US 8472792

I'll verify the patent's litigation status and check for any 2026 CAFC activity before summarizing.

Let me pin down the issued claim set and check whether the 2026 Federal Circuit activity actually involves this patent.

US 8,472,792 B2 — "Multimedia distribution system"

⚠️ Note on search scope: I searched for US 8,472,792 specifically. A 2026 Federal Circuit decision involving DivX and Netflix surfaces prominently in searches and does involve DivX patents — but not this one (see §5). I have flagged that distinction explicitly rather than conflating the cases.

1. Bibliographic data (verified)

Sources: Google Patents; FreePatentsOnline

2. Abstract (verbatim)

"A multimedia file and methods of generating, distributing and using the multimedia file are described. Multimedia files in accordance with embodiments of the present invention can contain multiple video tracks, multiple audio tracks, multiple subtitle tracks, a complete index that can be used to locate each data chunk in each of these tracks and an abridged index that can enable the location of a subset of the data chunks in each track, data that can be used to generate a menu interface to access the contents of the file and 'meta data' concerning the contents of the file. … One embodiment of a multimedia file in accordance with the present invention includes a series of encoded video frames, a first index that includes information indicative of the location within the file and characteristics of each encoded video frame and a separate second index that includes information indicative of the location within the file of a subset of the encoded video frames."

3. Plain-language overview

The patent is a container/file-format patent, not a codec patent. It describes a RIFF/AVI-style multimedia container (the DivX "extended AVI" family) that carries:

• Multiple video, audio, and subtitle tracks in one file (plus references to external tracks).
• Two indexes over the video:
  • an "idx1"-style complete index ("first index") giving the location and characteristics of every encoded frame — but conventionally placed after the media data;
  • a separate "index" chunk (fourcc like idxx) — the abridged index — covering only a subset of frames (e.g., scene/chapter starts, or frames ~1–10 s apart), placed before the media data.
• Each abridged-index entry is a "tag" containing the chunk offset, a pointer into the complete index, a frame number, and references to corresponding portions of the audio (and subtitle) tracks.
• DXDT "meta data" chunk expressed as subject–predicate–object–authority RDF-style statements.
• DMNU menu chunks defining navigable menus (a "lite" menu before the media data, a richer menu after), with background video/audio, buttons, and actions.

The commercial point: because the abridged index sits ahead of the movie data, a player that has only partially downloaded (or is streaming) the file can already seek, fast-forward, rewind, and skip scenes without waiting for the tail-end complete index. That indexing-ahead-of-media idea is what later DivX adaptive-streaming litigation turned on.

4. Independent claims — plain-language overview

⚠️ Uncertainty flag: I was not able to retrieve the granted claim text of US 8,472,792 in this session. The summaries below are based on the pre-grant publication US 2006/0129909 A1 for application 11/258,496 (which lists independent claims 1, 5, 9, 12, and 16), cross-checked against PTAB and district-court records. The granted patent appears to contain 23 claims, with the IPR identifying claims 1, 9, and 15 as independent among the challenged claims, while claim 16 was the claim asserted against Netflix in district court. The claim text as issued may therefore differ in numbering and wording from what is summarized below. Treat the claim numbers as approximate and verify against the official USPTO PatentCenter copy before relying on them.

Dependent claims add: per-tag frame numbering, per-tag audio-track and subtitle-track references, cross-references from the abridged index into the complete index, and external-file track references.

5. Litigation, PTAB, and Federal Circuit status

a) PTAB — IPR2020-00646 (Netflix, Inc. et al. v. DivX, LLC)

• Filed March 6, 2020; instituted September 15, 2020; petitioners Netflix and Hulu. Challenged claims 1, 5, 8, 9, 13–15, 18, 21–23.
• Original Final Written Decision (Sept. 9, 2021): Board rejected the obviousness grounds, holding Petitioner had not established Kaku was analogous art.
• Federal Circuit appeal (No. 2022-1138): decided September 11, 2023 — Netflix, Inc. v. DivX, LLC, 80 F.4th 1349 — vacated and remanded, holding the Board abused its discretion in finding Petitioner failed to articulate a field of endeavor; remand directed the Board to decide whether Kaku is in the same field of endeavor. Mandate issued Oct. 18, 2023.
• Final Written Decision on Remand — April 17, 2024: the Board determined all challenged claims unpatentable. Hulu had settled and been terminated in February 2024.
• Uncertainty: I did not verify whether the April 2024 remand decision was appealed and what its current status is. That materially affects the patent's practical enforceability.

Sources: PTAB remand FWD excerpt; IPVerse case record; Docket Alarm

b) District court — C.D. Cal.

• DivX, LLC v. Netflix, Inc., No. 2:19-cv-01602 (C.D. Cal.); companion DivX, LLC v. Hulu, LLC, No. 2:19-cv-01606 (dismissed Aug. 25, 2022).
• Jury trial commenced March 13, 2026; verdict returned March 25, 2026 — Netflix did not infringe the asserted claims of four patents, including '792.
• Final judgment (reported May 7, 2026, Judge George H. Wu): "Netflix has not infringed and does not infringe claim 16 of U.S. Patent No. 8,472,792."

Sources: Justia docket excerpt; CourtListener docket 2:19-cv-01602

c) ⚠️ 2026 Federal Circuit activity — IMPORTANT DISTINCTION
The notable 2026 CAFC decision involving these parties is Netflix, Inc. v. DivX, LLC, No. 24-1541 (Fed. Cir. Feb. 13, 2026) — and it concerns U.S. Patent 10,225,588, not 8,472,792. It addressed IPR2020-00558 and claim construction of "locating encryption information that identifies encrypted portions of frames of video within the requested portions of the selected stream of protected video." The court applied the nearest-antecedent rule, held the modifier attaches to "encrypted portions of frames of video," reversed the Board's construction, and vacated/remanded the non-obviousness determination.

I found no 2026 CAFC docket entry decided on the merits of US 8,472,792 itself. Secondary commentary (e.g., a PatSnap blog page) loosely frames appeal 22-1138 as the "US8472792B2 Multimedia Distribution Appeal," which is consistent with the IPR2020-00646 appeal described above, but that appeal was decided in 2023, not 2026. Treat any source that attributes the February 2026 Netflix v. DivX / '588 decision to '792 as conflating two different patents.

Sources: CAFC 24-1541 opinion (Feb. 13, 2026); Vorys case note; Lexology; PatSnap (secondary, treat with caution)

d) Other dockets listed on Google Patents (not independently verified): D. Del. 1:24-cv-01390; D. Del. 1:21-cv-01828. I did not confirm which patents are asserted in those cases or their current posture.

6. Summary of uncertainties

1. Granted claim text and independent-claim numbering of '792 were not retrievable in this session. The claim summary above is based on the pre-grant publication (application 11/258,496) and PTAB/litigation records; verify against USPTO PatentCenter.
2. Post-April-2024 status of IPR2020-00646 (whether the "all challenged claims unpatentable" remand decision was appealed, and the outcome) is unverified. This is the single most important open question for enforceability.
3. Claim 16 vs. claims 1/9/15: the district court record asserts claim 16, while the PTAB record describes 1, 9, and 15 as independent among the challenged claims. I cannot reconcile the complete independent-claim set from the records retrieved.
4. Date inconsistency in the task inputs: the prompt states "Current Date: April 26, 2026," yet search results include a May 7, 2026 final judgment in 2:19-cv-01602 (and the session header says 2026-09-22). I report the dates as they appear in the retrieved sources rather than resolving the discrepancy.
5. The "Adjusted expiration 2027-06-28" figure is Google Patents' own disclaimer-labeled assumption, not an official USPTO term calculation.

I'll research this. Let me start with targeted searches on the patent number and the litigation sources you named.

Good progress. Now let me dig into the specific case details, the ITC action, and the Delaware 2021 case.

Litigation involving U.S. Patent No. 8,472,792 ("Multimedia Distribution System")

Interpretation note: I treated the identifier literally as 8,472,792 / US 8,472,792 B2 (DivX, LLC; priority 2003‑12‑08; filed 2005‑10‑24; issued 2013‑06‑25). Everything below is limited to matters in which that specific patent was asserted or challenged. Sources are cited with URLs.

Related administrative (PTAB) and appellate proceedings

• IPR2020-00646 — Netflix, Inc. and Hulu, LLC v. DivX, LLC (U.S. Patent No. 8,472,792). Petition filed March 6, 2020; institution granted Sept. 15, 2020; challenged claims 1, 5, 8, 9, 13–15, 18 and 21–23. The Board's original Final Written Decision rejected the obviousness grounds, finding Petitioner had not shown Kaku was analogous art. Hulu was terminated as a petitioner on Feb. 22, 2024 after settlement (35 U.S.C. § 317), leaving Netflix as the sole petitioner.
• Netflix, Inc. v. DivX, LLC, No. 22-1138 (Fed. Cir.) — appeal from IPR2020-00646. Decided Sept. 11, 2023: the Federal Circuit vacated the Final Written Decision and remanded, holding the Board abused its discretion in finding Netflix failed to articulate a field of endeavor, and directing the Board to decide on remand whether Kaku is in the same field of endeavor (reported as Netflix, Inc. v. DivX, LLC, 80 F.4th 1362). The court did not reach the other issues. I was unable to confirm the outcome of the remand proceedings; the final validity posture of the '792 claims (and whether a further Federal Circuit appeal followed) should be verified against PTAB/PTACTS records.

Notes on scope and reliability

• The five district-court/ITC matters above are the ones I can confirm from retrieved sources as involving 8,472,792 specifically. The Google Patents litigation links for this patent list the C.D. Cal. cases (2:19-cv-01602, 2:19-cv-01606), the Delaware cases (1:21-cv-01828, 1:24-cv-01390), the Federal Circuit appeal (22-1138), and PTAB case IPR2020-00646 — consistent with the table above.
• Several other DivX matters are easy to confuse with this one but do not concern 8,472,792, e.g. DivX's later ITC investigation instituted Nov. 29, 2022 (asserting the '808, '297, '673, '588 and '553 patents; final ID of no violation May 29, 2024, affirmed and investigation terminated), and the Hulu II case (2:21-cv-01615, asserting other DivX patents such as the '987/'141 family). DivX also referenced PTAB proceedings IPR2020-00647 and IPR2021-01418 in third-party filings, but I did not verify whether '792 is at issue in those, so I have excluded them.
• Dates/statuses for matters still developing (Netflix post-trial motions and any appeal; the PTAB remand; the Top Victory action) should be re-checked against PACER and the PTAB PTACTS system before being relied upon.

I'll research this. Note that the patent's own Google Patents litigation block already flags PTAB case IPR2020-00646 (Final Written Decision) associated with US8472792 — so the "no PTAB activity" default from the ODP block needs to be tested against that. Let me search.

Excellent — the structured block's "no PTAB activity" default is wrong for this patent. Let me dig into the claim-level outcome and the appeal.

Now let me check for additional IPRs on this patent, the appeal of the 2024 remand FWD, and the district court / settlement picture.

PTAB / AIA Trial Proceedings — U.S. Patent No. 8,472,792 (DivX, LLC)

⚠️ Discrepancy flag up front. The structured "PTAB proceedings on file" block in this prompt states the USPTO ODP API returns no AIA trial proceedings for this patent. That is wrong, or at least stale. US8472792 is the subject of IPR2020-00646, which was instituted, went to a Final Written Decision, was appealed to the Federal Circuit, was remanded, and terminated with a second Final Written Decision on 2024-04-17 holding every challenged claim unpatentable. Google Patents' own litigation block for this patent lists "PTAB case IPR2020-00646 filed (Final Written Decision)." Do not rely on the ODP "no proceedings" default here.

Proceedings overview

One AIA trial proceeding is on file for this patent — IPR2020-00646 (Netflix, Inc. and Hulu, LLC v. DivX, LLC) — and its bottom line is adverse to the patent owner: all 11 challenged claims were held unpatentable in a Final Written Decision on Remand dated 2024-04-17 (0 active, 0 institution denials, 1 claims-invalidated, with Hulu's participation separately settled after institution in 2024-02-22). The defensive posture this gives a defendant today is: "claims 1, 5, 8, 9, 13–15, 18, and 21–23 are dead — any demand letter or infringement theory built on those claims is sanction-bait; the only live exposure is claim 16 (and any other never-challenged claims), and even claim 16 was adjudged not infringed by Netflix in the C.D. Cal. jury trial that concluded in March 2026."

IPR2020-00646 — Netflix, Inc. and Hulu, LLC v. DivX, LLC

• Type: Inter Partes Review (35 U.S.C. §§ 311–319)
• Filed: 2020-03-06 (Petition, 90 pp.)
• Status: Verbatim from the structured record — "PTAB case IPR2020-00646 filed (Final Written Decision)". Plain-English gloss: instituted, tried, appealed, remanded, and terminated 2024-04-17 with all challenged claims held unpatentable.
• Judge panel: Bart A. Gerstenblith, Monica S. Ullagaddi, and Iftikhar Ahmed, Administrative Patent Judges; APJ Ahmed authored both the 2021 FWD and the 2024 FWD on remand. Art Unit 2484, Tech Center 2400.
• Petition grounds (pre-AIA §§ 102/103 govern — the '792 patent's effective filing date precedes 2013-03-16):
  • Ground 1 — § 103(a): claims 1, 5, 8, 9, 13–15, 18, 21–23 obvious over Zetts (U.S. Pat. No. 7,212,726) in view of Kaku (U.S. Pat. No. 6,671,408).
  • Ground 2 — § 103(a): claims 5, 18, 21 obvious over Zetts, Kaku, and Seo (U.S. Pat. No. 8,286,213).
  • No § 102 anticipation ground and no § 112 ground were instituted. Independent claims at issue: claims 1, 9, and 15.
  • Petitioner's expert: Clifford Reader, Ph.D. Patent Owner's expert: Chandrajit Bajaj, Ph.D.
• Institution decision: Instituted 2020-09-15 ("Decision Granting Institution of Inter Partes Review," 57 pp.). The panel found a reasonable likelihood that claims 1, 5, 8, 9, 13–15, 18, and 21–23 were obvious over Zetts + Kaku and that claims 5, 18, 21 were obvious over Zetts + Kaku + Seo. (DivX had argued Kaku was merely cumulative art already before the examiner; that argument did not carry the day at institution.)
• Final Written Decision (original) — 2021-09-09, Paper 47: Patent Owner won. The Board held:
  • Claims 1, 5, 8, 9, 13–15, 18, and 21–23 — "have not been shown to be unpatentable."
  • Reasoning, verbatim: "Considering the full record before us, we conclude that Petitioner has not demonstrated by a preponderance of the evidence that one of ordinary skill in the art would have understood Kaku to be analogous prior art to the '792 patent." Specifically, the Board found petitioner and Dr. Reader "failed to identify the field of endeavor for either the '792 patent or Kaku — neither Petitioner's briefs nor either of Dr. Reader's two declarations address this threshold issue," and that "[t]he use of a common file format alone … does not say anything about the field of endeavor of either the '792 patent or Kaku." The Board likewise found Kaku not reasonably pertinent (it addressed digital-camera memory limitations and M-JPEG reproduction, not trick play in streamed multimedia).
  • DivX's Motion to Exclude was dismissed as moot.
  • Statutory deadline: institution 2020-09-15 → FWD due ~2021-09-15; issued 2021-09-09.
• Settlement / termination: Partial. On 2024-02-22 the Board issued "Settlement After Institution of Trial Only as to Hulu LLC, 35 U.S.C. § 317; 37 C.F.R. § 42.74," terminating Hulu from the proceeding after institution, on a joint motion to terminate. The settlement terms are confidential (subject to the § 42.74 joint-request-to-keep-separate practice); no terms are public. Separately, the underlying district court case DivX v. Hulu, No. 2:19-cv-01606 (C.D. Cal.) was dismissed 2022-08-25. Netflix did not settle — it litigated through remand and through district court trial.
• Appeal: Yes. Petitioners filed a Notice of Appeal 2021-11-08 → Fed. Cir. No. 2022-1138. Oral argument 2023-03-07; precedential opinion issued 2023-09-11, Netflix, Inc. v. DivX, LLC, 80 F.4th 1358 (Fed. Cir. 2023):
  • Affirmed-in-part: the Board's reasonable-pertinence holding was supported by substantial evidence.
  • Vacated-in-part: the Board abused its discretion in holding Netflix failed to articulate a field of endeavor — "our precedent does not require the use of magic words" — and the Board should have assessed Netflix's briefing as a whole (arguments that the patent and Kaku are both directed to AVI files and/or to encoding/decoding multimedia files).
  • Remanded with instructions to decide the factual question whether Kaku is in the same field of endeavor, expressly not to reconsider reasonable pertinence.
  • Mandate issued 2023-10-18.
  • Sources: CAFC opinion PDF (22-1138, 2023-09-11) · CAFC docket page · CourtListener 22-1138.
• Final Written Decision on Remand — 2024-04-17: "Final Written Decision on Remand Determining All Challenged Claims Unpatentable; Denying Patent Owner's Motion to Exclude" (35 U.S.C. §§ 144, 318). On remand the panel decided the remanded field-of-enumeration question for Netflix, concluded Kaku is analogous art, reached the merits issues the 2021 FWD had not decided (the joint list of issues on remand covered motivation to combine Zetts + Kaku with a reasonable expectation of success, and whether the combination teaches an index that "includes information … indicative of characteristics of each encoded video frame," etc.), and held all challenged claims — 1, 5, 8, 9, 13–15, 18, 21–23 — unpatentable. DivX's Motion to Exclude was denied this time. Proceeding terminated 2024-04-17.
  • Caveat I must be explicit about: I verified the claim set reached from the petition/institution record and the paper's title ("All Challenged Claims Unpatentable"), but I could not independently retrieve the remand FWD's claim-by-claim outcome table in the sources available to me. The FWD document itself (PTAB E2E, IPR2020-00646, Paper 58/59 area) should be pulled to confirm the per-claim disposition verbatim.
  • Also unverified: I found no notice of appeal from the 2024-04-17 remand FWD, and PTAB docket aggregators list the proceeding as terminated on 2024-04-17 with no subsequent papers. I also could not confirm whether the § 318(b) certificate cancelling claims 1, 5, 8, 9, 13–15, 18, 21–23 has issued. However, the district court record is consistent with issuance: DivX asserted five claims of the '792 patent (1, 9, 15, 16, 22); four of them were in the IPR, and claim 16 was the only '792 claim still addressed in the March 2026 verdict and May 2026 final judgment — i.e., the four IPR'd claims appear to have dropped out as canceled. Treat this as a strong inference, not a confirmed fact.
• Defensive value: Exceptionally high. Every claim Netflix and Hulu challenged is now held unpatentable — including independent claims 1, 9, and 15 — so any infringement theory, demand letter, or damages model built on those claims is untenable. Equally important, the Federal Circuit's precedential Netflix decision and the Board's remand FWD on the analogous-art question hand a future challenger a roadmap: the field of endeavor for this patent was successfully framed as AVI files / encoding-and-decoding of multimedia files, which neutralizes the exact "non-analogous art" defense that defeated the first FWD.

Strategic summary

Canceled vs. sustained vs. untested. Held unpatentable: claims 1, 5, 8, 9, 13–15, 18, 21, and 23 (the full challenged set). Note the claim-numbering quirk: claim 16 sits between challenged claims 15 and 18 and was never challenged in IPR — the stay-motion record expressly notes "Claim 16 (no IPR filed)." DivX asserted claims 1, 9, 15, 16, and 22 of the '792 patent in the C.D. Cal. litigation; claims 1, 9, 15, and 22 were all within the IPR's challenged set. Claim 16 is therefore the patent's practical residual scope, and it has been tested — and failed — in front of a jury: the C.D. Cal. jury returned a unanimous verdict on 2026-03-25 that Netflix did not infringe the asserted claims of the '792 patent among others, and the 2026-05-07 final judgment (No. 2:19-cv-1602-GW-AGRX) recites that "Netflix has not infringed and does not infringe claim 16 of U.S. Patent No. 8,472,792." That judgment is not issue-preclusive against a non-party, but it is powerful, admissible, non-infringement and credibility evidence. Consider also that Google Patents lists an adjusted expiration of 2027-06-28 — roughly nine months of nominal term remain, which further weakens leverage regardless of validity.

Estoppel landscape. § 315(e)(2) estoppel runs against Netflix, Hulu, and their privies/RPIs, and only as to claims that reached a Final Written Decision. Because the FWD covered claims 1, 5, 8, 9, 13–15, 18, 21, 22, and 23, Netflix/Hulu (and privies) cannot relitigate those claims on any ground raised or reasonably raisable — but that is largely academic, since those claims are invalidated. The critical asymmetry for a new, unaffiliated defendant: no estoppel applies to you at all, and no estoppel applies to claim 16 by anyone — Credit Acceptance Corp. v. Westlake Servs., 859 F.3d 1044, 1052 (Fed. Cir. 2017) (no IPR estoppel as to a claim for which no FWD issued). Your full invalidity toolkit is intact, and you can now deploy Zetts + Kaku (and Zetts + Kaku + Seo) as an IPR ground with the analogous-art fight already won by a prior petitioner — plus any new art, § 102 art, § 112 art, and subject-matter-eligibility challenges that were never in play. (Note the § 315(b) one-year bar from service of any complaint, and the disappearing runway given the 2027-06-28 expiration.)

Pattern signals. One IPR total on this patent, filed by a coordinated two-defendant coalition (Netflix jointly with Hulu — the same pair that also petitioned against sibling DivX patents, e.g., IPR2020-00614 on U.S. Pat. No. 7,295,673). No defensive aggregator is in the chain: the "Unified Patents Litigation Data" label in the Google Patents record is a data-source attribution, not a petitioner — do not misread it as Unified Patents having filed the IPR. DivX behaved as a persistent, aggressive patent owner: it litigated in C.D. Cal. and D. Del., appealed adverse PTAB outcomes, and pressed the '792 patent to a jury trial in March 2026 rather than settling. It also did not appeal the 2024-04-17 remand FWD as far as I can determine — an unusual posture that is consistent with the challenged claims being given up rather than defended further (again, unconfirmed).

Recommended next steps

1. If a demand letter or complaint cites claims 1, 5, 8, 9, 13–15, 18, 21, 22, or 23 — stop there. Those claims were held unpatentable in the April 17, 2024 FWD on remand: "Final Written Decision on Remand Determining All Challenged Claims Unpatentable." Obtain the paper from PTAB E2E (IPR2020-00646, Paper ~58) and confirm the per-claim table, then obtain the CAFC opinion, No. 22-1138 (2023-09-11) and the 2021 FWD (public copy) for the full record.
2. Pull the § 318(b) certificate from the '792 file wrapper (Patent Public Search / PatentCenter) to confirm cancellation of claims 1, 5, 8, 9, 13–15, 18, and 21–23 and to date-stamp the cancellation. This is the single most valuable document for a Rule 11 / § 285 posture against an assertion of canceled claims.
3. Confirm there was no appeal of the 2024-04-17 remand FWD by checking the CAFC docket and PTAB E2E (a notice of appeal would appear as a PTAB paper; aggregator data shows no activity after 2024-04-17). If an appeal exists, the cancellation is stayed pending resolution.
4. Claim 16 is the live target. It was never challenged in IPR (confirmed in the Dec. 2023 stay briefing: "Claim 16 (no IPR filed)"), so no § 315(e)(2) estoppel attaches to it, and it remains presumptively valid. If DivX (or any successor) asserts claim 16 against you, an IPR is available — subject to the § 315(b) one-year bar from service — and Zetts + Kaku is now an available, pre-vetted ground given the Board's remand holding that Kaku is analogous art. Buttress with the March 2026 jury non-infringement finding on claim 16 (C.D. Cal. No. 2:19-cv-1602, Dkt. 1197, judgment entered 2026-05-07).
5. No active proceedings mean no trial-stage milestones to track. There is no pending institution deadline, oral hearing, or statutory one-year FWD date on this patent. The only forward-looking event worth diarizing is the 2027-06-28 listed adjusted expiration; if a new case is filed against you in Delaware — Google Patents lists a 2024 Delaware action, No. 1:24-cv-01390, whose defendant I could not confirm — the § 315(b) clock starts on service.
6. Re-check the ODP feed. The "no AIA trial proceedings" result you were given is inconsistent with a proceeding that terminated on 2024-04-17; verify by respondent patent number (8472792) rather than by party name, since the petitioner of record is Netflix/Hulu, not Unified Patents.

I'll research the assignment record. Let me start with targeted searches on the USPTO assignment data and the parties involved.

Strong material. Now let me pin down the assignment reels/frames, correspondents, and the corporate history details.

Ownership & Assignment Record — U.S. Patent No. 8,472,792

Scope note on sources. I could not open the USPTO Assignment Center record directly in this session, and the recorded correspondent of record (the attorney/agent who filed each recording) was not retrievable from any source I reached. What follows reconstructs the chain from (a) the USPTO-derived assignment data mirrored in Google Patents' legal-events tables for the '792 and its DivX/Sonic siblings, (b) the USPTO-assignment chain-of-title recited by DivX itself in an IDS filed in App. No. 15/453,714 (reel/frame citations), and (c) party filings in DivX, LLC v. Netflix, Inc., No. 2:19-cv-01602 (C.D. Cal.), Dkt. 797-1 (Jan. 29, 2026). Reel/frame numbers I did not see stated on a '792-specific record are flagged as such. Nothing below is invented.

Inventors

• Applicant of record on the family's PCT filing (PCT/US2004/041667, filed 2004-12-08, published WO2005/057906) is DIVX, INC., 4780 Eastgate Mall, San Diego, CA 92121 — the same address and San Diego inventor cluster, which supports "employed by DivX, Inc." for all three.
• No fire-sale-precursor pattern. The assignment executions ran 2006-01-03 to 2006-01-28 — i.e., three of the three inventors assigned at/near filing, not on the way out the door. Butt and Fudge also continued to be named as co-inventors on later members of the same DivX family (e.g., U.S. Pat. No. 8,723,369, "Multimedia Distribution System for Multimedia Files Having Subtitle Information," listing "Ahsan Butt" and "Brian Fudge" alongside the broader DivX team), which is inconsistent with an "all inventors departed within 12 months" pattern. I found no evidence of any inventor departure, and I cannot determine employment end-dates from the sources retrieved.

Original assignee

DivX, Inc. (San Diego, CA). The issued patent's face reads DivX, LLC only because of the 2012 name change recorded in the chain (see timeline); the application was filed by DivX, Inc. on 2005-10-24.

• Product / line of business: DivX was an operating technology company — MPEG-4 video codec (the "DivX" format), the DivX Certified device program (press materials claim 300M+ certified consumer-electronics devices shipped), PC software, and content-protection/licensing services. The multimedia file structure claimed here — multiple audio tracks, multiple subtitle tracks, a menu chunk, an abridged frame index, and per-video-chunk DRM chunks — tracks the DivX Media Format / "DivX Ultra" generation of the format, i.e., the specification describes the assignee's own shipping file format rather than a paper design. (I did not retrieve a DivX product datasheet tying a specific SKU to these claims; that assessment rests on the specification's DivX-proprietary chunk set (DXDT, DMNU, nndr) and the applicant's business.)
• Current status of the original assignee: no longer operating under that name. Corporate chain (dates marked unverified were not re-confirmed in the sources retrieved for this task): DivX, Inc. → DivX, LLC (2012 conversion/name change, reel 032000/0334); DivX, Inc. was acquired by Sonic Solutions (agreement signed June 2010, closing reported ~Oct. 2010 at ~$322M); Sonic Solutions was acquired by Rovi Corp. (2011 — the March 25, 2011 JPMorgan credit facility below is consistent with this); the DivX business was sold by Rovi to NeuLion, Inc. (closing reported Jan. 2015); and per Netflix's filing in the C.D. Cal. case, the legacy DivX, LLC was renamed NLD, LLC on Feb. 13, 2018 — i.e., the original assignee's shell was retired the same week the '792 was transferred to Fortress. NeuLion in turn was acquired by Endeavor (reported 2018). No bankruptcy filed by any of these entities was found.

Assignment timeline

Reminder: the '792 sits inside a group-wide recording practice — the security agreement and releases below were recorded against large portfolios (All Media Guide, DivX, Sonic Solutions), not against this patent alone.

• 2006-01-03 to 2006-01-28 (executed) / recorded 2006-03-20 — Reel/frame not retrieved (portfolio inventor assignment; reel not exposed in the sources reached)

  • Conveyance: Assignment of Assignors' Interest
  • Assignor: Abou Ul Aala Ahsan Butt; Steve R. Bramwell; Brian T. Fudge
  • Assignee: DivX, Inc.
  • Correspondent: not retrieved
  • Context: ordinary employee/inventor assignment at filing — internal, not a transfer to a third party.

• 2011-03-25 — Reel 026026/0111 (reel/frame taken from the identical recorded transaction as it appears in the legal events of sibling DivX/Sonic patents; I did not open the '792-specific frame)

  • Conveyance: Security Agreement (secured-party lien, not an ownership transfer)
  • Assignors: All Media Guide, LLC; DivX, LLC; Sonic Solutions LLC
  • Assignee: JPMORGAN CHASE BANK, N.A., as Collateral Agent
  • Correspondent: not retrieved
  • Context: securitization — a Rovi/Sonic-group credit facility blanket-encumbering the group's patents, including the '792.

• 2012-09-15 (effective; recording date not retrieved) — Reel 032000/0334

  • Conveyance: Change of Name
  • Assignor: DivX, Inc.
  • Assignee: DivX, LLC
  • Correspondent: not retrieved
  • Context: internal corporate reorganization / change of name only — no substantive transfer of rights.

• 2013-06-07 (effective; '792 event table shows 2013-06-11) — Reel 030591/0534

  • Conveyance: Release by Secured Party
  • Assignor: JPMorgan Chase Bank, N.A., as Collateral Agent
  • Assignee: All Media Guide, LLC; DivX, LLC; Sonic Solutions LLC
  • Correspondent: not retrieved
  • Context: lien release on the group credit facility (partial — see next two entries).

• 2013-11-22 — Reel 032293/0557

  • Conveyance: Assignment of Assignors' Interest
  • Assignor: DivX, LLC
  • Assignee: SONIC IP, INC. (San Diego, CA; address of record 4790 Eastgate Mall, Suite 200, San Diego, CA 92121)
  • Correspondent: not retrieved
  • Context: internal reorg / IP-holding-company consolidation — DivX's patents were moved into a dedicated subsidiary (Sonic IP, Inc.) under the then-parent (Rovi/Sonic). This is the first time the '792 leaves the original operating entity.

• 2014-02-24 or 2014-03-31 (effective; '792 event table shows 2014-04-09) — Reel 032645/0559

  • Conveyance: Release by Secured Party
  • Assignor: JPMorgan Chase Bank, N.A., as Collateral Agent
  • Assignee: DivX, LLC
  • Correspondent: not retrieved
  • Context: final lien release tied to repayment/replacement of the group facility. ⚠️ Date inconsistency flagged: the '792 event date (2014-04-09), the sibling records' effective dates (2014-02-24 and 2014-03-31) do not line up; treat the exact execution date as unconfirmed.

• 2014-07-02 (effective; '792 event table shows 2014-07-22) — Reel 033378/0685

  • Conveyance: Patent Release
  • Assignor: JPMorgan Chase Bank N.A., as Collateral Agent
  • Assignee: All Media Guide, LLC; DivX, LLC; Sonic Solutions LLC
  • Correspondent: not retrieved
  • Context: second, portfolio-level release of the same security interest (clean-up recording).

• 2018-02-12 (effective) / recorded 2018-02-12; '792 event table shows 2018-02-13 — Reel 045310/0020

  • Conveyance: Assignment of Assignors' Interest
  • Assignors: SONIC IP, INC.; DIVX, LLC
  • Assignee: DIVX CF HOLDINGS LLC (New York)
  • Correspondent: not retrieved
  • Context: transfer-to-asserter / portfolio carve-out. DivX CF Holdings LLC was created Dec. 15, 2017 by Fortress Investment Group; the purchase agreement with NeuLion covering "certain patent assets and related patent rights" is dated Dec. 18, 2017 (NeuLion 8-K exhibit filed on Justia Contracts). Per Dkt. 797-1, signatory Constantine Dakolias was then Co-CIO of Fortress Credit Funds. This is the effective NPE formation event — hundreds of patents and applications moved in one recording.

• 2018-02-12 / 2018-03-05 (effective; '792 event table shows 2018-03-05) — Reel 045498/0560 (listed as reel 045498/0560 for a DivX CF Holdings LLC document in DivX's own IDS chain-of-title for App. No. 15/453,714; the IDS chain shows the transferee field blank, and the reel number is ~March 2018 — consistent with the name-change recording)

  • Conveyance: Change of Name
  • Assignor: DivX CF Holdings LLC
  • Assignee: DIVX, LLC (current owner)
  • Correspondent: not retrieved
  • Context: change of name only — the Fortress acquisition vehicle adopts the target's name, three weeks after acquiring the portfolio.
  • ⚠️ Two versions of this date exist: Dkt. 797-1 says the certificate of amendment was signed Feb. 12, 2018 ("the same day" as the assignment), while the USPTO-derived event table gives 2018-03-05. Both cannot be the recording date; treat as ~Feb–Mar 2018.

• 2019-02-28 — Reel/frame not retrieved (formality filing; no reel exposed)

  • Conveyance: Change of Principal Place of Business
  • Assignor / Assignee: DivX, LLC (no change in owner)
  • Correspondent: not retrieved
  • Context: administrative address update only — DivX, LLC's address of record moved to 4350 La Jolla Village Drive, Suite 950, San Diego, CA 92122 (address appears on the EP counterpart, EP2972960).

No later recordings were found. As of the March 25, 2026 Netflix verdict and the May 7, 2026 final judgment, DivX, LLC was still the party asserting the '792, and the 2024 Top Victory Investments v. DivX declaratory-judgment matter (D. Del. 1:24-cv-01390) turns on a DivX covenant covering the '792 — i.e., DivX, LLC remains the owner of record. No assignment to RPX, AST, Unified Patents, or LOT Network appears anywhere in the chain.

Timeline diagram

timeline
    title Ownership of US 8472792
    2005 : Application filed by DivX Inc
    2006 : Inventors assign to DivX Inc
    2011 : JPMorgan security interest recorded
    2012 : DivX Inc renamed DivX LLC
    2013 : Lien released by JPMorgan
         : Assigned to Sonic IP Inc
    2014 : Two secured party releases recorded
    2018 : Assigned to DivX CF Holdings LLC
         : DivX CF renamed DivX LLC
    2019 : First suits filed against Netflix and Hulu
    2026 : Jury finds no Netflix infringement

NPE / troll-pattern signals

1. Shell-entity transfer — PRESENT. Reel 045310/0020 (effective 2018-02-12): the portfolio moved from Sonic IP, Inc. and DivX, LLC to DIVX CF HOLDINGS LLC, an entity created by Fortress on Dec. 15, 2017 — nine weeks before the assignment — and the "Holdings" suffix appears in the recorded assignee name itself. Concrete non-name evidence: (i) Dkt. 797-1 (C.D. Cal., Jan. 29, 2026) states Plaintiff DivX "has no engineering department, employs only one engineer, is led by an attorney, and engages in litigation and patent licensing as its primary business"; (ii) it is a pass-through — 97% owned by DivX CF Investors LLC, whose only members are Fortress entities; (iii) Fortress personnel must constitute a majority of its management committee and are expressly relieved of fiduciary duties to it. That is documented single-purpose, fund-owned, licensing-only ownership — not naming alone.

2. Known asserter in the chain — PRESENT. Unified Patents states on its own PATROLL page: "The patent is owned by DivX, LLC, a subsidiary of well-known NPE, Fortress Investment Group" (Unified Patents, "$1,500 Awarded for DivX '792 prior art," Mar. 18, 2021), and Unified itself filed IPRs against DivX (e.g., IPR2021-01476, IPR2021-01418). DivX, LLC is also a serial plaintiff (C.D. Cal. 2:19-cv-01602 and -01606; ITC Inv. No. 337-TA-1297; D. Del. 1:21-cv-01828 and 1:24-cv-01390). Fortress is not one of the named legacy NPEs in the prompt's list (Acacia, Marathon, IV, Wi-LAN, etc.), so I am grounding this signal on Unified's published characterization and on the assertion record rather than on a directory match to your enumerated list.

3. Repeat correspondent across the chain — UNCLEAR / not established. I could not retrieve the correspondent of record for any of the nine recordings above; no source I reached exposes that field for these reel/frames. I want to be explicit that the attorneys visible in the litigation record are not evidence of this signal: Kenneth J. Weatherwax, Lowenstein & Weatherwax LLP (1880 Century Park East, Suite 815, Los Angeles) is DivX's PTAB counsel of record and Harper Batts, Sheppard Mullin (Reg. No. 56,160) is petitioners' counsel — these are IPR counsel, and nothing links either to an assignment recording. This signal should be re-run against the Assignment Center correspondent field; it is the single highest-value open item in this analysis.

4. Cascading transfers — UNCLEAR. The chain has two substantive transfers (2013-11-22 → Sonic IP, Inc., reel 032293/0557; 2018-02-12 → DivX CF Holdings LLC, reel 045310/0020) separated by more than four years, so the strict "<24 months through chained LLCs" test fails. However, the Feb. 2018 sequence — assignment on Feb. 12, name change within ~3 weeks, and the legacy DivX, LLC renamed NLD, LLC on Feb. 13 — is the classic clean-title packaging move. I cannot call a cascade without the correspondent data.

5. Pre-litigation transfer — NOT PRESENT (fails the 6-month test), but note the pattern. The transfer to DivX CF Holdings LLC was effective 2018-02-12 (reel 045310/0020); the first suits naming the '792 (DivX v. Netflix and DivX v. Hulu) were filed 2019-03-05 — ~12.6 months later, outside your 6-month window. The patent did, however, sit unasserted for its entire 2006–2018 life and was asserted within 13 months of moving to the Fortress entity, so the acquisition-to-assertion purpose is well supported even though the timing test is not met.

6. Bankruptcy fire-sale — NOT PRESENT. No Chapter 7/11 assignor appears. The 2018 transfer was a negotiated asset sale out of NeuLion (per the Dec. 18, 2017 purchase agreement and Dkt. 797-1), and NeuLion was subsequently acquired by Endeavor — a going-concern M&A path, not a bankruptcy estate sale.

7. Privateering — UNCLEAR (partial match). What is documented: the patents were carved out of the DivX operating business and sold to a financial sponsor's vehicle (NeuLion → DivX CF Holdings LLC, reel 045310/0020), after which the acquirer asserted them against Netflix, Hulu, TCL, and others. What is not documented: any assertion by Fortress on the original operating company's behalf against that company's competitors — the classic privateering element. The seller (NeuLion/DivX business) is not the beneficiary. I therefore do not score this as present.

8. Defensive aggregator (anti-NPE) — NOT PRESENT. The chain terminates at DivX, LLC, the Fortress-controlled plaintiff, which is actively asserting the '792 as of the March 2026 Netflix trial and the 2024 Top Victory covenant dispute. The patent has not been neutralized.

Verdict

NPE — high confidence.

Two independent strong signals drive this, both grounded in specific reel/frame entries: (1) the recorded transfer at Reel 045310/0020, effective 2018-02-12, moved the '792 from the operating DivX/Sonic IP entities to DivX CF Holdings LLC, a New York LLC created by Fortress Investment Group on 2017-12-15 solely to hold the purchased portfolio and renamed DivX, LLC weeks later (reel 045498/0560) — with the fund's own filing conceding an entity that has "no engineering department, employs only one engineer, is led by an attorney," and is 97% pass-through to Fortress; and (2) Unified Patents publicly identifies the current owner, DivX, LLC, as "a subsidiary of well-known NPE, Fortress Investment Group" and litigated IPRs against it, while DivX, LLC simultaneously maintains five-plus assertion matters including two 2019 C.D. Cal. suits and ITC Inv. No. 337-TA-1297. Signals 5 (pre-litigation transfer) and 7 (privateering) do not cleanly apply on the evidence — the transfer preceded suit by ~13 months and the seller was not the assertion beneficiary — and signal 3 (repeat correspondent) is unresolved because the Assignment Center correspondent field was not retrievable here.

Verify at: https://assignmentcenter.uspto.gov/ (search patent number 8472792) or https://assignment.uspto.gov/patent/index.html — specifically to capture the correspondent of record for reel/frame 045310/0020 and 045498/0560, and to confirm the recorder for the 2011 lien and 2013–2014 releases (reels 026026/0111, 030591/0534, 032645/0559, 033378/0685), which I transcribed from sibling-patent legal events rather than from '792-native records.

I'll research the prior art of US 8,472,792. Let me start with targeted searches on the patent's citation records.

Let me get the full citation lists and the specific IPR references.

I have the key IPR references. Let me now pin down the specific patent numbers for the IPR and prosecution references.

Prior Art Analysis — U.S. Patent No. 8,472,792 B2 ("Multimedia distribution system")

Note on the tool limit: I hit the search-step ceiling before I could run every query I intended. Below I distinguish carefully between (a) what I retrieved and verified this session, (b) what comes from the previously-generated sections of this analysis, and (c) what I could not verify and am therefore not asserting. I have not auto-corrected any identifier, and I have not filled gaps with recollection — where a patent number for a cited reference was not retrieved, I say so rather than guess.

Sourcing caveat (important): I could not query USPTO PatentCenter/PatFT directly. The front-page citation data below came from (i) Google Patents' page for US 8,472,792 and (ii) a Sumobrain mirror of the printed patent front page ("Domestic Patent References"), plus (iii) PTAB records for IPR2020-00646. Treat the examiner-cited list as partial — the Sumobrain scrape truncated partway through the "M" references.

0. Cross-reference flags against the earlier sections

Before the prior art, two corrections/refinements to what was generated earlier. These matter for prior-art mapping:

1. Granted claim 1 is narrower than the pre-grant publication suggested. The previously-generated "Patent summary" §4 correctly flagged that it could not retrieve granted claim text and summarized claim 1 from US 2006/0129909 A1 (file with first index + separate second index). The IPR2020-00646 petition element breakdown (retrieved this session) shows granted claim 1 additionally requires:

   • [1f] "the separate second index located prior to the series of encoded video frames and the first index"; and
   • [1g] "the first and second indexes enabling trick play functionality."

   These two limitations are the entire battleground of the prior art fight, and they do not appear in the pre-grant summary. Build on the granted version.

2. Claim-number mapping differs between the publication and the grant. The earlier section listed claim 5 = encoder, claim 9 = decoder (per the pre-grant publication). The IPR petition's element-by-element mapping ([9a] "An encoder for encoding a multimedia file comprising at least one video track and at least one audio track, the encoder comprising: … [9c] at least one sequence of encoded video frames and … [9d] a full index … [9e] wherein the processor is configured to generate an abridged index") shows claim 9 is the encoder claim in the granted patent. The two are inconsistent; the IPR record is the better evidence of the granted numbering. Flagging rather than reconciling.

1. The substantive prior art: the IPR2020-00646 / prosecution references

These are the references that actually drove validity. Note at the outset: no tribunal applied any of these as a §102 anticipation reference. The Board's analysis, and the Federal Circuit's 2023 opinion, were entirely §103 obviousness (Zetts as primary reference, Kaku as secondary). The §102 column below is therefore a screening-level mapping I am supplying, not a finding of record — and I have not read the full specifications of these references, only the passages quoted in the PTAB papers.

1a. Kaku — U.S. Patent No. 6,671,408 B1 ✅ verified

1b. Zetts — patent number not verified

1c. Abbott — patent number not verified

1d. Hallberg — patent number not verified

1e. A possible companion reference surfacing in the same search

U.S. Patent No. 6,728,471 B1 — an "image reproducing apparatus" (digital camera, CCD imager 10) describing normal vs. high-speed reproduction, generating decompression instructions either at a predetermined interval or each time one screen completes decompression. This is highly similar in subject matter to Kaku and may be a same-family/Sanyo reference. I did not verify its inventor, assignee, filing date, or whether it is cited in '792 — do not cite it as prior art to '792 without checking the front page.

2. Examiner-cited "Domestic Patent References" on the face of '792 (partial list)

These are the references the examiner listed under "References Cited." I retrieved the list only down to 1998 (it truncated), and I have only the titles — I have not read the specifications. The §102 column is therefore a coarse screening map.

3. Direct answer: "which claims does each reference potentially anticipate under §102?"

Read this first. Anticipation requires every claim element disclosed in a single reference, arranged as claimed. Nothing in the record of '792 does that for the independent claims. The record shows only §103 combinations:

• IPR2020-00646 Ground 1: Zetts + Kaku against claims 1, 5, 8, 9, 13–15, 18, 21–23 — as obviousness.
• Prosecution: examiner rejections over Abbott + Hallberg, and separate consideration of Kaku.

The mapping below is my own §102 screening — treat as a research lead, not a conclusion.

4. Bottom line and open items

• Most relevant prior art to US 8,472,792: the Zetts + Kaku combination used in IPR2020-00646. Per the previously-generated sections (which I did not re-verify this session), the Board's Final Written Decision on remand (April 17, 2024) held all challenged claims unpatentable as obvious over that combination. If that holds, the practical prior-art answer to your question is not "which reference anticipates" but "the claims were cancelled on §103."
• No §102 anticipation of the granted independent claims was found by the examiner, the Board, or the Federal Circuit. The Board's original FWD rejected the §103 grounds for a procedural reason (failure to articulate a field of endeavor), and the Federal Circuit affirmed the "not reasonably pertinent" finding while vacating only the field-of-endeavor finding — Netflix, Inc. v. DivX, LLC, No. 22-1138 (Fed. Cir. Sept. 11, 2023).
• The nearest thing to an anticipation-type disclosure is Kaku for the AVI-index/frame-number/evenly-spaced-frame elements, and Zetts for the "subset index ahead of the media data" element. Neither alone reaches granted claim 1.
• I could not obtain (i) the patent numbers for Zetts, Abbott, or Hallberg, (ii) the filing dates for Kaku's §102(e) date, (iii) the truncated remainder of the examiner-cited front page (including any foreign patent documents and NPL), or (iv) a direct look at the USPTO PatentCenter "References Cited" tab. Each of these should be pulled from PatentCenter / Google Patents "References Cited" before this analysis is relied on, since the §102 analysis — particularly Kaku's date problem (issued 12/30/2003 vs. priority 12/08/2003) — turns on facts I could not confirm here.
• Reminder on the earlier section's open question: whether the April 2024 remand FWD was appealed, and the post-April-2024 enforceability status, remains unverified and is the most consequential gap for the prior-art story.

I'll ground this analysis in the actual prior-art record for '792. Let me pull the cited references and the IPR art first.

Key art identified. Let me pin down the remand outcome and the granted claim text.

Now let me confirm the identities of the three IPR references and the remand outcome.

Obviousness Analysis — U.S. Patent No. 8,472,792 B2 (35 U.S.C. § 103)

0. Source note, scope, and contradictions to flag

On "the Prior Art section of this page." The authoritative page text supplied to me contains the Google Patents prior-art taxonomy fields — Prior art keywords: chunk, encoded video, index, file, frame; Prior art date 2003-12-08 — and the litigation/PTAB/CAFC links, but it does not contain a rendered "References Cited" or "Prior Art" listing. I therefore reconstructed the operative prior-art record from the actual challenge record: the IPR2020-00646 Petition and Final Written Decision, the Federal Circuit opinion in No. 22-1138, and prosecution-history excerpts reproduced in those papers. Where I rely on something I could not verify this session, I say so.

⚠️ Contradiction with the earlier "Patent summary" section — material. The earlier section's §4 claim table was built from pre-grant publication US 2006/0129909 A1 and states claim 1 is a multimedia file claim, claim 5 an encoder, claim 9 a decoder, claim 12 a method of encoding, and claim 16 a method of decoding/seeking. In the granted patent this is wrong. The PTAB and the Federal Circuit both quote the issued claim 1 (Ex. 1001, 13:7–22 / 51:31–45) as a decoder claim, and the Board states expressly: "Petitioner challenges claims 1, 5, 8, 9, 13–15, 18, and 21–23, of which claims 1, 9, and 15 are independent claims." There are no method claims in the issued patent, and claim 16 is a dependent decoder claim (locate reference information in the complete index using the abridged index). The earlier caveat that the pre-grant text "may differ in numbering and wording" was correct — and it does differ. The corrected claim set is set out in §3 below and should supersede §4 of the earlier summary.

(Sources: PTAB FWD, IPR2020-00646; CAFC No. 22-1138 opinion; RPX litigation document reproducing claims 1–23.)

1. Governing framework and the effective filing date

Because the '792 patent's effective filing date precedes March 16, 2013, pre-AIA § 103(a) governs — the Board said so expressly, citing the AIA transition. (FWD fn. 1.)

Effective filing date is in dispute and matters here. The '792 patent claims a Dec. 8, 2003 priority chain, but Petitioner contended the challenged claims are not entitled to it, because the examiner found the earlier applications "fail[ed] to provide adequate support or enablement … for … the 'first index and a second index' of claim 1." Petitioner argued the effective date is the Oct. 24, 2005 CIP filing date. (Ex. 1002 at 0559; Petition quoting same.) Petitioner added: "Regardless, the challenged claims are obvious even with a 2003 priority date."

Consequence: the safest § 103 case runs off Oct. 24, 2005, under which all three asserted references are unambiguously § 102(b)/§ 102(e) art. If Patent Owner successfully claims Dec. 8, 2003, then Kaku (issued Dec. 30, 2003 — after that date) survives only via its own US filing date, and Seo (US 8,286,213, issued Oct. 9, 2012) becomes more vulnerable as § 102(e) art. Verification item: confirm Kaku's and Seo's US filing/priority dates against the references themselves.

2. Level of ordinary skill (POSITA)

The Board's framing, consistent with the record: a POSITA by 2003/2005 would have had a bachelor's in electrical engineering, computer science, or equivalent, plus about two years of experience with multimedia container formats and video compression, or equivalent education/experience. Critically, the FWD records that MPEG standards were commonly used to compress digital video, AVI was "a widely-used file format" for storing encoded audio and video, "trick play" was a known term of art, and scene-skipping to I-frames/GOP boundaries was conventional. (Ex. 1003 ¶¶ 14–23.) This is the level of skill the analysis below assumes.

3. The issued claim set (independents 1, 9, 15)

Which claims the Board actually adjudicated: 1, 5, 8, 9, 13–15, 18, 21–23. Claims 2–4, 6–7, 10–12, 16–17, 19–20 were never challenged in the IPR — DivX itself pointed this out, noting the tag→first-index cross-reference "is recited in unchallenged dependent claims, such as claim 4." Any § 103 challenge to those claims is therefore untested and must be built independently (see §9).

4. The prior art

The last two matter because the '792 specification itself concedes that "the 'idx1' chunk can be implemented as specified in the AVI format" (22:19–27) and that its chunks "are defined as part of the AVI file format" (5:33–6:17). That is a specification admission that the first index — the "location and characteristics of each encoded video frame" — is conventional AVI idx1.

5. Combination A (primary): Zetts in view of Kaku — § 103(a)

This is the combination Petitioner actually ran, against claims 1, 5, 8, 9, 13–15, 18, 21–23.

Claims 9 and 15 are met by the same combination viewed from the encoder side (Zetts teaches an MPEG encoder that creates and inserts the table before the stream; a POSITA would implement the encoder to produce the claimed file) and by the decoder side (Kaku's AVI-file reader + Zetts's table lookup, with playback starting from the located frame).

Dependents:

• 5 / (6) — evenly spaced tags: Zetts's table has an entry per GOP; Kaku organizes evenly spaced groups (15 frames/1 s). Petitioner's rationale: by 2003, evenly spaced GOPs with a fixed I-frame interval were the norm — commonly an encoder setting — because they trade compression ratio against access granularity and improve fast-forward/rewind by distributing intra-coded frames predictably.
• 8 / 14 / 23 — trick play = FF/rewind/scene skipping: MPEG/DVD trick play via I-frames was conventional; Zetts FIG. 12; Seo.
• 13 — insert key frames on scene change or when a threshold interval is exceeded: Zetts teaches scene-change detection (catalogers performing "scene change analysis and creating a bitmap … of the first frame after each cut or major scene transition"), and Kaku's fixed 15-frame/1-second grouping supplies the threshold-interval branch.
• 18 / 21 / 22 — desired frame → closest frame in the abridged index → display; playback without the complete index: Zetts's "nearest preceding GOP header" binary search does exactly this, and because the table is at the front of the file, playback begins "once the last GOP offset table padding packet has been received" — i.e., without the end-of-file idx1. Note the '792 patent concedes the identical point: front-loading the index "can enable a device to start playing … prior to the downloading of the 'idx1' chunk" (16:10–15).

6. Combination B: Zetts + Kaku + Seo — claims 5, 18, 21

The Board adopted Petitioner's reasoning here in full and expressly found a rational motivation and reasonable expectation of success:

• Seo supplies periodic I-frames (reinforcing claim 5's "evenly spaced").
• Seo supplies the "requested frame → not an I-frame → stream from the closest I-frame" teach- ing that directly maps to claim 18's "determine the encoded video frame that is closest to the desired video frame in the abridged index."
• Motivation: Zetts and Seo are complementary for scene-skipping — Zetts marks where the I-frames are; Seo teaches what to display when the requested frame isn't one. Seo further teaches that some users prefer faster playback from the nearest I-frame because it "feels more responsive" — a direct, rational, user-preference-based reason to combine.
• Reasonable expectation of success: implementing the user-preference option was "a simple matter of implementing user preferences using conventional programming."

7. Combination C (recommended alternative): the AVI/RIFF specification + Zetts, with Kaku used only evidentially

This framing hedges the one legal weakness in the record. The Federal Circuit affirmed the Board's finding that Kaku is not reasonably pertinent to the '792 problem (trick play in downloaded/streamed multimedia), because Kaku's problem was image compression to fit a digital camera's limited memory. So Kaku can only be carried by the field-of-endeavor prong — which the court vacated and remanded, and which (per the earlier-generated section, §5a, unverified by me this session) the Board resolved against DivX in April 2024.

A petitioner can sidestep that fight entirely: the AVI/RIFF format itself is a § 102(b) printed publication, and the '792 specification admits idx1 is AVI-format-specified, that the file is AVI-based, and that AVI is extensible. Paired with Zetts (which supplies the front-of-file abridged index and the express motivation), the combination needs no Kaku for the container teaching at all; Kaku becomes corroborating evidence that MPEG-in-AVI indexing was known. Hallberg US 2006/0093320 (if its § 102(e) date holds) independently supplies the "first index = location + characteristics of each frame" limitation, again reducing dependence on Kaku.

8. Why a POSITA would have combined these references

1. Same field of endeavor. Both the '792 patent and the container references are directed to AVI files and to the encoding/decoding of multimedia files — the two fields the Federal Circuit expressly said were "fairly presented" and must be decided on remand. The patent's own Background frames it as "encoding, transmission and decoding of multimedia files," and its embodiments are AVI.
2. Express motivation in Zetts. Zetts states the table is compact (~3 bytes/entry), is placed near the start of the file so a user can jump to a desired location, and avoids maintaining a separate index file — a direct efficiency teaching.
3. Compatibility, not replacement. The AVI idx1 at the end is retained; adding Zetts's table does not disable it. Dr. Reader explained the added structure "can be exploited by devices that are aware of [this] capability, but doesn't prevent the file being played on legacy playback devices"; Dr. Bajaj conceded "vanilla" decoders simply ignore it. There is therefore no loss of function and no teaching-away.
4. Extensibility is the whole point of RIFF/AVI. Unknown chunks are ignored by compliant decoders — so adding a chunk before movi is a low-risk, predictable modification "well within the level of ordinary skill."
5. Known technique / predictable result (KSR). Indexing I-frames/GOP boundaries for trick play was conventional (DVD trick play; MPEG random access only at GOP headers); relocating an index to the front of a file to enable early seeking is a predictable use of a known technique to improve a known technology. Tightening the index granularity (every GOP → evenly spaced subset) is "a simple matter of … conventional programming."
6. Seo adds a user-preference rationale for the closest-frame behavior of claims 18 and 21.

9. My independent § 103 read on the untested claims

Because claims 2–4, 6–7, 10–12, 16–17, 19–20 were not in the IPR, no tribunal has ruled on them. On the face of the art:

• Claim 2 (tag = file location + frame number): Kaku reads image data "based on a frame number," and Zetts's entries carry address offsets — likely obvious; the strongest single-reference hook is Kaku + AVI idx1 (offset + per-chunk identity).
• Claim 3 (tag → accompanying audio portion): AVI interleaving plus Kaku's 1 s of audio per 15 frames makes this a near-trivial design consequence, but the "portion … accompanies the encoded video frame" correspondence is a mapping I would want to see worked element-by-element rather than asserted.
• Claim 4 (tag cross-references the first index): this is the "index offset" idea; AVI idx1 is an ordered, per-chunk table, so an entry pointing into it is straightforward — but this claim was deliberately left out of the IPR, which may reflect Petitioner's own confidence assessment.
• Claim 6 (≥ 10 seconds apart) is the thinnest mapping: neither Zetts nor Kaku teaches a 10-second tag spacing. It would rest on a POSITA selecting index granularity to a desired seek latency — defensible under KSR, but the weakest link in the set.
• Claims 10–12 (encoder generates a complete index; entries cross-reference it; sequence numbers; audio reference) track claims 4/2/3 from the encoder side and rise or fall with them.
• Claims 16, 19, 20 (locate reference info in the complete index via the abridged index; use the abridged reference to find the frame; closest preceding frame) map closely to Zetts's binary search for "the nearest preceding GOP header." Claim 20 in particular is close to a verbatim description of Zetts. Note: claim 16 was asserted against Netflix and the jury found no infringement — which says nothing about validity.
• Claim 22 ("without receiving the complete index") is met almost textually by Zetts's front-placed table, and by the patent's own admission about avoiding the idx1 download.

10. Anticipated rebuttals and how they fare

Objective indicia: I found no evidence of unexpected results, long-felt need, industry praise, or licensing-due-to-validity in the records retrieved. Commercial success of the DivX platform is plausible but no nexus showing is in the record. Absent that, the Graham secondary-considerations factor is neutral-to-negative for Patent Owner.

11. Bottom line

• Claims 1, 5, 8, 9, 13–15, 18, 21–23: a strong § 103 case over Zetts + Kaku (§ 103(a)), with Zetts + Kaku + Seo as a reinforced ground for claims 5, 18 and 21. The Board adopted these grounds on the merits on remand (per the earlier-generated section — the April 17, 2024 remand FWD is reported as holding all challenged claims unpatentable; I could not independently re-verify that in this session, since my searches returned the Nov. 30, 2023 remand procedure order and the joint list of issues on remand, not the remand FWD itself). The single most important open item — flagged in the earlier section and still open — is whether the remand FWD was appealed.
• Claims 2–4, 6–7, 10–12, 16–17, 19–20: unadjudicated. My independent view is that most are obvious over the same art (claim 20 is essentially Zetts's "nearest preceding GOP header"), with claim 6's 10-second spacing and claim 3's audio-correspondence as the two limitations warranting element-by-element work before asserting a § 103 case.
• The recommended framing for any future challenge is a container-centric primary reference set (AVI/RIFF specification + Zetts, with Hallberg as backup for the first index), using Kaku only as corroboration — that neutralizes the analogous-art fight that has consumed this patent's validity litigation since 2020.

Verification checklist

1. Pull the certified claim set from USPTO PatentCenter for claims 1–23 of the '792 patent (all quotations above come from PTAB/CAFC/docket reproductions of Ex. 1001).
2. Confirm Kaku's (6,671,408) and Seo's (8,286,213) US filing and priority dates.
3. Confirm the April 2024 remand FWD and any appeal/appeal deadline — this determines present enforceability.
4. Confirm Hallberg US 2006/0093320's § 102(e) date before relying on it.
5. Note the date discrepancy carried from the task inputs (prompt says April 26, 2026; session header says 2026-09-22; the earlier section reports a May 7, 2026 final judgment) — I report dates as they appear in the retrieved sources rather than resolving it.