpha>=245; both regional fractions are separately reported.
  • At least 95% of the inset handle-hole and at least 95% of exterior-background have alpha<=10; both regional fractions are separately reported.
  • In every pre-frozen source-RGB group within each foreground reference core, alpha>=245 coverage meets that region's fixed minimum fraction (95% bottle/body, 90% mug handle). Of those retained opaque pixels, at least 99% differ from the source RGB by at most 2 in every channel. All per-group numerators and denominators are reported; zero opaque pixels cannot pass. Transparent-pixel hidden RGB outside these cores is not scored.

Failure conditions

  • The source, reference masks, thresholds or measurement code change after the first product call; the operator supplies a mask, repairs/edits output, uses another engine/product, re-imports/restarts/removes again for quality, enters an optional FullSize/login/credit/contribution/batch path, or substitutes an old output. This makes the run nonconforming, not passed.
  • The first native result or PNG download fails a fixed dimension/alpha/IoU/region/color condition. Preserve the actual original artifact/error and each failed check; an overall average cannot hide a region failure or replace the first output.
  • Missing native output, pending service work, unverified current UI/provenance or uncertain action-time authorization remains explicitly unexecuted/incomplete/unverified. A wait timeout alone is not terminal failure and must not trigger another import/removal. Do not invent an image, successful download or pass score.

Permissions and failure boundary

  • Documented access: Official hosted browser webapp. Reviewed source routes an ordinary local file through FileReader into a browser worker; page initialization may load processing resources. Self-hosting, complete offline operation, current model identity and network isolation have not been established.; Web app. Confirm the actual scopes for the selected account and plan.
  • Acceptance boundary: A genuine first unedited native BgSub PNG download, exactly 512x512 with real hole/exterior transparency, retained low-contrast body/dark stripe/narrow handle and all six fixed checks satisfied. Lost handle, opaque hole or changed source color remains a separately reported failure.
  • Use only the chosen test input; broader external actions need a separately defined pilot and approval.
Source access and directory page checks
Product performancePublic vendor UI product test · 2 cases executed

2 of 2 defined cases have actual product execution records. Inspect each outcome, access method, inputs and limits below.

Official-source access14 of 14 URLs checked

Current HTTP/readability checks are listed below. They establish access, not the truth of every vendor claim.

uAgentKit profilePage checks passed

Checked 2026-10-06T01:30:33.049Z. Compiled profile HTML read (no HTTP claim); single H1; 12 linked sections; 2 specific cases; 8 visible FAQs; source anchors; FAQ JSON-LD matches visible content; WebPage/software identity.

Official-page checks

Page accessibility checks for BgSub; these are separate from product performance testing.
SourceAccess statusEvidence and scope
Official BgSub background-removal overview and linked blogaccessibleHTTP 200 · 2026-10-02T20:46:17.900272+00:0032120 source bytes. Retained official public HTML bytes and derived readable content were reviewed. Vendor statements and dated historical attribution are not independently measured product results or current company-size certification. Original retrieval time is preserved; 2026-10-03 is the profile editing date.
Official BgSub pages entry shell; full About and policies are in separately retained route chunkslimitedHTTP 200 · 2026-10-02T20:46:56.081456+00:002003 source bytes. Retained official entry shell only. HTTP 200 establishes entry availability, not rendered policy/About text, live controls or inference. Route content is supported by the separately retained official chunks. Original retrieval time is preserved; 2026-10-03 is the profile editing date.
Official BgSub native webapp entry shell; not a product resultlimitedHTTP 200 · 2026-10-02T20:46:56.124867+00:002814 source bytes. Retained official entry shell only. HTTP 200 establishes entry availability, not rendered policy/About text, live controls or inference. Route content is supported by the separately retained official chunks. Original retrieval time is preserved; 2026-10-03 is the profile editing date.
Official pages bundle: FAQ, free processed-image statement, maximum size and policy route loaderaccessibleHTTP 200 · 2026-10-02T21:31:45.282269+00:0012521 source bytes. Retained official JavaScript bytes read as inert source. Content and loader/code statements were reviewed; this is not JavaScript execution, a live UI observation, current model identity or a product-quality result. Original retrieval time is preserved; 2026-10-03 is the profile editing date.
Official webapp route loader: native page and layout chunk identityaccessibleHTTP 200 · 2026-10-02T21:31:45.283131+00:0011893 source bytes. Retained official JavaScript bytes read as inert source. Content and loader/code statements were reviewed; this is not JavaScript execution, a live UI observation, current model identity or a product-quality result. Original retrieval time is preserved; 2026-10-03 is the profile editing date.
Official-home-linked BgSub blog and public author attributionaccessibleHTTP 200 · 2026-10-02T21:31:45.283796+00:0075262 source bytes. Retained official public HTML bytes and derived readable content were reviewed. Vendor statements and dated historical attribution are not independently measured product results or current company-size certification. Original retrieval time is preserved; 2026-10-03 is the profile editing date.
Official About route text: mission and vision without named legal operatoraccessibleHTTP 200 · 2026-10-02T21:32:34.316283+00:003511 source bytes. Retained official JavaScript bytes read as inert source. Content and loader/code statements were reviewed; this is not JavaScript execution, a live UI observation, current model identity or a product-quality result. Original retrieval time is preserved; 2026-10-03 is the profile editing date.
Official Terms full static route text: by-use conditions, material license and logo-linking restrictionaccessibleHTTP 200 · 2026-10-02T21:32:34.317084+00:0011844 source bytes. Retained official JavaScript bytes read as inert source. Content and loader/code statements were reviewed; this is not JavaScript execution, a live UI observation, current model identity or a product-quality result. Original retrieval time is preserved; 2026-10-03 is the profile editing date.
Official English Privacy full static route text: logs, cookies, advertising and contact/account informationaccessibleHTTP 200 · 2026-10-02T21:32:34.317553+00:009261 source bytes. Retained official JavaScript bytes read as inert source. Content and loader/code statements were reviewed; this is not JavaScript execution, a live UI observation, current model identity or a product-quality result. Original retrieval time is preserved; 2026-10-03 is the profile editing date.
Official 2022-08-21 article: independently developed origin and historical implementationaccessibleHTTP 200 · 2026-10-02T21:32:34.998374+00:0088250 source bytes. Retained official public HTML bytes and derived readable content were reviewed. Vendor statements and dated historical attribution are not independently measured product results or current company-size certification. Original retrieval time is preserved; 2026-10-03 is the profile editing date.
Public Blogger profile linked by official BgSub blog: DeepKnight aliasaccessibleHTTP 200 · 2026-10-02T21:32:35.005292+00:00521851 source bytes. Retained official public HTML bytes and derived readable content were reviewed. Vendor statements and dated historical attribution are not independently measured product results or current company-size certification. Original retrieval time is preserved; 2026-10-03 is the profile editing date.
Official native image import, worker, account/credit, optional contribution and PNG download implementationaccessibleHTTP 200 · 2026-10-02T21:33:16.326190+00:00121657 source bytes. Retained official JavaScript bytes read as inert source. Content and loader/code statements were reviewed; this is not JavaScript execution, a live UI observation, current model identity or a product-quality result. Original retrieval time is preserved; 2026-10-03 is the profile editing date.
Official native webapp layout module retained as inert static sourceaccessibleHTTP 200 · 2026-10-02T21:33:16.327252+00:00887 source bytes. Retained official JavaScript bytes read as inert source. Content and loader/code statements were reviewed; this is not JavaScript execution, a live UI observation, current model identity or a product-quality result. Original retrieval time is preserved; 2026-10-03 is the profile editing date.
Official browser-worker source: ordinary image route and small-image Authorization branchaccessibleHTTP 200 · 2026-10-02T21:33:44.172974+00:0060963 source bytes. Retained official JavaScript bytes read as inert source. Content and loader/code statements were reviewed; this is not JavaScript execution, a live UI observation, current model identity or a product-quality result. Original retrieval time is preserved; 2026-10-03 is the profile editing date.

Evidence

What “official sources” means We read vendor material for the claims cited below. This is a documentation review. No independent product test or professional endorsement is implied. Read our method →

Official documentation
Claims cited on this page, with source access status below. URL accessibility is separate from a substantive claim review.
Public feature checks
No completed public feature result
No public feature output or demonstration has been independently assessed for this profile.
uAgentKit product execution
Public vendor UI product test · 2 cases executed
Public vendor UI product test · 2 cases executed. 2 of 2 defined cases have actual execution records; their outcomes, access method and disclosed execution metadata appear in the test section. Two original frozen 512x512 synthetic BgSub cases, one native Open Image automatic removal and first Save Image > Save PNG export per case, with all six unchanged acceptance conditions. Both original cases failed, each passing 3/6 conditions. Bottle IoU 0.838686 and label opaque retention 0.433129% fail. Mug IoU 0.958243 passes, while body opaque retention 92.0616%, hole transparency 89.2258% and source-color retention fail. First native outputs are retained unchanged.
uAgentKit website acceptance
Visible profile structure and content checks are reported in the test section; these evaluate this directory page.
Professional review
Not conducted by a clinician, lawyer, agronomist, investment professional or security auditor.

Commercial use: The official FAQ says users own rights to images they process with BgSub. This vendor statement does not supply missing rights to an input image or certify every commercial use. The Terms separately restrict use of website material and require a trademark license for logo/artwork linking. uAgentKit uses an existing generic icon and ordinary source links without implying endorsement. Review current image-processing, account and export conditions for the intended route.

Limitations and checks

  • Both original fixed 512x512 cases were executed on 25 October 2026 and failed, each passing 3/6 conditions. Bottle IoU is 0.838686 and cream-label opaque retention is 0.433129%. Mug IoU 0.958243 passes, but body opacity 92.0616%, hole transparency 89.2258% and source-color retention fail. The first PNGs were not repaired or replaced.
  • The finite geometric bottle and low-contrast mug cases do not establish natural-photo quality, hair/fur performance, complex scenes, broad accuracy, processing speed or a product ranking.
  • The 2022 independent-development statement and DeepKnight alias establish historical attribution only. Current real name, legal operator, staffing and controlling ownership remain unknown.
  • The observed controls were Open Image and Save Image > Save; both cases used the same ordinary task session and produced 512x512 PNGs with stored alpha. Exact model/build, inference runtime, other layout/platform behavior and full-size defaults remain unverified. Save PNG was not the literal button label.
  • Import can invoke automatic removal. For the frozen cases preserve the first result and first native download; manual repair, alternate engines, repeated imports, quality reruns and second exports would change the evaluated workflow.
  • Thin handles, internal holes, low-contrast edges, labels and source colors need individual review. Good overall appearance or an average score cannot hide loss of a required region.
  • The FAQ free and maximum-size claims do not establish universal account-free access or all-size export entitlement. FullSize, autoFullSize, account and credit paths remain outside the frozen pilot.
  • Ordinary browser processing is documented by source and a vendor no-upload statement, but current network behavior has not been audited. Optional Contribute uploads an image through a distinct path.
  • Website logs, IP/browser details, cookies, advertising and possible account/contact information are described separately by Privacy. Do not interpret local image processing as whole-site zero collection or guaranteed offline isolation.
  • The reviewed FAQ describes batch/developer access as coming soon. No public inference endpoint, store integration, autonomous publishing or business-action execution was established.
  • Input-image and third-party rights remain necessary. The processed-image FAQ does not grant permission to reuse BgSub website material, logo or artwork; the Terms require a trademark license for logo/artwork linking.

Field-level unknowns identify gaps in this review. They do not imply the vendor lacks the capability.

Alternatives and comparisons

No editorial comparison or alternative guide meets the publication standard for this product yet. Build an instant fact comparison.

Questions about BgSub

What is BgSub and who is it useful for?

BgSub is a specialist browser image tool for automatic background removal and subject cutouts. An independent seller can prepare an authorized product cutout for a listing; a creator can isolate a subject for a thumbnail or composite; an individual can clean a simple background before reuse. The reviewed sources also document optional image adjustments, but this profile focuses on preserving the first automatic cutout. Inspect the outline, small parts, labels and colors, then check the native PNG. The tool does not establish autonomous publication, store management or actual commercial results.

Is BgSub an independently created project?

An official blog article dated 2022-08-21 explicitly calls BgSub an independently developed project. The homepage links that blog, whose public author and linked Blogger profile use the alias DeepKnight. This supports a dated independently developed origin and public author attribution. It does not establish a current one-person company, the author's real name, the legal operating entity, present headcount, financing or controlling ownership. uAgentKit groups the profile under BgSub project so the company page describes the documented project relationship without certifying an unverified corporate identity.

Is BgSub free, and can it export a full-size image?

The retained FAQ advertises free processing, processed-image rights and support up to 4096x4096, while native source also contains account, credit and FullSize routes. The 25 October 2026 pilot obtained two ordinary 512x512 first PNGs without login, purchase or FullSize. Both failed the fixed quality conditions. This establishes that limited observed route, not universal free entitlement, current quotas or full-size export rights. Per-run charge/credit debit and total operating cost were not measured.

How do I make and save a cutout with BgSub?

Read the current Terms, preserve an authorized source image and inspect current size/account conditions. In the 25 October 2026 pilot, native Open Image imported each local source once and automatic processing produced the first result. The observed export path was Save Image, then Save in the Save image dialog. Inspect the untouched PNG for dimensions, real alpha transparency, labels, narrow parts and internal holes. No optional editing or second export was used; those features remain separate from these first-result cases.

Has uAgentKit tested BgSub?

BgSub: 2/2 defined cases completed. Latest completed result per original case: 0 passed, 2 failed, 0 partial. Recorded scope: official public vendor UI. Completion dates (UTC): 2026-10-25. Hidden backend calls and server actions were not observed in the public UI scope. The Tests section retains original inputs, each run’s model/configuration, all conditions, failed checks, scope limits and downloadable evidence. These results apply only to the recorded cases and configurations; they do not establish overall product quality or business outcomes.

Does BgSub upload my image or collect website data?

The official historical blog describes ordinary processing in the browser without uploading the source image, and native source shows a local FileReader-to-worker route. That is source support plus a vendor statement; current network isolation has not been audited. Contribute is a separate optional image-upload path for algorithm improvement, with upload-information and file-upload requests. The Privacy text separately describes IP/browser/ISP/time/referrer logs, cookies, advertising and possible contact/account information. Keep ordinary image processing, optional contribution and website data conditions separate. None of these sources prove every feature is offline, the whole site collects nothing or all retention is zero.

What should I review before using a BgSub product cutout?

Compare the result with the untouched source: retain the full subject, product proportions, cap or narrow handle, label details and original colors. Inspect background inside real holes as well as outside the subject, and verify stored PNG alpha instead of relying on a checkerboard. The frozen finite cases require 512 x 512 dimensions, silhouette IoU at least 0.90, fixed foreground/background region thresholds and source-color checks; a failed handle or hole remains a separate failure. These geometric checks do not validate natural photographs or broad quality. Review input rights, current export conditions and listing requirements before publishing the image.

Does BgSub have a public API, batch workflow or autonomous agent?

The reviewed official FAQ labels API and batch processing as coming soon. Static account and credit requests do not establish an available public image-inference API. The documented native route is a supervised image tool; background edits, transforms, filters and AI Coloring are separate optional controls, while a store connection, autonomous publishing, scheduled batch execution and external business actions were not established. Historical implementation names and processing-time claims do not identify the current model or establish measured speed. Evaluate any newly offered endpoint, account feature or automation from its current official documentation and actual behavior.

Sources and change history

  1. Official BgSub background-removal overview and linked blog

    BgSub project · bgsub.com · Read · 2026-10-02

  2. Official BgSub pages entry shell; full About and policies are in separately retained route chunks

    BgSub project · bgsub.com · Limited accessible content · 2026-10-02

  3. Official BgSub native webapp entry shell; not a product result

    BgSub project · bgsub.com · Limited accessible content · 2026-10-02

  4. Official pages bundle: FAQ, free processed-image statement, maximum size and policy route loader

    BgSub project · bgsub.com · Read · 2026-10-02

  5. Official webapp route loader: native page and layout chunk identity

    BgSub project · bgsub.com · Read · 2026-10-02

  6. Official-home-linked BgSub blog and public author attribution

    DeepKnight / official BgSub blog · bgsub.blogspot.com · Read · 2026-10-02

  7. Official About route text: mission and vision without named legal operator

    BgSub project · bgsub.com · Read · 2026-10-02

  8. Official Terms full static route text: by-use conditions, material license and logo-linking restriction

    BgSub project · bgsub.com · Read · 2026-10-02

  9. Official English Privacy full static route text: logs, cookies, advertising and contact/account information

    BgSub project · bgsub.com · Read · 2026-10-02

  10. Official 2022-08-21 article: independently developed origin and historical implementation

    DeepKnight / official BgSub blog · bgsub.blogspot.com · Read · 2026-10-02

  11. Public Blogger profile linked by official BgSub blog: DeepKnight alias

    DeepKnight / official BgSub blog · www.blogger.com · Read · 2026-10-02

  12. Official native image import, worker, account/credit, optional contribution and PNG download implementation

    BgSub project · bgsub.com · Read · 2026-10-02

  13. Official native webapp layout module retained as inert static source

    BgSub project · bgsub.com · Read · 2026-10-02

  14. Official browser-worker source: ordinary image route and small-image Authorization branch

    BgSub project · bgsub.com · Read · 2026-10-02

· Executed both original 512x512 synthetic cases once after explicit Terms approval. Both failed, each passing 3/6 fixed conditions: the bottle lost most of its cream label and the mug missed body-opacity and handle-hole-transparency thresholds despite passing overall silhouette IoU. Retained original cases, first automatic results, native PNGs, evaluator-only masks and exact failed conditions. No login, purchase, FullSize, contribution upload, manual repair or quality rerun; model, inference runtime and total cost remain unknown. Historical official-source check dates are unchanged.

Suggest a sourced correction →