US 12288421

Search results

• USPTO / Google Patents / Justia hits: US 12288421 resolves unambiguously to US12288421B2 — "Optical skin detection for face unlock" (TrinamiX GmbH family). No conflicting numbers surfaced; I did not auto-correct any identifier.
• CAFC 2026 docket search: Searches for "12288421" CAFC, "12288421" appeal, and TrinamiX litigation queries returned no CAFC (or other court) docket results for this patent. I found no evidence of any appeal, IPR, or litigation involving patent 12288421 as of the current date.
• Sources: patents.google.com/patent/US12288421B2, patents.justia.com/patent/12288421, patentimages.storage.googleapis.com (family publication US20240037201A1 PDF), patentguru/patentleaderboard index pages.

Patent summary

Abstract (as indexed for this patent family; wording is approximate because the abstract block was not fully captured in my primary source — the specification and claim summary confirm the substance): A method for face authentication is proposed. The method includes at least one face detection step including determining at least one first image with a camera, determining at least one material property from a second image, where the second image is recorded while projecting at least one illumination pattern including a plurality of illumination features on the scene, and at least one authentication step including authenticating the detected face using the face detection and the material property. (If you need the verbatim abstract text, it should be confirmed against the USPTO Patent Center record.)

Independent claims — plain-language overview

Caveat: The full claims section was not retrievable in the materials I obtained (the Google Patents text I worked from contained the specification but was truncated before the claims; web snippets confirmed the "Summary of the Invention" but not the granted claim set verbatim). The overview below is therefore reconstructed from the specification's independent-claim summary, which closely mirrors the claims. Exact claim numbering and boundaries should be verified against the USPTO full-text claims.

Claim 1 — Method for face authentication (four required steps):

• (a) Face detection step: Capture at least one 2D image of a scene with a camera and, using a processing unit, detect a face in that image by identifying predefined/determined geometrical features characteristic of faces (e.g., eyes, nose, mouth).
• (b) Skin detection step: Project an illumination pattern made of multiple illumination features (e.g., NIR laser spots) onto the scene using an illumination unit; capture a second ("reflection") image with the same camera; for at least one reflection feature located in the image region corresponding to the detected face, analyze its beam profile to derive first beam-profile information and determine a material property (e.g., sub-surface scattering/translucency versus surface-only reflection). The face is characterized as skin only if the material property matches a property characteristic of skin — this defeats photos/screens and masks made of non-skin materials.
• (c) 3D detection step: Analyze the beam profiles of at least four reflection features inside the face region to derive second beam-profile information (e.g., depth-from-photon-ratio) and determine a depth level; the face is characterized as a 3D object if the depth level deviates from a predefined depth level of plane objects — this defeats flat 2D photograph/video presentation attacks.
• (d) Authentication step: Authenticate the detected face (via an authentication unit) only if the face was characterized as skin (b) and a 3D object (c) — i.e., a two-factor anti-spoof check (material + 3D geometry) before biometric verification/device unlock.

Likely additional independent claims (aspects recited in the specification; exact text unverified):

• Mobile device configured to carry out the method (e.g., smartphone with camera, illumination unit projecting the pattern, and processing unit/authentication unit securing device access).
• Computer program / computer-readable medium with instructions causing a computer to perform the method.
• Possibly a use claim (e.g., use of the method for face authentication / presentation-attack detection), based on the specification's statement that the invention "relates to a method for face authentication, a mobile device and various uses of the method."

Uncertainties to flag: (1) verbatim abstract and full independent claim set were not confirmed against USPTO text; (2) no CAFC or district-court docket activity was found for this patent, but that is a negative result from general web searches, not an exhaustive court-docket query; (3) the related continuation US12530925B2 (same title, filed 2025-03-24 as US 19/088,110) appears in the same family but is a different patent number and was not analyzed here.

I'll search multiple sources for litigation involving US Patent 12288421.

The initial searches surfaced news of a trinamiX (BASF subsidiary) v. Apple suit filed in the Western District of Texas (Midland) around September 3, 2026, alleging infringement of 7 face-authentication patents. The articles don't name the specific patent numbers, so I need to verify whether US 12288421 is among the asserted patents. Let me dig into the docket and case details.

I've reached the search-step limit, so I'll answer based on the information retrieved, flagging clearly what is and is not confirmed.

Litigation involving US Patent 12288421 — Findings

Date discrepancy note (important)

Your task states "Current Date: April 26, 2026," but the live search environment and the retrieved documents are dated September 3–4, 2026 (the news reports I found are from those dates). Per the operating rules, I treat the live search results as ground truth and flag the discrepancy. If your analysis date is truly April 26, 2026, then the case described below would be after that date — but it is the only litigation the live searches surfaced, so I report it.

Known litigation (only matter found)

TrinamiX GmbH v. Apple Inc. — assertion details unconfirmed as to patent 12288421 specifically

• Plaintiff(s): trinamiX GmbH (wholly owned subsidiary of BASF). News reports describe BASF as the plaintiff and identify trinamiX GmbH as the entity bringing the suit.
• Defendant(s): Apple Inc.
• Jurisdiction: U.S. District Court for the Western District of Texas, Midland Division (reported as "U.S. District Court in Midland, Texas" / "德州米德兰").
• Case number: Not retrieved. No docket number appears in any source I obtained.
• Filing date: Reported as September 3, 2026 (local time; some outlets dated it Sept 4 in Asia time zones).
• Asserted patents: Seven (7) trinamiX face-authentication patents. The specific patent numbers are not named in any source I retrieved, so I cannot confirm that US 12288421 is among the seven asserted patents. This is the critical open question.
• Subject matter overlap: The suit is about Apple's Face ID (iPhone 15/16/17 series, iPad Pro) allegedly incorporating trinamiX's patented "material and skin detection" liveness technology (anti-spoofing against photographs, 3D-printed masks, silicone replicas; NIR backscattering/Beam Profile Analysis). That is squarely the technology of US 12288421 ("Optical skin detection for face unlock," assigned to TrinamiX GmbH) — but topical similarity is not proof the patent is asserted.
• Relief sought: Unspecified damages and injunctive relief halting further infringement.
• Current status (as of reporting, Sept 3–4, 2026): Newly filed; Apple had not yet responded or commented per Reuters and follow-on outlets. No answer, no motions, no scheduling orders, and no CAFC appeal reported.

Negative results for other litigation

• No CAFC appeal, IPR/PGR, ITC proceeding, or other district-court case specifically tied to patent number 12288421 was found in any search.
• An earlier search in this analysis (see prior sections) likewise found no court activity for 12288421 — but that predates the September 2026 trinamiX v. Apple filing, so it is now superseded.

Bottom line

• The only known litigation plausibly involving US 12288421 is trinamiX GmbH (BASF) v. Apple Inc., W.D. Tex. (Midland), filed ~Sept 3, 2026.
• I cannot state with confidence that 12288421 is one of the seven asserted patents — no retrieved source lists the patent numbers. To confirm, you would need the actual complaint (PACER/CM/ECF for W.D. Tex., or the court's Public Access to Court Electronic Records), or a docket aggregator (Docket Navigator, Unified Patents portal, RPX) once the case is indexed.
• Suggested verification queries once you have docket access: search W.D. Tex. filings around Sept 3–4, 2026 by party "TrinamiX" and "Apple"; or run the patent number 12288421 against RPX/Docket Navigator litigation databases.

Proceedings overview

There are zero AIA trial proceedings (IPR / PGR / CBM) on file for US 12288421 — no active trials, no petitions, no institutions, no settlements, no appeals. The USPTO Open Data Portal record contains no proceeding, and independent web searches (2026-09-04) surfaced no petition against this patent or any related TrinamiX patent. The defensive posture this gives a defendant is a clean slate: no claim of 12288421 has ever been tested, narrowed, or canceled at the PTAB, and no estoppel under § 315(e)(2) binds any would-be petitioner — but that also means the patent is completely untested and, as of yesterday, its owner is actively asserting the family in district court (see Strategic summary below).

No proceeding-specific sections follow because there are no proceedings. I will not invent numbers, panels, or Final Written Decisions.

---

Strategic summary

Which claims are CANCELED vs. SUSTAINED vs. UNTESTED. All claims of US 12288421 are UNTESTED. No IPR/PGR/CBM petition has been filed against the patent, so no claim has been canceled and no claim has been affirmatively sustained by a PTAB panel. The patent issued 2025-04-29 (priority date 2021-02-18) — it is roughly sixteen months old as of today (2026-09-04), which is within the ordinary one-year post-service window for IPR filings. The absence of PTAB activity is therefore not yet a meaningful signal that the claims are strong; it is equally consistent with the patent simply being young and only now entering active litigation.

Estoppel landscape. Because there are no completed or pending AIA trials, there is no § 315(e)(2) estoppel anywhere. For a defendant facing assertion of 12288421 today, every prior-art ground remains available — there is no petitioner or privy that has already burned § 102/§ 103 grounds in a PTAB proceeding. One practical wrinkle: a defendant that is sued in district court and then files an IPR must still contend with § 315(b)'s one-year bar measured from service of the complaint (the suit was reportedly filed 2026-09-03, so the bar would run to 2027-09-03 for the served defendant), plus the current discretionary-denial climate at the PTAB (Fintiv considerations, and the post-March-2025 Stewart Memo process under which roughly 60% of petitions considered through October 2025 were discretionarily denied).

Pattern signals. There are no pattern signals from PTAB filings — no serial petitioner, no Unified Patents or other defensive aggregator in the chain, no patent-owner PTAB appeals. The important pattern signal is in district court, not the PTAB: on 2026-09-03, trinamiX GmbH (the BASF subsidiary and assignee of 12288421) sued Apple in the [U.S. District Court for the Western District of Texas, Midland Division](/courts/western-district-of-texas-midland), alleging that Face ID in iPhone 15/16/17-series and iPad Pro products infringes seven trinamiX face-authentication patents directed to material/skin detection (per Reuters and contemporaneous reports). Whether US 12288421 is one of the seven asserted patents was not confirmed by the sources I retrieved — the articles name trinamiX and the technology but do not list the patent numbers. If 12288421 is among them, this is the first assertion event for the patent and likely the trigger for any future IPR; if it is not, the patent may still be asserted later or never. Do not represent that 12288421 is or is not in suit until the complaint's exhibit list is checked.

---

Recommended next steps

• There is no FWD to cite and no canceled claim to leverage. If you are a defendant receiving a demand letter citing 12288421, you cannot point to any PTAB disposition — say so plainly to anyone expecting one.
• Confirm whether 12288421 is asserted in the new W.D. Tex. case. Pull the docket for the 2026-09-03 trinamiX v. Apple complaint (W.D. Tex., Midland Division — search PACER/CM/ECF and CourtListener for trinamiX GmbH v. Apple Inc.). The complaint reportedly asserts seven patents on Face ID material/skin detection; 12288421's title and classification (G06V40/40 spoof detection, G06V40/16 face recognition) make it a plausible member of that set, but this must be verified against the complaint's exhibit list, not assumed.
• If you are (or become) the defendant in that suit and 12288421 is asserted: the § 315(b) IPR clock starts at service — mark 2027-09-03 as the hard deadline for any petition. Given the PTAB's current discretionary-denial posture, seriously evaluate a Sotera-type stipulation (stipulating not to pursue in district court the same or reasonably-raisable grounds) to neutralize Fintiv, and consider whether the strongest art is § 102 (single-reference) versus § 103 combinations, since compelling-merits showings weigh against discretionary denial.
• Watch the family, not just this patent. TrinamiX holds a dense family here — the related continuation US12530925B2 (same title, filed 2025-03-24, granted 2026-01-20) and parent publication US20240037201A1 (US 18/264,326). If Apple or another defendant challenges the family, petitions may be filed against multiple members in parallel; grounds and expert positions will be cross-admissible, so any invalidity study should be built family-wide.
• Monitor from 2026-09-03 forward. The Apple suit is one day old as of this analysis. Within the next several months expect either (a) an answer and early motions in Midland, and/or (b) IPR petitions if the asserted set includes these patents. Re-run the ODP PTAB query and a CourtListener search at 30-day intervals — the first petition against 12288421, if it comes, will most likely be filed by Apple or an RPX/Unified-type entity within the next 6–12 months.

---

Caveats on sourcing

• The "no PTAB activity" conclusion rests on (1) the USPTO ODP ingest reported in the structured block, and (2) general web searches run 2026-09-04. I did not run a live PTAB E2E/API query in this session, so treat the negative result as high-confidence but not an exhaustive docket audit.
• The Apple suit details come from Reuters and syndicated coverage dated 2026-09-03/04 (g1.globo.com, investing.com, marketscreener.com, sina/eastmoney aggregations). None of those sources list the seven asserted patent numbers, so 12288421's presence in that case is unconfirmed.
• No proceeding numbers, judge panels, institution decisions, or FWDs are reported above because none exist on the available record — per the operating rules, I have not fabricated any.

Inventors

All six named inventors — Peter Schillen, Benjamin Guthier, Friedrich Schick, Manuel Guenther, Lars Diesselberg, Christian Lennartz — are listed as trinamiX GmbH inventors across the trinamiX face-authentication patent family (the same six appear on the sibling patent US12530925B2, "Optical skin detection for face unlock," granted 2026-01-20, per patentguru assignee records). trinamiX GmbH is headquartered in Ludwigshafen am Rhein, Germany; the inventors' assignments identify trinamiX GmbH as the recipient of their rights.

• Unusual departure pattern: none observed. All six continue to appear on trinamiX filings dated well after this application (e.g., 2023–2024 filings, per patentguru), which is inconsistent with a mass inventor exodus preceding a portfolio sale.

Original assignee

The entity named on the issued patent and on the recorded assignment is TrinamiX GmbH (also spelled trinamiX GmbH), Ludwigshafen, Germany — a wholly owned subsidiary of BASF SE (BASF's own releases state trinamiX "was founded in 2015 as a wholly owned subsidiary of BASF SE"; trinamiX's November 2022 backgrounder likewise says founded 2015 as a 100% BASF SE subsidiary, while the 2026 complaint narrative says BASF "created trinamiX in 2014" — a minor date discrepancy in the sources, same entity).

• Primary line of business: biometric sensing — "trinamiX Face Authentication" (2D face recognition plus liveness detection via beam-profile-analysis-based skin detection) — and mobile NIR spectroscopy. More than 200–240 employees at sites in Germany, China, Japan, Korea and the US; 650+ granted/pending patents.
• Products embodying the claims: yes. trinamiX Face Authentication is a commercialized, certified product line (FIDO Biometric Component Certification, IIFAA, Android Class 3 "strong"), publicly demonstrated integrated behind OLED/foldable displays jointly with TCL CSOT (IFA Berlin 2024, Display Week San Jose), and press coverage states the technology "has been applied in Android smartphones." The patent's claims (optical skin detection using projected NIR illumination patterns and beam-profile analysis) map directly onto this commercialized technology.
• Current status: operating and actively asserting. On 2026-09-03 (one day before the date of this analysis), trinamiX/BASF filed a patent-infringement suit against Apple in the US District Court for the Western District of Texas, Midland Division, alleging Apple's Face ID (iPhone 15/16/17, iPad Pro) infringes seven trinamiX face-authentication patents involving material and skin detection (Reuters, Sept. 3, 2026). The press reports do not enumerate the seven asserted patents, so I cannot confirm US 12288421 is among them.

Assignment timeline

I could not retrieve reel/frame numbers or correspondent-of-record data for this patent: searches of the USPTO Assignment Center indices returned no citable reel/frame entries in the materials available to me (the Assignment Center is not directly machine-queryable here). The chronological record below is therefore reconstructed from the USPTO/Google Patents legal-events feed for US 12288421, which lists one recorded conveyance. Treat reel/frame and correspondent as unverified unless confirmed on the Assignment Center.

• (execution date not shown in available records) / recorded 2025-01-28 — reel/frame not retrievable in my sources
  • Conveyance: Assignment of Assignors' Interest (recorded as "ASSIGNMENT OF ASSIGNORS INTEREST (SEE DOCUMENT FOR DETAILS)")
  • Assignor: Friedrich Schick; Benjamin Guthier; Peter Schillen; Manuel Guenther; Lars Diesselberg; Christian Lennartz (the six named inventors)
  • Assignee: TRINAMIX GMBH
  • Correspondent: not retrievable from my sources. (Context only, not a finding: the prosecution agent of record for sibling family member US12530925B2 is Armstrong Teasdale LLP per patentguru; whether that firm also served as assignment correspondent here is unverified.)
  • Context: standard inventor-to-employer assignment recorded ~3 months before grant (grant 2025-04-29) — normal chain-of-title cleanup, not a portfolio transfer.

No post-issuance assignments exist in the record I could verify. No transfer to any LLC, holding company, licensing vehicle, or defensive aggregator appears for this patent. The original assignee (TrinamiX GmbH) is still the owner.

Timeline diagram

timeline
    title Ownership of US 12288421
    2021 : Priority date claimed
    2024 : US continuation filed
    2025 : Inventors assign to TrinamiX GmbH
         : Patent issued Apr 29
    2026 : TrinamiX and BASF sue Apple over face authentication

NPE / troll-pattern signals

1. Shell-entity transfer — not present. The only recorded conveyance is inventors → TrinamiX GmbH (recorded 2025-01-28). There is no transfer to an "IP / Holdings / Licensing / Ventures" LLC, no registered-agent address, no single-purpose Delaware/Texas LLC anywhere in the chain.
2. Known asserter in the chain — not present. No current or prior assignee matches the public NPE lists cited (Acacia, Marathon, Intellectual Ventures, IPNav, Wi-LAN, Conversant/Mosaid, Vringo, Pendrell, Innovatio, MPHJ, Lumen View, Round Rock, Document Generation Corp., Spangenberg entities, or RPX/Unified high-frequency plaintiffs). The assignee, TrinamiX GmbH, is a BASF SE operating subsidiary, not an entity on those lists.
3. Repeat correspondent across the chain — not present / unclear. Only one recorded conveyance exists, so no recurrence of a correspondent can be established from this patent's own record (recurrence, not a single appearance, is the signal). I could not retrieve the correspondent name from available sources, so this factor is unclear only in the sense that the underlying data was not accessible; nothing suggests a repeat-player NPE filer.
4. Cascading transfers — not present. A single inventor-assignment link, not chained LLC transfers.
5. Pre-litigation transfer — not present. The Apple suit was filed 2026-09-03; the only recorded assignment (2025-01-28) predates it by ~19 months, well outside the 6-month window, and is a routine inventor-to-employer assignment rather than a transfer to an assertion vehicle to set venue or establish standing.
6. Bankruptcy fire-sale — not present. TrinamiX's parent, BASF SE, is the world's largest chemical producer by revenue and is solvent and operating.
7. Privateering — not present. No operating company transferred patents to an NPE to sue competitors. The inverse is true: the operating subsidiary (TrinamiX GmbH, a BASF company) is asserting in its own name against Apple (Reuters, 2026-09-03).
8. Defensive aggregator (anti-NPE) — not present. The chain does not end at RPX, AST, LOT, Unified Patents, or OIN; it terminates at TrinamiX GmbH, which is actively asserting rather than neutralizing.

Verdict

Operating-company assertion.

The ownership record is clean and uncomplicated: a single inventor-to-employer assignment recorded 2025-01-28, with TrinamiX GmbH — the original and current assignee — remaining the owner through grant and today. TrinamiX GmbH is an operating technology subsidiary of BASF SE that develops and commercializes the very face-authentication/skin-detection technology this patent protects (FIDO/IIFAA/Android-certified products demonstrated with TCL CSOT), and it filed suit against Apple on 2026-09-03 in the W.D. Tex. (Midland) over face-authentication patents. No shell LLCs, no cascading transfers, no known-NPE assignees, and no bankruptcy or privateering indicators exist in the record. Two caveats: (1) I could not independently confirm that US 12288421 is one of the seven patents asserted against Apple, since press reports did not enumerate them; and (2) Apple is best characterized as an unlicensed implementer rather than a direct competitor of the B2B component supplier trinamiX — but the asserting entity is nonetheless an operating company practicing the technology, not a patent-acquisition vehicle.

Verification: USPTO Assignment Center search by patent number at https://assignmentcenter.uspto.gov/ (index: https://assignment.uspto.gov/patent/index.html); legal-events record at https://patents.google.com/patent/US12288421/en; Reuters litigation coverage of 2026-09-03 (e.g., unionleader.com / marketscreener.com summaries).

Scope note on what I could and could not retrieve

I ran targeted web searches of USPTO/Google Patents/Justia records for patent 12288421. The identifier resolves unambiguously to US12288421B2 — "Optical skin detection for face unlock" (TrinamiX GmbH; inventors Schillen, Guthier, Schick, Guenther, Diesselberg, Lennartz; granted 2025-04-29; priority 2021-02-18). I did not auto-correct any identifier and found no similar-number confusion.

Important retrieval limitation (stated explicitly): The structured "References Cited" (backward-citation) block that appears on the Google Patents/USPTO front-page record for US12288421B2 was not retrievable through the searches available to me — the query results returned the specification text and family records but not the examiner's citation list. I am therefore not able to certify the complete examiner "Patent Citations" list. What I can provide with high confidence is the set of references affirmatively cited inside the specification itself (which are the applicant-identified references the examiner considered, and which normally appear on the face of the granted patent), plus family members that must be excluded as prior art. I flag each item's evidentiary basis below rather than inventing a citation list.

---

Family members — NOT prior art (do not count under §102)

• US 2024/0331450 A1 (application publication of this very patent) and US 2024/0037201 A1 (parent application US 18/264,326) — same disclosure, earlier publications of this family.
• WO 2022/175386 A1 (PCT/EP2022/053941, filed 2022-02-17, published 2022-08-25) — the PCT counterpart.
• EP 4295252 A1 / B1 / C0, ES 3065568 T3, PL 4546296 T3 — European/regional national-phase counterparts.
• US 12530925 B2 (US 19/088,110, filed 2025-03-24, granted 2026-01-20) — a later-filed continuation in the same family (filed after US12288421B2's priority date), so it cannot be §102 prior art.
• EP 4700712 A2, EP 4479951 B1, EP 4740123 A1, WO 2023/156317 A1 — later trinamiX applications in the broader "face authentication / material data" program; all post-date the 2021-02-18 priority and are not §102 art against this patent.

---

Applicant-cited prior art identified from the specification

The specification cites the following patent documents (verified in the spec text of the family, e.g., the full text mirrored at freepatentsonline.com/y2024/0037201 and patents.justia.com/patent/12288421):

1. US 2019/0213309 A1

(Printed in the patent literally as "US 2019/213309 A1" — I keep the identifier as printed rather than "correcting" it.)

• Type: US pre-grant publication; described in the patent's background section as "a system and method of authenticating a user's face with a ranging sensor."
• Disclosure: A ranging sensor comprising a time-of-flight sensor and a reflectance sensor transmits a signal reflected off a user; the received signal yields (i) a distance between user and sensor and (ii) a reflectance value of the user; a processor activates a facial-recognition process in response to the distance and/or reflectivity.
• Basis for inclusion: Directly quoted and discussed in the specification background (¶[0009] of the family publication).
• §102 analysis vs. the reconstructed method claim 1: Discloses step (a)-type 2D face recognition and a rudimentary "reflectance" gating that overlaps with the idea of using a light-material interaction for liveness. It does not disclose: patterned illumination with a plurality of illumination features; beam-profile analysis of reflection features; classifying skin from a material property (sub-surface scattering/translucency); a 3D-vs-plane depth determination from at least four reflection features; or the conjunctive skin-and-3D authentication gate of step (d). It would not anticipate claim 1 as a whole; at most it anticipates the generic face-detection portion. It is the closest single prior-art document discussed in the patent.

2. WO 2018/091649 A1 (depth-from-photon-ratio "DPR" family)

• Type: PCT publication, trinamiX GmbH (BASF group).
• Disclosure: Cited in the specification for the "depth-from-photon-ratio (DPR) technique" — determining a longitudinal coordinate (distance) of an object by analyzing the beam profile of a reflection feature, e.g., forming a quotient Q of a center region and an edge region of a light spot on a sensor and mapping Q to distance via a predetermined relationship.
• §102 analysis: This reference family teaches the beam-profile depth-measurement mechanism that reads on elements of step (c) (determining depth from beam-profile information of reflection features). It does not disclose face detection, skin/material classification from beam profiles, or the authentication combination of claim 1.
• Common-ownership caveat: Under AIA 35 U.S.C. §102(b)(2)(C), subject matter owned by (or under an obligation of assignment to) the same person as this application at the effective filing date may be excluded from §102 prior art — likely applicable here given trinamiX ownership, so these references may be citable only for what they teach, not as statutory §102 anticipations.

3. WO 2018/091638 A1 and 4. WO 2018/091640 A1

• Type: PCT publications in the same trinamiX DPR "detector" family (cited together with WO 2018/091649 A1 in the specification's DPR discussion).
• Disclosure: Companion implementations of the beam-profile/depth-from-photon-ratio technique (optical detectors determining object position from light-spot beam-profile quotients).
• §102 analysis: Same as item 2 — they anticipate the underlying depth-measurement sub-technique used in step (c) but not the claimed face-authentication method; likely excluded under §102(b)(2)(C) by common ownership.

5. WO 2020/187719 A1

• Type: PCT publication, trinamiX/BASF group; referenced in the specification for material classification by beam-profile analysis.
• Disclosure: Classifying the material of a surface (including distinguishing biological tissue/human skin from other materials) by analyzing the beam profile/back-scattering of a projected light feature — the "skin detection" technology that step (b) builds on.
• §102 analysis: Reads on the step (b) sub-technique (determining a material property from beam-profile information to identify skin). It does not disclose the face-detection → skin-gated → 3D-gated → authentication pipeline of claim 1. Same common-ownership §102(b)(2)(C) caveat as items 2–4.

6–8. European applications incorporated by reference (EPO numbers as printed in the spec)

• EP application No. 20 159 984.2 (filed 2020-02-28); EP application No. 20 154 961.5 (filed 2020-01-31); EP application No. 20 170 905.2 (filed 2020-04-22, describing a periodic/hexagonal grid of laser spots for reflection imaging).
• Disclosure: Incorporated by reference for material/beam-profile classification and for the patterned-spot illumination used in the reflection image.
• §102 analysis: All are trinamiX's own earlier-filed applications; each discloses component technology (spot-grid illumination, beam-profile material classification) but not the claimed combination. As unpublished-or-published commonly-owned applications, they are subject to the §102(b)(2)(C) exclusion analysis and would not, individually, anticipate claim 1.

Non-patent literature cited in the spec (for completeness)

• Masi, Iacopo, et al., "Deep face recognition: A survey," 2018 31st SIBGRAPI Conference on Graphics, Patterns and Images (SIBGRAPI), IEEE, 2018 — incorporated for 2D face-detection/geometrical-feature techniques (step (a)).
• J. Eichler & T. Seiler, "Lasertechnik in der Medizin" (Springer, 1991), pp. 171–266 — incorporated for the skin reflection model (surface reflection + diffuse subsurface back-scattering) underlying step (b).

---

Bottom-line §102 assessment

Against the reconstructed independent method claim (the four-step face-detection → skin-detection → 3D-detection → authentication pipeline):

• No single reference retrieved anticipates all four steps. The closest is US 2019/0213309 A1 (distance + reflectance gating of facial recognition), which fails to disclose patterned-illumination beam-profile skin classification, the ≥4-reflection-feature 3D-vs-plane depth test, and the conjunctive authentication condition.
• The trinamiX WO 2018/091649/091638/091640 and WO 2020/187719 references each anticipate only an individual sub-technique (DPR depth; beam-profile material/skin classification) and lack the face-authentication integration; they are also likely disqualified as §102 prior art by common ownership under §102(b)(2)(C).
• If the full examiner "References Cited" list becomes available from the USPTO Patent Center record for US12288421B2, this analysis should be re-run against the verbatim granted claims (which were truncated in my source material) — the claim-by-claim mapping above is keyed to the specification's independent-claim summary and the family publication's claim language, not to the certified USPTO claim text.

Key URLs: patents.google.com/patent/US12288421B2/en • patents.justia.com/patent/12288421 • freepatentsonline.com/y2024/0037201.html (full family specification) • patentimages.storage.googleapis.com/a5/9c/d6/934ad01ea375a2/US20240037201A1.pdf

§ 103 Obviousness Analysis — US 12288421 B2 ("Optical skin detection for face unlock")

0. Threshold notes: date environment and claim-set discrepancy (flagged)

Date environment. The task block states "Current Date: April 26, 2026," while the system environment and the live search results (Reuters/Biometric Update/digit coverage of the trinamiX v. Apple complaint) are dated September 2026. I treat the live search results as ground truth per the operating rules. For § 103 purposes the operative date is the effective filing date, 2021-02-18 (EP priority 21157800.0), which is unambiguous regardless of the discrepancy. Nothing in this analysis depends on the difference between April and September 2026.

Claim-set discrepancy (flagged against earlier sections). The previously generated sections reconstructed Claim 1 as a four-step claim (face detection → skin detection → 3D detection → authentication, with explicit "beam profile information" and "at least four reflection features" recitations). The verbatim granted claim set retrieved live from Justia (patents.justia.com/patent/12288421#10) differs materially. Granted Claim 1, as retrieved, is a three-step claim:

1. A method for face authentication comprising:

• a) a skin detection step: projecting an illumination pattern comprising a plurality of illumination features (pattern selected from point/line/stripe/checkerboard/periodic-or-non-periodic patterns) on the scene with an illumination unit; determining a second image with the camera, the second image comprising a plurality of reflection features generated in response to the illumination; determining at least one material property of the reflection features with the processing unit; characterizing the detected face as skin if the material property corresponds to a property characteristic for skin;
• b) a 3D detection step: determining at least one depth level from the reflection features of the second image, characterizing the detected face as a 3D object if the depth level deviates from a pre-determined depth level, wherein depth information is determined by structured light; and
• c) an authentication step: authenticating only if the face is characterized as skin and as a 3D object.

Caveat: the retrieved text is partial (it runs through mid-Claim 17). Claim 4's reference to "the first image" implies Claim 1 likely contains a first-image/face-region recitation not visible in the snippet; I note where this matters. The analysis below keys to the retrieved text and flags where the earlier four-step reconstruction would change the outcome.

---

1. Legal standard and the person of ordinary skill

Under 35 U.S.C. § 103, a claim is unpatentable if the differences between it and the prior art are such that the claimed invention as a whole would have been obvious at the effective filing date (2021-02-18) to a PHOSITA, evaluated under Graham v. John Deere (scope and content of prior art, differences, level of skill, secondary considerations) with the flexible motivation/ reasonable-expectation-of-success framework of KSR Int'l Co. v. Teleflex (2007).

PHOSITA profile. An engineer/scientist with a B.S./M.S. in electrical engineering, computer science, or optical engineering and 2–5 years in biometric authentication, computer vision, and optical sensing — conversant in: 2D face detection/recognition; structured-light and time-of-flight depth sensing; NIR cameras and VCSEL/DOE dot projectors as deployed in mobile "face unlock"; presentation-attack detection/liveness (photo/video/mask attacks); machine-learning classifiers; and the reflectance/scattering optics of skin. This profile matters because the central § 103 question is whether a person building a face-unlock anti-spoofing pipeline in early 2021 would have found it obvious to (i) classify the presented surface as skin vs. mask material from the reflected light pattern, (ii) derive structured-light depth from the same reflected features, and (iii) AND-gate authentication on both.

---

2. Prior-art corpus relevant to this analysis

Grounding references are those catalogued in the Prior Art section (itself drawn from the patent specification's own citations), plus two verified live-search supplements. I flag each item's status.

A. Applicant-acknowledged references (in-spec, Prior Art section).

1. US 2019/0213309 A1 (printed in the patent as "US 2019/213309 A1") — face authentication with a ranging sensor (ToF + reflectance); facial recognition is activated in response to distance and reflectivity of the user. This is the closest single framework reference: it already teaches gating face authentication on physical attributes of the presented object (a material-ish attribute — reflectance — and a depth-ish attribute — distance), i.e., the PAD-gating concept underlying claim step (c).
2. WO 2018/091649 A1, WO 2018/091638 A1, WO 2018/091640 A1 — trinamiX depth-from-photon-ratio (DPR) disclosures: per-point distance from analysis of the beam profile (e.g., quotient of center/edge areas) of a projected light spot captured by a camera/sensor.
3. WO 2020/187719 A1 — trinamiX disclosure of classifying surface material, including distinguishing human skin from other materials, by beam-profile/back-scattering analysis of projected light features (sub-surface scattering/translucence).
4. EP 20159984.2, EP 20154961.5, EP 20170905.2 — trinamiX applications incorporated by reference for skin/material classification by beam profile and for a periodic (e.g., hexagonal) grid of laser spots.
5. NPL: Masi et al., "Deep face recognition: A survey" (2018) (2D face detection); Eichler & Seiler, Lasertechnik in der Medizin (Springer, 1991) (skin optics: surface reflection + diffuse subsurface backscatter).

B. Verified live-search supplements (not in the Prior Art section; added under the operating rule that live results are ground truth).
6. Verbatim granted claims (Justia, retrieved 2026) — used above; also confirms dependent claims on ~940 nm illumination, machine-learning skin classification, facial-landmark/region-specific models, specular/diffuse reflectivity or translucence, spot-shape and Law's-energy filters.
7. trinamiX's own product datasheet, "Behind OLED Face Authentication," Release date July 2020 (trinamixsensing.com/media/datasheet_smartphone.pdf) — i.e., before the 2021-02-18 priority date. It publicly describes a face-authentication solution combining, from one Beam-Profile-Analysis frame: a flood-illuminated NIR 2D image, 3D depth, and skin detection, where "only if all three checks are passed in a single unlock process, the device unlocks"; hardware = 940 nm dot projector (1591/509 points), NIR camera; PAD vs. photos/videos and "even 3D printed masks." This is powerful state-of-the-art evidence that the combination claimed here was already public and commercialized before the priority date — by the applicant itself (self-collision implications discussed in § 6).
8. Apple Face ID / TrueDepth public disclosures (commercial state of the art, 2017–2021) — iPhone X (2017) onward: a dot projector emitting tens of thousands of IR dots plus flood illuminator and NIR camera; structured-light 3D depth map; a dedicated anti-spoofing neural network trained against photos and masks. Per Biometric Update's coverage of the trinamiX suit, Apple has a patent family claiming priority to September 2018 describing facial authentication using both flood IR and patterned IR illumination for 3D depth — predating this patent's 2021 priority. No specific publication number was retrievable; I reference this only generically and flag that the precise Apple family number must be verified from the litigation docket before it is used in an IPR.

C. Disqualification risk (flagged). Items 2–4 and 7 are trinamiX's own. Under AIA § 102(b)(2)(C), subject matter disclosed in commonly-owned § 102(a)(2)-type documents is not prior art under § 102(a)(2). Critically, that exception is cabined to § 102(a)(2); it does not remove subject matter that is independently prior art under § 102(a)(1) as a printed publication. WO publications publicly disseminated by WIPO and the publicly posted July-2020 datasheet are prima facie § 102(a)(1) printed publications. Whether the (b)(2)(C) shield nevertheless bars their use — including in an IPR — is a live legal question (see § 6). I analyze both availability scenarios.

---

3. Claim 1 element chart

---

4. The combinations and why a PHOSITA would make them

Combination C1 — US 2019/0213309 A1 + WO 2020/187719 A1 (core skin-gating)

Resulting disclosure: a face-authentication system in which authentication/recognition proceeds only after a material attribute of the presented face passes a check — US '213309 already gates facial recognition on the user's reflectance measured from a transmitted signal. WO '87719 teaches that human skin is reliably discriminated from other materials (silicone, latex, plastics, 3D-print media) by analyzing the back-scattering/beam profile of projected light features, exploiting skin's characteristic sub-surface scattering.

Motivation. The gap in US '213309 is obvious on its face: a scalar reflectance threshold is spoofable by any material engineered to reflect like skin (precisely the 3D-mask problem the patent's own background identifies as the unsolved attack class — ¶[0004], [0010]). A PHOSITA seeking to harden US '213309's reflectance gate against masks would consult the known body of material-classification-by-reflection art (WO '87719; the Eichler/Seiler two-component skin-reflection model is the physical basis) and substitute/augment the scalar reflectance test with a per-feature material classifier. This is the KSR paradigm of "known technique, known problem, predictable result": both references operate on the same modality (projected light, reflected signal), so integration requires no new hardware concept. C1 supplies elements (a) (pattern + second image + material property + skin characterization) and the skin half of (c).

Combination C2 — US 2019/0213309 A1 + WO 2018/091649 A1 (with /091638, /091640) (3D-from-pattern gate)

Resulting disclosure: US '213309 gates recognition on a single ToF distance of the user. The DPR disclosures teach deriving per-point longitudinal coordinates from the beam profiles of a plurality of projected reflection features captured in one image by one sensor — i.e., pattern-based (structured-light-type) depth over an extended surface.

Motivation. US '213309's scalar distance cannot distinguish a flat photograph of a face (presented at the same range) from a real face; the recognized fix in the art (acknowledged in the patent background ¶[0004]) is 3D sensing that "can clearly distinguish between planar photos or tablets with playback videos and a 3D face." A PHOSITA combining the ranging-gated authenticator with DPR (or with any commercial structured-light depth method, e.g., Face ID's dot-projection depth mapping, in the public domain since 2017) obtains a per-feature depth map from the same reflected pattern, and a "3D vs. plane" test is the natural classifier over such a map. C2 supplies element (b) and the 3D half of (c).

Combination C3 — Full combination: US 2019/0213309 A1 + WO 2020/187719 A1 + WO 2018/091649 A1 (+/091638, /091640) (all elements)

Resulting disclosure: A face-authentication method that (i) projects a multi-feature illumination pattern on the scene, (ii) records the reflection features in a second image, (iii) derives a material property per reflection feature (WO '87719) to confirm the presented surface is skin rather than a mask material, (iv) derives per-feature depth from the same reflection features (DPR) to confirm the surface is 3D and not flat media, and (v) authenticates only when both conditions hold (US '213309's conjunctive gating concept, now applied to the two PAD-relevant attributes).

Motivation to combine — the core § 103 inquiry. Several independent rationales converge:

1. The patent itself documents the modular prior-art problem/solution structure. Its background (¶[0004]–[0010]) concedes: 3D cameras solve flat-photo/video attacks; "these types of masks can be rejected by a PAD system that is able to classify human skin from other materials such as described in [the incorporated EP applications]"; and the residual problem is that no single system "reliably detect[s] spoof attacks by 3D masks" at acceptable speed. In other words, both component solutions were known in the same technical field and for the same purpose; the claimed contribution is their co-deployment with a conjunctive gate. Where known components are each deployed to solve the known sub-problems they were designed for, the combination is obvious (KSR; PRG v. Philips line of cases is inapposite because there is no teaching away here).
2. Hardware and data synergy make the combination the obvious engineering path. Both the skin check (WO '87719) and the depth check (DPR) consume the same input — a patterned-illumination reflection image — and both were developed by the same research program for the same projector/camera geometry. A PHOSITA facing the documented speed/power constraints of face unlock (¶[0006]–[0007]) has a direct design incentive to compute both attributes from a single captured frame rather than running two sensing modalities. The trinamiX July-2020 datasheet confirms that by the priority date the "2D + 3D depth + skin detection, only if all three checks pass in a single unlock" architecture was not merely conceivable but commercialized — strong evidence of the combination's obviousness (and, if the datasheet is citable, of its having been in the public domain).
3. Market/industry pressure. By early 2021, PAD for face unlock was a mature, standards-driven field (FIDO biometric component certification, Android Biometric Class 3), and mask attacks were the acknowledged next frontier. A PHOSITA designing a mask-resistant system would layer the known material test onto the known structured-light depth test precisely because the two attack classes (flat media, volumetric masks) map one-to-one onto the two known sensor-side countermeasures (depth, material). This is the standard "defense-in-depth" motivation courts routinely credit under § 103.

Under C3, the remaining differences from the prior art are: (i) performance of the material classification on reflection features within the face region of the same frame used for structured-light depth; and (ii) the specific conjunctive skin-AND-3D authentication condition. Both are ordinary applications of disclosed techniques to disclosed problems, with predictable results; neither requires new hardware, new physics, or an unexpected interaction.

Combination C4 — Third-party-only fallback (if trinamiX-owned art is excluded)

US 2019/0213309 A1 + Apple Face ID/TrueDepth public disclosures (structured-light dot-pattern face unlock with depth map and anti-spoof NN; Apple family priority Sept. 2018) + skin-optics literature (Eichler/Seiler 1991) and the general published art on reflectance-based skin/liveness detection. C4 comfortably supplies the face-unlock context, patterned illumination, structured-light 3D-vs-plane depth, and the anti-spoof/AND-gate concept. Its weak point is element (a)'s skin-classification-from-reflection-features limitation: Eichler/Seiler supplies the physical signature of skin (surface reflection + diffuse sub-surface backscatter) but not a ready-made per-feature classifier, and US '213309's scalar reflectance is only a coarse proxy. C4 is a genuine but weaker obviousness case; a challenger would need additional third-party art (e.g., published reflectance-spectroscopy or multi-spectral liveness work) to close the gap cleanly. I did not retrieve such third-party references in this session; that is an open item, not a finding that none exist.

---

5. Dependent-claim analysis

• Claim 2 (device-secured access; permit access if authenticated) and Claim 3 (smartphone): The standard commercial setting of Face ID-class face unlock (2017 onward) and of every face-authentication reference discussed; obvious.
• Claim 4 (enrollment storing "3D or material-derived feature"; identification by comparison with first image): Enrollment of a 3D face model is standard Face ID practice. Enrollment of a material fingerprint is less clearly taught by third-party art; this claim is the most defensible dependent claim in the C4 world and moderately defensible even under C3 (WO '87719's material classification is per-material-class, not per-user identity). Lower confidence of obviousness.
• Claim 5 (region-of-interest selection), and the recurring ROI limitations (13, 16, …): routine image-analysis step, applied in every face-PAD pipeline; obvious.
• Claim 6 (~940 nm): The specification itself states that at 940 nm melanin absorption "runs out" so skin-tone variation is minimized. NIR ~940 nm was the standard face-unlock wavelength (Face ID's VCSELs; trinamiX's own 940 nm dot projectors per the 2020 datasheet). Selecting a wavelength for sensor sensitivity, eye safety, and skin-tone invariance is a routine design choice; obvious.
• Claims 7–9 (parametrized/machine-learned skin classification model; facial-landmark detection; region-specific models): Machine-learned classifiers for liveness/spoof resistance were standard (Apple's anti-spoof NN, publicly documented; the patent admits second-generation face recognition uses deep CNNs). Feeding a classifier with the face-region features identified by a landmark detector is conventional; region-specific training is an optimization, not an inventive leap. Obvious if the underlying skin-classification concept is in the art (WO '87719 uses ML per the family's own description).
• Claim 10 (material property = specular reflectivity, diffuse reflectivity, or translucence): These are precisely the beam-profile-derived quantities in WO '87719 / the incorporated EP applications and correspond to the Eichler/Seiler two-component skin-reflection model; obvious.
• Claims 11, 14 (spot-shape filter; Law's energy filter or correlate): The family specification describes the spot-shape filter as measuring translucence and the Law's-energy/speckle-contrast filter as measuring volume-scattering smoothness — i.e., implementations of the very skin/material signatures the incorporated material-classification applications teach. If those applications (or equivalent third-party speckle/texture analysis art) are available, obvious.
• Claims 12, 15, … (redundant structured-light recitations): dependent claims that re-add a limitation already in Claim 1 — no independent patentable weight; they track the amendment history and are obvious for the same reasons as Claim 1's step (b).

---

6. Threshold legal issues that govern the strength of any § 103 case

1. Common ownership (§ 102(b)(2)(C)) vs. § 102(a)(1) availability. The most on-point component references (WO '87719; WO '1649/'1638/'1640; the incorporated EP applications) are trinamiX's own, and trinamiX owned them at the 2021-02-18 effective filing date. The (b)(2)(C) exception on its face removes only § 102(a)(2) status. Publicly disseminated WO publications and the July-2020 datasheet are prima facie § 102(a)(1) printed publications, which the exception does not textually reach. However, the interaction is genuinely unsettled (whether (b)(2)(C) shields subject matter that is also (a)(1) art is litigable), and in an IPR the PTAB would scrutinize this closely. Any invalidity theory must plead the availability theory expressly. This is the single largest swing factor: if trinamiX's own disclosures are fully available, C3 is a strong case; if they are wholly excluded, only C4 remains and element (a) is under-supported.
2. Applicant admissions. The specification's own "Prior Art" and "Problem Addressed" sections concede: 3D cameras already distinguish flat media from 3D faces; skin-vs-material PAD is "described in" the incorporated trinamiX applications; the residual problem is reliable, fast 3D-mask rejection. In litigation these are party admissions usable to show the state of the art, the known problem, and the obviousness of combining known countermeasures — though they are not themselves prior art for examination purposes.
3. Self-collision / on-sale-type evidence. The July-2020 datasheet (pre-priority, public, describing the "all three checks" architecture with 940 nm dot projector, skin detection, and 3D depth from one frame) is potentially devastating to patentability if its publication/availability date is proven and its disclosure is enabling — including as § 102(a)(1) art or as evidence of public use/offer for sale. It also undermines any secondary-consideration argument built on commercial success of the combination, because the combination's commercial embodiment predates the priority date. Its use is subject to the same common-ownership analysis (though (b)(2)(C) does not cover (a)(1) marketing materials as a matter of text) and to an enabling-disclosure inquiry: the datasheet describes the architecture at a high level and points to the underlying patents for implementation.

---

7. Secondary considerations (Graham factor four)

• Long-felt need / failure of others: Weak for the patentee. The field's acknowledged need (mask-resistant, fast PAD) was being addressed by layered depth+material approaches before the priority date, including by trinamiX itself (July-2020 product) — undercutting any narrative that others failed where this patent succeeded.
• Commercial success: trinamiX's certified, commercialized Face Authentication product and the Apple allegations post-date the priority date but the combination's commercial embodiment (datasheet) pre-dates it; success is not attributable to the claimed invention as against prior art that already embodied the combination. No nexus.
• Unexpected results / teaching away / skepticism: None identified in the corpus. The references describe layered PAD as a natural progression (photo → video → mask → material). No reference teaches away from combining depth and material checks; US '213309's dual-condition gating affirmatively points toward multi-attribute gating.
• Industry praise/copying: Certification (FIDO, IIFAA) is of a product line, not of the claimed method per se, and post-dates the relevant prior-art disclosures.

On the available record, secondary considerations do not rescue the claims from a C3-type obviousness challenge if the trinamiX-owned art is available.

---

8. Bottom-line assessment

Verdict. Claim 1 of US 12288421 is prima facie obvious under § 103 on Combination C3 — US 2019/0213309 A1 (dual-attribute, reflectance-and-distance gated face authentication) in view of WO 2020/187719 A1 (skin vs. material classification from reflection-feature beam profiles) and WO 2018/091649 A1 / /091638 / /091640 (per-feature structured-light-type depth from the same reflection features) — with the trinamiX July-2020 product datasheet and Apple Face ID public disclosures corroborating that the combined architecture was known and commercialized in the face-unlock field before the 2021-02-18 priority date. The motivations are conventional KSR motivations: known modular solutions to known, acknowledged sub-problems (flat-media attacks → 3D/depth test; volumetric-mask attacks → material/skin test), a shared projector/camera data modality that makes single-frame co-computation the obvious low-power/high-speed engineering choice, and market/standards pressure for mask-resistant PAD. The conjunctive "skin AND 3D" gate is a straightforward application of US '213309's existing multi-condition gating concept to the two PAD-relevant attributes.

The decisive open question is legal, not technical: whether trinamiX's own foundational disclosures (WO '87719; WO '1649 family; incorporated EP applications) are usable as § 102(a)(1) printed publications notwithstanding § 102(b)(2)(C), and whether the July-2020 datasheet's public availability and enabling character can be proven. If yes, the obviousness case is strong. If no, a challenger must build element (a) from third-party skin/material-sensing art not yet identified in this record — and claims such as Claim 4 (material-fingerprint enrollment) may survive even a successful C3 challenge.

Open items to strengthen/confirm this analysis: (1) retrieve the full verbatim granted claim set (through the last dependent claim) from USPTO Patent Center; (2) obtain the actual text of WO 2020/187719 and WO 2018/091649 to map disclosures claim-element-by-claim-element rather than through the patent's own description of them; (3) verify the July-2020 datasheet's public availability date and enabling detail; (4) identify third-party (non-trinamiX) references disclosing skin/material classification from reflected patterned light (e.g., spectral-reflectance liveness literature) to build the C4 fallback; (5) confirm whether US 12288421 is among the seven patents asserted in trinamiX GmbH et al. v. Apple Inc. (W.D. Tex., filed 2026-09-03) and, if so, monitor for IPR petitions — Apple's predictable response — which would test these exact combinations before the PTAB.

Sources: patents.justia.com/patent/12288421#10 (verbatim claims); patentimages.storage.googleapis.com/a5/9c/d6/934ad01ea375a2/US20240037201A1.pdf (family publication, spec ¶[0004]–[0010] admissions); trinamixsensing.com/media/datasheet_smartphone.pdf (July-2020 product datasheet); biometricupdate.com/202609/trinamix-accuses-apple-of-infringing-face-id-skin-detection-patents (Apple Face ID background and Sept-2018 Apple family); Prior Art section of this file (reference corpus and § 102 analysis).