US 8634462

I'll verify the patent record and check for any current litigation activity before summarizing.

US Patent 8,634,462 — Verification Summary

I confirmed the record against the Google Patents full text for US8634462 and cross-checked with Justia and the litigation record. One important caveat up front: I was cut off before completing my final verification search on 2025–2026 docket activity, so my litigation-status statements below rest on the sources I did retrieve and are flagged where incomplete.

1. Bibliographic data (as recorded)

Note: the 2008 filing date sits inside the 12‑month Paris window from the 2007 priority, consistent with a national-stage/continuation-type filing of the earlier US application; I do not have the underlying PCT number in the retrieved record, so I am not asserting one.

2. Abstract (verbatim)

"Method for coding a video signal using hybrid coding, comprising: reducing temporal redundancy by block based motion compensated prediction in order to establish a prediction error signal; performing quantization on samples of the prediction error signal or on coefficients resulting from a transformation of the prediction error signal into the frequency domain to obtain quantized values, representing quantized samples or quantized coefficients respectively; calculating a quantization efficiency for the quantized values; calculating a zero efficiency for a quantization, when the quantized values are set to zero; selecting the higher efficiency; and maintaining the quantized values or setting quantized values to zero, for further proceeding, depending on the selected efficiency."

3. Independent claims — plain-language overview

Independent claim 1 (the verified principal independent claim). This is a method for coding video using hybrid coding, and in the granted version it is narrower than the published application's claim 1. Steps:

1. Reduce temporal redundancy using block-based motion-compensated prediction to produce a prediction error signal.
2. Quantize either the samples of that prediction error signal directly, or coefficients obtained by transforming it into the frequency domain — producing "quantized values" (quantized samples or quantized coefficients). Critically, the claim specifies the prediction error signal includes a plurality of subblocks, each with a plurality of quantized values.
3. Calculate a first quantization efficiency for the quantized values of at least one subblock.
4. Set the quantized values of that subblock to all zeroes.
5. Calculate a second quantization efficiency for that subblock while all values are zero.
6. Compare the two efficiencies and pick the higher.
7. Output the subblock either with its original quantized values (if the first efficiency is higher) or with it zeroed out (if the second is higher).

In plain terms: instead of using the H.264/AVC-style "importance sum" heuristic (Tables 1–2 in the spec) to decide whether to blank a block, the coder explicitly computes a cost/benefit both ways — keep the quantized data, or zero it — and takes whichever scores better. Claim 1 as granted is tied to the subblock-level comparison.

Verified dependent claims (illustrative): claim 2 adds the macroblock-level two-stage decision (per-subblock comparison over all subblocks, then an overall quantization efficiency vs. an overall zero efficiency for the whole macroblock); claim 3 recites that efficiency is based on a cost function; claim 4 the cost function is rate-distortion cost; claim 5 rate distortion is distortion plus weighted required rate; claim 6 gives C_j = D_j + L·R_j (D = distortion, R = required rate, L = Lagrange parameter, j = subblock index); claim 7 the distortion measure is summed squared quantization error or mean absolute quantization error.

Uncertainty flag: The '462 is a multi-facet disclosure — the specification and summary of invention also describe (a) a coder with quantization means plus control means for comparing quantization vs. zero efficiency, (b) a decoder with adaptive control means and a scan control unit, (c) a switch between spatial-domain and frequency-domain coding selected by rate-distortion cost, (d) an adaptive scanning scheme derived from a prediction image / prediction error image / motion vector, (e) subjectively-weighted or MSE-optimized quantizer design for spatial-domain samples (FIG. 8, FIG. 9, Table 3), and (f) a data signal / bitstream with side-information flags (Slice_FD_SD_coding_flag, MB_FD_SD_coding_flag, FD_or_SD_flag, Tables 4–6). I could not retrieve the complete granted claim set verbatim (e.g., the exact wording and numbering of any independent coder, decoder, or data-signal claims). I therefore describe claim 1 with confidence and treat the other facets as embodiments/aspects rather than asserting their exact claim language. Anyone relying on claim scope should pull the granted claims from the USPTO Patent Center or the full-text PDF.

Also note the divergence between the published application claim 1 (US 2010/0189180 A1), which was broad — no subblock limitation, just "calculating a quantization efficiency … calculating a zero efficiency … selecting the higher efficiency" — and the granted claim 1, which added the subblock/subblock-zeroing structure. That narrowing matters for any claim-scope analysis.

4. Technical context

The patent addresses the "second quantization" step in H.264/AVC, where the reference software zeroed out blocks using a threshold on an "importance" sum I_j (Tables 1 and 2), with thresholds of 5 (per 8×8 block) and 6 (per macroblock). The inventors' premise is that this heuristic is "unsatisfactory" because it does not weigh the actual bit cost against the actual distortion. Their contribution is to compute explicit rate-distortion efficiencies for both options (keep vs. zero) at subblock level, then repeat the comparison at macroblock level, choosing the lower-cost outcome. Related aspects cover choosing between spatial-domain and frequency-domain coding of the prediction error, adaptive scan ordering, and quantizer design for spatial-domain samples.

5. Litigation / docket status

• The '462 was one of eight Realtime patents in the consolidated MDL: In re Realtime Adaptive Streaming LLC Patent Litig., 326 F. Supp. 3d 1374 (J.P.M.L. 2018), which lists U.S. Patents 7,386,046; 8,867,610; 8,929,442; 8,934,535; 9,762,907; 9,769,477; 8,634,462; and 9,578,298 (I am reading the '462 exactly as listed there). Accused products were alleged to implement standards such as H.264 and H.265.
• District-court matters listed on the Google Patents record citing this patent include D. Del. 1:17-cv-01519, 1:17-cv-01520, 1:17-cv-01692, 1:17-cv-01693; D. Colo. 1:17-cv-02869 and 1:18-cv-00927; E.D. Tex. 6:17-cv-00591; D. Mass. 1:18-cv-10355; C.D. Cal. 2:17-cv-07611 and 2:18-cv-09344; plus the Federal Circuit entry 21-2268.
• The '462-specific infringement allegations appear in the Netflix complaint analysis (D. Del. 1:17-cv-01692), where the '462 claim 1 elements were mapped to HEVC/H.265 block-based motion compensation and rate-distortion-optimized quantization.
• Important correction: Federal Circuit Appeal No. 21-2268, Realtime Adaptive Streaming LLC v. Sling TV, L.L.C. (D. Colo. 1:17-cv-02097), decided May 11, 2023 (per curiam, affirmed), concerned the '610 patent and § 101 eligibility/attorney-fees issues — not the '462 patent. The companion fee-shifting decision is Realtime Adaptive Streaming L.L.C. v. Sling TV, L.L.C., No. 2023-1035 (Fed. Cir. Aug. 23, 2024) (vacating and remanding the $3.9M § 285 fee award). So the CAFC docket link on the '462 family page does not by itself establish an appeal of the '462.
• On 2026 CAFC dockets: I did not find any 2026 Federal Circuit docket or filing specifically involving U.S. 8,634,462. My final targeted search for 2025 activity on "8,634,462"/"8634462" was not completed before I ran out of tool steps, so I cannot state with certainty that no such 2025–2026 activity exists. Treat this as "not found," not "confirmed absent." A PACER/Docket Navigator or USPTO Patent Center check would be needed to confirm current status.

6. Explicit uncertainty list

1. Full granted claim set of US 8,634,462 — verified only claim 1 (and dependents 2–7) verbatim; independent coder/decoder/data-signal claims and their exact numbering are not confirmed.
2. Underlying PCT application number — not in the retrieved record.
3. Current 2025–2026 litigation/docket activity — no 2026 CAFC docket located for the '462; verification incomplete.
4. Legal status (Ceased / expiration 2029-05-22) is Google Patents' stated assumption, expressly disclaimed by Google as not a legal conclusion; treat as indicative only.
5. Whether the '462 was actually asserted (versus merely listed in the MDL/portfolio) in each of the enumerated district-court cases — the complaint-level mapping I found was in the D. Del. Netflix matter; other entries are docket listings.

I'll research litigation involving US Patent 8634462. Let me search multiple sources.

Let me dig deeper into the specific cases and verify which ones asserted the '462 patent.

Litigation Involving U.S. Patent No. 8,634,462

Patent identification (for reference)

Note: the '462 patent is a Narroschke/Musmann patent and is distinct from the Fallon family ('046, '442, '535, '610, '907, '477) that dominates most Realtime Adaptive Streaming litigation. It was asserted in the same multi-defendant campaign, but often as one of the "Non-Fallon" patents.

Known litigation asserting the '462 patent

The '462 patent was part of Realtime Adaptive Streaming LLC's 2017–2018 enforcement campaign. It was asserted in the following actions (all Realtime as plaintiff):

Related multi-district proceeding: In In re Realtime Adaptive Streaming LLC Patent Litig., 326 F. Supp. 3d 1374 (J.P.M.L. Aug. 1, 2018), the JPML denied Realtime's §1407 motion to centralize twelve actions in five districts. The Panel expressly listed U.S. Patent No. 8,634,462 among the eight patents at issue (MDL No. 2845). Centralization was denied. This is significant because the '462 was held over for defendant-by-defendant adjudication rather than consolidated.

Important caveats

1. No litigation outcome on the '462 patent's merits was found. The high-profile decisions in this campaign — e.g., Realtime Adaptive Streaming LLC v. Sling TV L.L.C., No. 21-2268 (Fed. Cir. May 11, 2023) and Adaptive Streaming Inc. v. Netflix, Inc., 836 F. App'x 900 (Fed. Cir. 2020) — concern the '610 and '535 patents (Fallon family), not the '462. I found no Federal Circuit or district-court merits ruling specifically invalidating or upholding the '462 claims.

2. The Court of Appeals case linked on Google Patents (CAFC 21-2268) is the Sling TV/DISH appeal concerning the '610 patent, not the '462. It appears on the '462 page only because of family-level grouping. I would not report it as "'462 appellate litigation."

3. Two possible identification issues to flag rather than resolve:

   • The Google Patents "family" record shows a reissue, US RE46,777 E1 (from application 15/013,843, filed Feb. 2, 2016). In the C.D. Cal. Google/Adobe claim-construction order, the five patents asserted against both Google and Adobe were listed as '046, '535, '298, '477, and RE46,777 — suggesting the '462 patent may be related to (or replaced by) RE46,777 in those pleadings. I cannot confirm from the sources retrieved whether RE46,777 is a reissue of the '462 or of a sibling patent; treat this as uncertain.
   • Filing dates for Brightcove, Apple, Cisco, and Sony are approximate because the complaints themselves were not directly retrieved; the Nov. 21, 2017 date for Sony is confirmed from Sony's motion-to-dismiss brief.

4. Verification limits. I could not confirm current docket status (settled vs. voluntarily dismissed) for several cases (Wowza, Samsung, Hulu, Apple, Cisco, Google/YouTube, Adobe) beyond the events shown above. Following your strict rule, I have not auto-corrected or assumed any case numbers; all numbers are reported as found.

Bottom line

Yes, litigation involving US 8,634,462 is well documented — but as part of a mass multi-defendant campaign rather than any single headline case. The plaintiff in every action was Realtime Adaptive Streaming LLC. The '462 was asserted against at least 11 defendant groups (Netflix, Haivision, Wowza, Samsung, Google/YouTube, Hulu, Brightcove, Sony, Apple, Cisco, Adobe) across the District of Delaware, District of Colorado, Central District of California, Eastern District of Texas, and District of Massachusetts, all filed 2017–2018. The only clear disposition I could confirm for the '462 itself is the Sony case, dismissed with prejudice on November 5, 2018. The omnibus MDL request (No. 2845) that named the '462 was denied.

If you need docket-level confirmation of settlements and dismissals, the primary sources to pull are: (a) the district dockets via PACER for each case number above, (b) the JPML order (MDL No. 2845), and (c) the Unified Patents litigation portal entries corresponding to these case numbers — the Google Patents page lists the Delaware, Colorado, Texas, Massachusetts, and California case links individually.

I'll start by verifying the structured PTAB data against what web sources show for this patent.

Let me dig deeper on possible PTAB filings involving this patent and its owner.

The portfolio IPRs I'm finding target sibling patents ('046, '442, '535, '907, '477, '298). Let me specifically test whether any IPR names the '462 patent itself.

PTAB / AIA Trial Analysis — US 8,634,462 ("Quantization for hybrid video coding")

Proceedings overview

Total AIA trial proceedings on US 8,634,462: zero (0). The USPTO Open Data Portal returns no IPR, PGR, or CBM petition naming this patent, and my independent web checks for PTAB filings quoting the "'462 patent" surfaced none — the only case numbers that appear in searches are district-court dockets, not PTAB trials. Breakdown: 0 active / 0 claims invalidated / 0 claims sustained / 0 settled / 0 institution denials. The bottom line for a defendant: this is not a "hardened patent" story and it is not a "claims canceled" story — it is an untested-at-PTAB patent. All claims of the '462 remain presumptively valid and stand exactly as issued, because no AIA trial has ever been instituted against it. That cuts both ways: the patent owner has no adverse PTAB record to explain away, but you also have no free kill-shot from a prior FWD — any invalidity attack must be built from scratch, and served within your § 315(b) window.

Note on process: the canonical structured block states the ODP API shows no AIA trial proceedings, and I must not invent proceeding numbers. I found none, so there are no per-proceeding entries to report below. Instead I've documented what does exist around this patent (the litigation campaign and the invalidated sibling patents), because that is the actionable intelligence.

Important distinction — the Realtime Adaptive Streaming IPRs did NOT target the '462

Realtime Adaptive Streaming LLC waged a large multi-defendant campaign (2017–2019) and its portfolio drew a substantial wave of IPRs — but those petitions were aimed at other patents in the family, not the '462. Confirmed targets among siblings include:

• US 9,762,907 — Netflix, Inc. v. Realtime Adaptive Streaming LLC, IPR2018-01817; Final Written Decision (2020-03-30) held all challenged claims 1–14 unpatentable. Panel: Anderson, Baer, Jivani. FWD
• US 9,769,477 — Netflix, Inc. v. Realtime Adaptive Streaming LLC, IPR2018-01187 (instituted 2019-02-04 on claims 1–6, 9–14, 20–22, 25–27; panel Braden, Cherry, Jivani); plus IPR2018-01413 (Sony, terminated pre-institution) and IPR2018-01630 (Netflix). Institution decision
• US 7,386,046 — Netflix IPR2018-01169 / IPR2018-01170; Unified Patents, Inc. v. Realtime Adaptive Streaming, LLC, IPR2018-00883 (judgment 2019-01-31).
• US 8,929,442 — Adobe, Inc. v. Realtime Adaptive Streaming, LLC, IPR2019-00712; also IPR2019-01586/-01587.
• US 8,934,535 and US 9,578,298 — challenged by defendants in the same campaign (Samsung et al. suits).

None of these estops or otherwise affects the '462 — IPR estoppel under § 315(e)(2) runs only to the patent for which the IPR was instituted. A defendant facing the '462 cannot borrow an FWD against a sibling.

Litigation exposure (context, not PTAB)

The '462 was asserted widely by Realtime Adaptive Streaming LLC, including in Realtime Adaptive Streaming LLC v. Netflix, Inc., C.A. No. 17-1692-CFC-SRF (D. Del.) — where the '462 was treated as one of the "Non-Fallon patents" alongside the '298 — and against Haivision (D. Del. 17-1520), Brightcove (17-1519), Sony (17-1693), Adobe (D. Mass. 18-10355), Google (C.D. Cal. 18-3629), Amazon, Cisco (E.D. Tex. 17-591), Samsung (E.D. Tex. 18-113), Hulu (C.D. Cal. 17-7611), Polycom (D. Colo.), Comcast (D. Colo.), and Sling TV (D. Colo.). Realtime's MDL consolidation bid was denied (2018-08-01), and several cases were stayed pending IPR — but on the sibling patents. There is also a Federal Circuit appeal, No. 21-2268, associated with the family; I could not confirm from available sources that it arises from a PTAB FWD (consistent with there being no '462 FWD to appeal), so treat it as an appeal of unknown subject matter rather than a PTAB disposition. A related reissue (US RE46,777 E1) appears in the family chain — verify separately whether it reissues the '462 before relying on it.

Strategic summary

Claim status of the '462: every claim is UNTESTED at the PTAB. No independent claim has been canceled, no dependent claim has been held unpatentable, and none has been affirmed as patentable. The patent stands as granted with its full original claim set. Because there is no FWD, there is no claim-level PTAB disposition to cite and none to fear — you cannot point a court or a plaintiff to a canceled claim, and the plaintiff cannot point to a PTAB affirmance either.

Estoppel landscape is clean for both sides. No § 315(e)(2) estoppel attaches to the '462 from any AIA trial, because no trial was ever instituted on it. Practically, that means (a) a defendant is not blocked from raising any prior-art ground against the '462, and (b) the patent owner is not shielded by any Board finding. The real gating constraint is § 315(b) — a petition must be filed within one year of service of a complaint alleging infringement of this patent. The 2017–2019 assertion wave means many would-be petitioners' one-year windows have almost certainly lapsed; if you are newly served (or newly named in a threat), the clock is running and a PGR is unavailable (patent has pre-AIA priority of 2007-03-13, so only IPR is available, not PGR/CBM — and CBM is gone after SAS/Unwired Planet sunset anyway). Watch the reissue RE46,777 angle: reissue claims can carry a fresh vulnerability window and different § 315(b) timing if the reissue has been asserted.

Pattern signals. The same petitioners (Netflix, Sony, Adobe, Google) layered multiple IPRs across the sibling patents, and Unified Patents appears in the chain via IPR2018-00883 — so defensive-aggregator coverage of this portfolio existed, but it stopped short of the '462. The absence of any IPR on the '462, despite heavy assertion, suggests either the one-year window was missed across the board, the asserted '462 claims were narrower or more claim-construction-dependent than the siblings, or the cases resolved/stayed before defendants pulled the trigger. I cannot confirm which; treat it as an open question worth diligence, not a conclusion.

Recommended next steps

1. Correct any assumption that the portfolio IPRs cover you here. If a demand letter or complaint asserts the '462, none of the Netflix/Sony/Adobe/Google/Unified FWDs are yours to cite; the brother patents' invalidation does not invalidate this one. Build an independent invalidity position.
2. Calendar the § 315(b) deadline immediately. If you were served with a '462 complaint, your IPR window is one year from that service date. If you're outside it, your invalidity case lives in the district court / ITC, not the PTAB.
3. Verify the reissue. Confirm whether US RE46,777 E1 is a reissue of the '462 and, if so, whether it has been asserted — reissue claims change both the claim set and your challenge strategy.
4. Search PTAB E2E and PTAB Final Written Decisions directly before relying on the "zero proceedings" result, and re-check ODP as new filings are indexed; if any '462 petition appears, note the institution decision deadline and the statutory one-year trial clock from institution.
5. Note the patent's status: per Google Patents (fetched 2026-09-28), the '462 is Ceased, with an adjusted expiration of 2029-05-22 — relevant to damages-tail and FRAND/monitoring analysis, though expiration does not affect validity challenges to past infringement.
6. If no '462 assertion is live against you, say so plainly on the record: the absence of PTAB activity is itself a signal, but here it reflects an untested patent rather than a vindicated one — do not over-read it as weakness.

I'll research the assignment chain and ownership history for US 8,634,462, attempting to pull actual USPTO assignment records (reel/frame, correspondent) and cross-referencing litigation and NPE directories.

Let me dig into the specific assignment records and the corporate structure behind Realtime Adaptive Streaming.

Ownership & Assignment Analysis — US 8,634,462 ("Quantization for hybrid video coding")

Source limitation (read first). I was unable to open the USPTO Patent Assignment Search results page for US 8,634,462 before my tool budget ran out, and I could not verify reel/frame numbers or the recorded correspondent for any link in this chain. Everything below is reconstructed from (a) the Google Patents legal-events record supplied in the patent text, (b) IPR filings and district-court records that quote the assignment history, and (c) the Realtime family's litigation corpus. The reel/frame fields and correspondent fields are marked "not retrieved" rather than guessed. To complete this properly, pull the record at the USPTO Assignment Center and search patent number 8634462: https://assignmentcenter.uspto.gov/ (mirror: https://assignment.uspto.gov/patent/index.html).

Inventors

Both inventors appear on the priority application US 11/685,566 (published US 2008/0225947 A1) and on the '462 itself. The issued patent names "Individual" as the original assignee, which is the key fact for this chain: the patent was held personally by the two inventors, not by an operating company.

Unusual pattern noted: there is no corporate original assignee. A patent that issues to two individual natural persons, is left unassigned for ~4 years post-issuance (issue 2014-01-21), and is then swept up on a single day in 2018 into a two-step licensing chain is the classic profile of a personal-hold / academic portfolio that was sold to an aggregator, not of an operating company monetizing its own R&D. That said, I could not confirm from the record whether either inventor was employed by an entity that took a shop right, so I do not assert an employer at filing.

Original assignee

• Entity on the issued patent: "Individual" — i.e., Musmann and Narroschke personally.
• Product embodying the claims: None. The inventors are research scientists; there is no record of a product company behind the '462.
• Primary line of business at filing: academic/industrial video-coding research (H.264/AVC standardization context — the specification expressly targets the H.264/AVC reference-software "second quantization" step and cites JVT doc. JVT-K049).
• Current status: the individual inventorship was extinguished by the 2018-11-16 assignments (see below). The inventors themselves are not shown in the record as holding any residual interest.

This is significant: the "original assignee" was not an operating company at all, so the classic "operating company sells to a shell" framing is only half the story — here the property started outside the corporate world and was routed through a professional licensing/licensing-aggregation layer before reaching an asserter.

Assignment timeline

There are recorded post-issuance assignments (the Google legal-events feed shows two, both dated 2018-11-16), so I do not stop here — but I must be candid that I did not retrieve the Reel/Frame or the correspondent of record.

• 2018-11-16 (executed) / recorded date not retrieved — Reel not retrieved / not retrieved

  • Conveyance: Assignment
  • Assignor: Musmann, Hans-Georg and Narroschke, Matthias (the two natural-person inventors)
  • Assignee: S.I.SV.EL. Società Italiana per lo Sviluppo dell'Elettronica S.p.A. (Turin/None, Italy; the Sisvel group)
  • Correspondent: not retrieved (the USPTO record would name the recording attorney/firm here)
  • Context: transfer-to-monetization intermediary — the inventors sold their personal patent rights to an Italian patent-licensing house (Sisvel), the first step of a two-hop 2018 chain.

• 2018-11-16 (executed) / recorded date not retrieved — Reel not retrieved / not retrieved

  • Conveyance: Assignment
  • Assignor: S.I.SV.EL. Società Italiana per lo Sviluppo dell'Elettronica S.p.A.
  • Assignee: Realtime Adaptive Streaming LLC
  • Correspondent: not retrieved
  • Context: transfer-to-asserter — the same-day "flip" out of the Sisvel licensing vehicle into the Realtime assertion entity that was already litigating this patent.

Important timing wrinkle / possible contradiction to flag. The Google legal-events feed dates both assignments 2018-11-16, yet the '462 was asserted ~a year earlier — Realtime's D. Del. 1:17-cv-01692 complaint against Netflix (originally filed 2017-11-21) pleads that Realtime "is the owner by assignment" and asserts the '462 as a "Non-Fallon patent." If the assignment truly executed on 2018-11-16, Realtime was asserting a patent it did not yet own in 2017. The likely explanations are that (i) the 2018-11-16 date is the recordation date and execution was earlier, or (ii) there is an earlier, separate assignment not surfaced in the Google feed. Either way, the executed-date vs. recorded-date distinction on these two entries must be checked against the Assignment Center, because it changes the "pre-litigation transfer" analysis below.

No other assignments (no security agreements, no mergers, no name changes, no releases) appear in the Google legal-events record for the '462. Note also the related reissue US RE46,777 E1 (from application 15/013,843), which is part of the same chain and should be checked separately.

Timeline diagram

timeline
    title Ownership of US 8634462
    2007 : Priority application filed by inventors
    2008 : Application filed at USPTO
    2014 : Patent issued to individual inventors
    2017 : First infringement suits asserting patent
    2018 : Inventors assign to Sisvel SpA
         : Same day transfer to Realtime Adaptive Streaming LLC

NPE / troll-pattern signals

1. Shell-entity transfer — PRESENT. The patent moved out of individual ownership (2018-11-16) first to S.I.SV.EL./Sisvel (a pure patent-licensing house) and then, the same day, to Realtime Adaptive Streaming LLC, a licensing-only litigant. Realtime Adaptive Streaming is described in its own IPR filings as "a Texas limited liability company," with a principal place of business given as 1828 E.S.E. Loop 323, Tyler, Texas 75701 (IPR2018-01187 Power of Attorney, signed 6/21/18) and later 66 Palmer Avenue, Suite 27, Bronxville, NY 10708 (Updated Power of Attorney, IPR2019-01035, Feb. 3, 2020). A single-purpose litigating LLC with no products is exactly the profile. The suffix is "LLC" rather than "IP Holdings," but the absence of any commercial product is established elsewhere in the record.

2. Known asserter in the chain — PRESENT (strong). Realtime Adaptive Streaming LLC is a well-documented high-volume plaintiff, associated with the Realtime Data LLC d/b/a IXO family. It appears as assignee on the Unified Patents portal (e.g., https://portal.unifiedpatents.com/patents/patent/US-[6175856](/patent/6175856)-B1) as owner of a large video/compression portfolio (US 8,867,610; 8,929,442; 8,934,535; 9,762,907; 9,769,477; 9,578,298; 10,212,417; etc.). It was petitioner/patent-owner in numerous PTAB proceedings (IPR2018-01187, -01299, -01331, -01332, -01342; IPR2019-00712, -01033, -01035, -01036) and was the subject of the consolidated **MDL In re Realtime Adaptive Streaming LLC Patent Litig., 326 F. Supp. 3d 1374 (J.P.M.L. 2018)**. The Sisvel / S.I.SV.EL. layer is likewise a large professional patent-licensing and pool operator (not an operating company). Both links qualify as professional monetizers.

3. Repeat correspondent across the chain — UNCLEAR / not retrievable. I could not obtain the recorded correspondent for either 2018 assignment, so I cannot test the "same attorney filed both links" hypothesis. Do not collapse this into the litigation counsel. The patent owner's PTAB counsel is Russ, August & Kabat (Philip X. Wang, Kent Shum, C. Jay Chung, Neil A. Rubin, Reza Mirzaie — e.g., IPR2019-01035 Updated POA), and the recorded assignment correspondent may be an entirely different firm. A recurrence finding requires the reel/frame correspondent fields, which are outstanding.

4. Cascading transfers — PRESENT. Two consecutive assignments dated the same day (2018-11-16) carried the patent from the inventors → Sisvel → Realtime. A same-day, back-to-back pass-through is the textbook pattern for importing an asset into an assertion vehicle while leaving the intermediate licensor as a conduit.

5. Pre-litigation transfer — UNRESOLVED / possibly inverted. If the execution date is genuinely 2018-11-16, the transfer post-dates the earliest suits (Realtime v. Netflix filed 2017-11-21 asserting the '462), which would be anomalous and would suggest either an earlier unrecorded assignment or that 2018-11-16 is a recordation date. Because I could not confirm execution vs. recording, I mark this unclear rather than present or absent. Either way, the 2018 transfer was followed by a documented wave of 2018 suits (e.g., D. Colo. 1:18-cv-00927 and others), consistent with coordinated assertion.

6. Bankruptcy fire-sale — NOT PRESENT. No evidence any assignor entered Chapter 7/11, and no sale-in-bankruptcy record appears.

7. Privateering — NOT PRESENT (no operating-company trigger). There is no operating-company plaintiff here; the inventors were individuals and the first corporate owner in the chain (Sisvel) is itself a licensing company. No evidence of an operating company funding assertions behind a proxy.

8. Defensive aggregator — NOT PRESENT. The chain terminates at an asserter, not at RPX, AST, LOT, Unified, or OIN. The patent was weaponized, not neutralized.

Verdict

NPE — high confidence.

Two strong signals are established on the record: (i) a same-day, two-hop 2018-11-16 transfer (inventors → S.I.SV.EL./Sisvel → Realtime Adaptive Streaming LLC) that routed a personally-held, product-less patent into a known high-volume litigating LLC, and (ii) the terminal assignee is the documented Realtime assertion family (Texas LLC; Tyler TX then Bronxville NY addresses per its own IPR Powers of Attorney; owner in MDL 326 F. Supp. 3d 1374 and numerous PTAB proceedings), with no product and a track record of suing H.264/H.265 implementers (e.g., the '462 mapping in Realtime Adaptive Streaming LLC v. Netflix, Inc., D. Del. 1:17-cv-01692). The one signal I could not close — the recorded correspondent recurrence test — is pending retrieval of the reel/frame data, so it is scored unclear, not absent.

Verify against the primary source: USPTO Assignment Center, search by patent number 8634462 — https://assignmentcenter.uspto.gov/ (mirror https://assignment.uspto.gov/patent/index.html). Pull the Reel/Frame, the execution date vs. recording date, and the correspondent of record for both 2018-11-16 entries (and for the RE46,777 chain) to complete signals #3 and #5.

I'll research US 8,634,462 and its cited prior art. Let me start with several searches.

I found the patent's front-page "References Cited" list. Let me verify details on each citation and look for the examiner-cited vs. applicant-cited distinction.

I hit my search-step limit mid-verification, so I'm reporting what I retrieved and flagging what I could not confirm. Here is the prior-art analysis for US 8,634,462 B2.

Prior-Art Analysis — US 8,634,462 B2 ("Quantization for hybrid video coding")

0. Target identification (literal)

• US 8,634,462 B2 — "Quantization for hybrid video coding"
• App. 12/531,025, filed 2008-03-10 (national stage of PCT/EP2008/052824, per the ISR/Written Opinion listed in the record); priority claimed from US 11/685,566 filed 2007-03-13; granted 2014-01-21.
• Inventors: Matthias Narroschke, Hans-Georg Musmann. Current assignee: Realtime Adaptive Streaming LLC.

I did not return results for similar numbers (e.g., 8,634,463 or 8,644,462); everything below keys on 8634462 / 12/531,025.

1. Source of the citation list

The authoritative full text is the Google Patents record: https://patents.google.com/patent/US8634462/en.

The front-page "References Cited" list was retrieved verbatim from the patent's own front page as reproduced in the litigation exhibit Realtime Adaptive Streaming LLC v. Sony Electronics Inc., D. Del. 1:17-cv-01693, Doc. 1-2 (https://www.docketalarm.com/cases/Delaware_District_Court/1--17-cv-01693/Realtime_Adaptive_Streaming_LLC_v._Sony_Electronics_Inc/docs/1/2.pdf). It reads, as printed:

U.S. PATENT DOCUMENTS

FOREIGN PATENT DOCUMENTS

OTHER PUBLICATIONS

• Lim et al., "Text Description of Joint Model Reference Encoding Methods and Decoding Concealment Methods," Study of ISO/IEC 14496-10 and ISO/IEC 14496-5/AMD6, Mar. 2004, pp. 2-45.
• Narroschke, "Adaptive coding of the prediction error for H.264/AVC," Institut für Informationsverarbeitung, Universität Hannover, Dec. 2, 2005, 15 pages.
• Narroschke et al., "Adaptive prediction error coding in spatial and frequency domain for H.264/AVC," ITU-T SG16/Q6 VCEG, 29th Meeting, Bangkok, Jan. 16-20, 2006, 14 pages.
• Naroschke, "Adaptive prediction error coding in the spatial and frequency domain in the KTA reference model," ISO/IEC, Montreux, CH, Apr. 2006, 16 pages.
• Naroschke, "Extending the prediction error coder of H.264/AVC by a vector quantizer," Proc. SPIE 5960, Jul. 2005, 12 pages.
• Ostermann et al., "Video coding with H.264/AVC: Tools, Performance, and Complexity," IEEE Circuits and Systems Magazine 4(1): 7-28, 2004.
• International Search Report, mailed Aug. 13, 2008, for PCT/EP2008/052824, 4 pages.
• Written Opinion, mailed Aug. 13, 2008, for PCT/EP2008/052824, 8 pages.

⚠️ Two caveats. (1) The asterisk (*) appears on only 7,929,776 B2 in the exhibit extract, marked "cited by examiner" — I could not cleanly reconstruct which of the other documents were examiner-cited vs. applicant IDS-cited, because the exhibit's OCR/text is partial. (2) I could not re-verify the complete list against the raw USPTO front page before exhausting my search steps; treat the above as "as printed on the reproduced front page," not as an independently re-pulled USPTO record.

Do not confuse two different citation directions. Google Patents also lists documents that cite the '462 (e.g., RU2653236C2, JP4963526B2, US10021424, KR100964641B1). Those are later documents — they are not prior art to the '462 and are irrelevant to § 102. Only the "References Cited" (backward) list above matters here.

2. Legal framework (why the dates matter)

The '462 has an effective filing date of 2008-03-10 and is pre-AIA (filed before 2013-03-16). Because the priority claim is an external/§ 119-style priority to US 11/685,566 (2007-03-13) rather than a § 120 continuation, the one-year § 102(b) statutory-bar window runs from 2008-03-10, i.e., anything published/patented before ~2007-03-10 is a § 102(b) bar. Section 119 priority can antedate § 102(a)/(e) references but (per In re Hilmer) cannot remove a § 102(b) statutory bar. Consequently:

• References patented/published before 2007-03-10 → potentially § 102(a) and § 102(b).
• References published between 2007-03-10 and 2008-03-10 (e.g., WO 2007/079964 A1, US 2007/0133891 A1) → § 102(a) (if before the invention), or § 102(e) only if they have an earlier effective U.S. filing.
• U.S. patents/publications granted after filing (7,929,776 B2; 2010/0220784 A1) → § 102(e) only, and only if their effective U.S. filing predates the invention.

Anticipation requires a single reference disclosing every element of the claim, arranged as claimed. The only claim I have verified verbatim is claim 1 (and dependents 2–7) from the earlier section; mapping below is against that grant-version claim 1 unless stated.

3. Reference-by-reference analysis

⭐ 1) US 5,691,770 A — Keesman, Coene, Salomons — the closest prior art

• Full citation / dates: US 5,691,770 A, "Device and method for coding video pictures," G. J. Keesman, W. M. J. Coene, E. W. Salomons, assignee U.S. Philips Corp.; NL priority 1995-04-25, U.S. filed 1996-04-24, issued 1997-11-25.
• Description: An MPEG-type hybrid video coder: motion-compensated prediction residue → block DCT → quantization, with coefficients modified so the Lagrangian cost D + λR is minimized, where λ is computed from the coefficient amplitude distribution. Critically (FIG. 10), it compares the Lagrangian cost L1 of keeping a quantized coefficient against L2 of forcing it to zero, and further states: "An even further improvement is obtained by comparing the Lagrangian cost of a picture block with that of a picture block all coefficients of which have the value of 0… the number of bits involved in encoding coded_block_pattern is taken into account in the λR term." It also confirms hybrid structure: "The applied picture is then first subtracted from a motion-compensated prediction picture, whereafter the residue thus obtained is coded."
• § 102 mapping: This is the single most damaging reference. It arguably discloses claim 1 in full: reduce temporal redundancy by motion-compensated prediction → quantize (coefficients) to obtain quantized values → compute a "quantization efficiency" (keep-cost L1 / C_j) → compute a "zero efficiency" (all-zero-block cost L2) → compare → select. It also squarely meets dependents claim 3 (efficiency based on a cost function), claim 4 (rate-distortion cost), claim 5 (distortion + weighted rate), and claim 6 (C = D + λR form). It is a § 102(b) reference (issued 1997, far more than one year before filing). Claim 2 (subblock-level pass then macroblock-level pass) is the weakest mapping, since Keesman's explicit comparison operates at the coefficient/block level with a picture-level coded_block_pattern term.

2) Lim et al., JVT-K049 (2004)

• Citation / date: K.-P. Lim, G. Sullivan, T. Wiegand, "Text Description of Joint Model Reference Encoding Methods and Decoding Concealment Methods," Joint Video Team (JVT) doc. JVT-K049, March 2004 (also cited as "Kunststoff, Germany" in the patent text and "Munich" in the description — the record is internally inconsistent on the city).
• Description: The H.264/AVC reference-software encoding description. This is the document the '462's own Background identifies as the source of the "second quantization step" that zeroes blocks using the "importance" sum I_j with thresholds 5 (8×8 block) and 6 (macroblock) — i.e., the exact heuristic the patent calls "unsatisfactory."
• § 102 mapping: A § 102(b) reference. It discloses the hybrid coding, subblock/macroblock structure, and the zeroing of blocks, but it zeroes via the importance threshold, not by computing and comparing two efficiencies. It therefore cannot anticipate claim 1 (its "second quantization" is the admitted prior art, not the claimed efficiency comparison). Its best role is background/§ 103 motivation, not § 102 anticipation. It is, however, strong intrinsic evidence of what a skilled artisan already knew and of the "hybrid coding + subblocks" preamble elements.

3) The four Narroschke papers (inventor's own publications, 2005–2006)

• Citations / dates: (a) "Adaptive coding of the prediction error for H.264/AVC," Univ. Hannover, Dec. 2, 2005; (b) Narroschke et al., "Adaptive prediction error coding in spatial and frequency domain for H.264/AVC," VCEG, Bangkok, Jan. 16-20, 2006; (c) "…in the KTA reference model," ISO/IEC MPEG (Montreux), Apr. 2006; (d) "Extending the prediction error coder of H.264/AVC by a vector quantizer," Proc. SPIE 5960, Jul. 2005.
• Description: The inventor's own earlier work on adaptively switching prediction-error coding between the spatial and frequency domains and on vector-quantizer extension of the H.264/AVC prediction-error coder — i.e., the subject matter of the related application (US 11/685,566 / US 2008/0225947 A1; EP 1,982,524 "Adaptive coding of a prediction error in hybrid video coding"), not the efficiency-comparison quantization.
• § 102 mapping: These are § 102(b) statutory-bar candidates against the later-filed '462 claims (all predate 2007-03-10), and because they are the inventor's own disclosures within the same project they are the most factually "on point" as to the spatial/frequency switching and scan aspects. They would only anticipate a claim if they describe the efficiency-vs-zero-efficiency comparison of claim 1 — which the titles/framing suggest they do not; they appear to address the domain switch and prediction-error coding instead. Treat as high-relevance § 103 art / possible § 102 for any spatial-vs-frequency-domain claim, but not established as anticipating granted claim 1. ⚠️ I could not open the full texts before running out of steps, so this is provisional.

4) Ostermann et al. (2004)

• Citation / date: "Video coding with H.264/AVC: Tools, Performance, and Complexity," IEEE Circuits and Systems Magazine 4(1): 7-28, 2004.
• Description: General overview of H.264/AVC tools (transform, quantization, CABAC/CAVLC).
• § 102 mapping: § 102(b) background art. Discloses the general hybrid-coding framework (claims 1's preamble) but not the efficiency comparison. Not anticipating.

5) US 7,203,374 B2 — Haaba (4/2007; 382/248)

• Citation / dates: US 7,203,374 B2, inventor "Haaba," granted 2007-04-10 (per front page month/year), class 382/248. Effective filing date not verified.
• § 102 mapping: Because it issued after the ~2007-03-10 statutory-bar line, it can only be § 102(e) art, and only if its effective U.S. filing predates the invention. Its class (382/248, image compression) is consistent with quantization/transform art. ⚠️ I could not retrieve its content, so I cannot map it to any claim — unverified.

6) US 7,929,776 B2 — Sethi et al. (4/2011; 382/232) — examiner-cited (*)

• Citation / dates: US 7,929,776 B2, Sethi et al., granted 2011-04-19, class 382/232; marked "* cited by examiner."
• § 102 mapping: Granted after filing → § 102(e) only, if its effective U.S. filing predates the invention; likely the same family as the US 2006/0233239 A1 (Sethi et al.) publication below. ⚠️ Content not verified; no claim mapping asserted.

7) US 2006/0233239 A1 — Sethi et al. (10/2006)

• Citation / date: Published 2006-10-19. Pre-dates the 2007-03-10 line → potential § 102(a) and (as a U.S. publication) § 102(e) art. ⚠️ Content not verified. Likely related to 7,929,776 B2.

8) US 2006/0098733 A1 — Matsumura et al. (5/2006)

• Citation / date: Published 2006-05-11. Potential § 102(a)/(b)/(e) art (published more than one year before the 2008-03-10 filing). ⚠️ Content not verified.

9) US 2002/0114392 A1 — Sekiguchi et al. (8/2002; 375/240.15)

• Citation / date: Published 2002-08-22, class 375/240.15 (video coding). Clean § 102(b) reference. ⚠️ Content not verified.

10) US 2007/0133891 A1 — Jeong (6/2007)

• Citation / date: Published 2007-06-14 — after the 2007-03-13 priority and before the 2008-03-10 filing. Available as prior art only under § 102(a) (if before the invention) or § 102(e) (if it has an earlier effective U.S. filing). ⚠️ Content not verified.

11) US 2010/0220784 A1 — Tanimoto et al. (9/2010)

• Citation / date: Published 2010-09-02 — after filing. § 102(e) only, and only with an earlier effective filing date. ⚠️ Content not verified.

12) WO 96/34495 (10/1996)

• Citation / date: Published 1996-10-31. Clean § 102(b) reference (printed publication in a PCT application). ⚠️ Content not verified.

13) WO 2007/079964 A1 (7/2007)

• Citation / date: Published 2007-07-12 — after the 3/13/2007 priority, before the 3/10/2008 filing. Available only under § 102(a) (invention-date dependent) or via an earlier effective filing. ⚠️ Content not verified.

4. Ranking / bottom line

1. US 5,691,770 (Keesman) — by a wide margin the most relevant piece of prior art. It is a § 102(b) reference that expressly teaches comparing the Lagrangian cost of retaining a block's (quantized) coefficients against the cost of an all-zero block, with the coded-block-pattern bits folded into the rate term — i.e., it reads directly on claim 1's "quantization efficiency vs. zero efficiency … select … maintain or set to zero" and on dependents 3–6. This is the reference to lead any invalidity theory with.
2. Lim et al., JVT-K049 (2004) — the admitted-prior-art "second quantization" threshold scheme; strong § 103/background art but not an anticipatory § 102 reference for claim 1.
3. The four Narroschke papers (2005–2006) — inventor's own § 102(b) disclosures; relevant to the spatial/frequency-switch and scan aspects, not verified as anticipating the granted quantum-efficiency claim.
4. Ostermann et al. (2004) and the remaining U.S./foreign patent documents — secondary/background; several (7,203,374; 7,929,776; 2010/0220784; 2007/0133891; WO 2007/079964) are post-priority and reachable only via § 102(e) or § 102(a).

5. Explicit uncertainty list (not confirmed)

1. I could not retrieve the contents of most of the cited patent documents (7,203,374; 7,929,776; 2002/0114392; 2006/0098733; 2006/0233239; 2007/0133891; 2010/0220784; WO 96/34495; WO 2007/079964) — I hit the tool-step limit after confirming the citation list and fully retrieving only US 5,691,770 (Keesman). No § 102 claim mapping is asserted for those; they are marked unverified.
2. Examiner-cited vs. IDS-cited — only 7,929,776 B2 carries the "*" mark in the reproduced front page; the split for the rest is unresolved.
3. The complete granted claim set — only claim 1 and dependents 2–7 were verified earlier; independent coder/decoder/data-signal claims are not confirmed, so § 102 mappings are stated against claim 1 unless noted.
4. The underlying PCT number — the record shows PCT/EP2008/052824 via the cited ISR, but I did not independently confirm the US 12/531,025↔PCT correspondence.
5. Effective filing dates of the post-priority references — needed to decide § 102(e) reach; not verified.
6. Whether the four Narroschke papers actually disclose the efficiency-comparison of claim 1 — titles suggest not, but full texts were not reviewed.

If you want, I can (in a follow-up pass) pull the full text of US 5,691,770 claim-by-claim against the '462's granted claims, and retrieve the content of the remaining cited U.S. documents to complete the § 102(e) date analysis.

I'll research the prior-art references associated with US 8,634,462 before constructing the § 103 analysis.

I now have the patent's front-page "References Cited." Let me pull content for a few of the key cited references to ground the § 103 combinations.

Obviousness Analysis — U.S. Patent 8,634,462 (35 U.S.C. § 103)

Analyst note on sources. This builds on the previously generated summary (bibliographic data, claim 1 overview, litigation). Two updates/corrections to that earlier work, then the analysis:

• Resolved uncertainty #1 (partial): I retrieved the granted claim 1 and claim 11 verbatim (via Patexia's full-text of the patent, https://patexia.com/us/patent/08634462, and the Google Patents record). Claim 11 is a coder claim (means-plus-function). This upgrades the earlier "could not retrieve" flag.
• Resolved uncertainty #2: The PCT number is PCT/EP2008/052824 — it appears on the patent's own face: "International Search Report, mailed Aug. 13, 2008, for PCT/EP2008/052824."
• ⚠️ Contradiction flagged: The earlier summary stated the priority application US 11/685,566 published as US 2008/0225947 A1 titled "Adaptive coding of prediction error in hybrid video coding." The search results instead show US 2008/0225947 A1 titled "Quantization for hybrid video coding," priority 2007‑03‑13 (e.g., the family tables at https://patents.google.com/patent/KR100964641B1/en and https://patents.google.com/patent/[US10021424B2](/patent/US10021424B2)/en). "Adaptive coding of a prediction error in hybrid video coding" is a separate Musmann/Narroschke family (priority 2006‑01‑09; e.g., US 9,749,660 B2, WO 2007/079964 A1). Treat the title attribution in the earlier summary as likely erroneous; the two families must be kept distinct because they play different roles as prior art below.

1. The claims at issue

Claim 1 (method). Steps, verbatim structure:

1. reducing temporal redundancy by block-based motion-compensated prediction → prediction error signal;
2. performing quantization on (a) samples of the prediction error signal or (b) coefficients from a transform to the frequency domain → "quantized values," "wherein the prediction error signal includes a plurality of subblocks each including a plurality of quantized values";
3. calculating a first quantization efficiency for the quantized values of at least one subblock;
4. setting the quantized values of that subblock to all zeroes;
5. calculating a second quantization efficiency for that subblock while all quantized values are zeroes;
6. selecting which of the first/second efficiencies is higher;
7. outputting the subblock with its quantized values if the first is higher, or zeroed if the second is higher.

Claim 11 (coder). The apparatus counterpart: means for reducing temporal redundancy + quantization means + control means performing the same five-step efficiency computation/selection.

Dependents (per the earlier summary, claims 2–7): 2 = macroblock-level aggregate comparison; 3 = efficiency based on a cost function; 4 = rate-distortion cost; 5 = distortion + weighted rate; 6 = C_j = D_j + L·R_j; 7 = summed-squared or mean-absolute quantization error.

Effective date: March 13, 2007 (priority) / March 10, 2008 (filing). Pre‑AIA § 103(a) governs; KSR Int'l v. Teleflex, 550 U.S. 398 (2007) applies.

PHOSITA: a video-coding engineer with a B.S./M.S. in EE/CS and ~2–4 years of H.264/AVC codec experience, familiar with the JM reference software, Lagrangian rate-distortion optimization, and transform/quantization.

2. The prior art of record

The "References Cited" on the face of the '462 (from the patent copy filed as an exhibit, https://www.docketalarm.com/cases/Delaware_District_Court/1--17-cv-01693/Realtime_Adaptive_Streaming_LLC_v._Sony_Electronics_Inc/docs/1/2.pdf) is the statutory art the examiner considered:

U.S. patents/applications

Foreign

• WO 96/34495 (10/1996)
• WO 2007/079964 A1 (7/2007) — "Adaptive coding of the prediction error in hybrid video coding" (Musmann/Narroschke family; priority 2006‑01‑09)

Non-patent literature

• Lim, Sullivan, Wiegand, "Text Description of Joint Model Reference Encoding Methods and Decoding Concealment Methods," JVT‑K049, March 2004 — the H.264/AVC JM reference-software description; this is the admitted prior art and the closest art.
• Narroschke, "Adaptive coding of the prediction error for H.264/AVC," Univ. Hannover, Dec. 2, 2005 (15 pp.)
• Narroschke et al., "Adaptive prediction error coding in spatial and frequency domain for H.264/AVC," ITU‑T VCEG, 29th Meeting, Bangkok, Jan. 16–20, 2006 (14 pp.)
• Narroschke, "Adaptive prediction error coding in the spatial and frequency domain in the KTA reference model," ISO/IEC, Montreux, Apr. 2006 (16 pp.)
• Narroschke, "Extending the prediction error coder of H.264/AVC by a vector quantizer," Proc. SPIE 5960, Jul. 2005 (12 pp.)
• Ostermann et al., "Video coding with H.264/AVC: Tools, Performance, and Complexity," IEEE Circuits & Systems Magazine 4(1):7–28, 2004

3. Element-by-element mapping of claim 1

Bottom line: every element of claim 1 except the metric used for the keep/zero decision is expressly in Lim/JVT‑K049. The case therefore turns on whether substituting the RD-cost metric for Lim's importance-sum heuristic is obvious. Under KSR, it is.

4. The obviousness combinations

Combination A — Lim (JVT‑K049) + H.264/AVC Lagrangian rate-distortion optimization (Wiegand; Ostermann)

What Lim teaches. The hybrid pipeline and the "second quantization step," expressly framed as a rate/distortion trade-off: "The coding of these single quantized coefficients unequal to zero may require a high data rate and may reduce the distortion only marginally." Lim then resolves that trade-off with an ad hoc proxy — the importance sum I_j vs. thresholds 5/6.

What the secondary art teaches. H.264/AVC coder control universally decides coding choices by minimizing the Lagrangian cost J = D + λR with λ = 0.85·2^((QP−12)/3) — see Wiegand et al., "Rate-Distortion Optimized Mode Selection for Very Low Bit Rate Video Coding and the Emerging H.263 Standard," IEEE TCSVT 6(2):182–190 (Apr. 1996) and Ostermann 2004. The '462 specification itself concedes this: "As a Lagrange parameter L, the commonly used Lagrange parameter for the coder control of H.264/AVC is applied."

Why a PHOSITA would combine them.

1. Same problem, same variable. Lim's decision is a rate-vs-distortion trade-off; RD cost is the standard, already-implemented way to quantify exactly that trade-off. Substituting a known, more accurate metric for an admitted heuristic in an otherwise-unchanged decision is the paradigm KSR case ("familiar elements according to known methods… predictable results").
2. The inventor's own framing supplies the motivation. The '462 Background calls Lim's procedure "unsatisfactory" and the Summary says the goal is to "avoid selecting a choice which is alleged to be a good choice even if the other choice turns out to be even better" — i.e., to stop guessing and actually evaluate both outcomes. That is the definition of replacing a proxy with the true criterion.
3. KSR "finite number of identified, predictable solutions." For a given block there are exactly two outcomes — keep the coefficients or zero them. Computing the cost of each and choosing the lesser is the most direct, predictable implementation; the expected improvement (better R‑D performance at the same complexity) is not surprising.
4. Architectural compatibility. The JM encoder already computes rate (bit costs) and distortion (SSE) for mode decision; reusing them for the zero-block decision requires no new hardware.

Result: claim 1 would have been obvious over Lim in view of the H.264 Lagrangian coder-control art.

Combination B — Lim + the inventors' own earlier Narroschke publications (Dec. 2005 / Jan. 2006 / Apr. 2006)

What the Narroschke papers teach (all pre‑2007, hence § 102(b) art). They disclose adaptive prediction-error coding in the spatial and frequency domain and, critically, that the decision is made by rate-distortion cost — including the option of coding nothing (all-zero). The '462's own description (FIG. 4 discussion) articulates this exactly: "there are at least three rate distortion costs calculated — … the spatial domain, … the frequency domain and … the case when all values are set to zero. The resulting three calculated costs are compared and the coding resulting in the lowest rate distortion costs is selected." That passage mirrors the Narroschke/KTA work.

Why a PHOSITA would combine them.

1. The Narroschke papers establish that comparing an RD cost that includes a "zero" candidate was known; Lim supplies the subblock/macroblock quantization and zeroing structure. Both address the same coding stage (prediction-error quantization), so combining is not an unpredictable field-jump (cf. KSR "obvious to try").
2. The two references are from the same technical community and same authors, and the January 2006 VCEG paper explicitly builds on the H.264/AVC residual coder — so a PHOSITA working on the JM coder would naturally read them together.
3. Motivation: extend the RD-based domain selection already described to the block-zeroing decision, reusing the existing 4×4/8×8/16×16 partition (Lim) — a simple, expected optimization.

Caveat on WO 2007/079964 A1. It is on the face of the '462, but its publication date (July 2007) post-dates the March 13, 2007 priority. As a printed publication it is therefore not § 102(a)/(b) art. It may still qualify as § 102(e)/(pre‑AIA) art as of its international filing date (2006‑01‑09) if it was published in English and designated the U.S. — I have not verified the publication language/designations, so I flag this rather than assert it. The Dec. 2005/Jan. 2006/Apr. 2006 Narroschke papers, by contrast, are unambiguously prior art and independently supply the RD-with-zero-option teaching.

Combination C — Lim + block-skip / zero-block RD-decision art (art not of record)

If a broader combination is needed, the following pre‑2007 references are on point (I flag that these were not cited on the '462's face, so they would be "new" art in any challenge):

• EP 1 761 069 A1 (Thomson, published 2007‑03‑07), "Method and apparatus for encoding video data using block skip mode" — RD-based decision to skip (zero) a block. (Located via the citation network around H.264 skip-mode art; I did not independently verify the full text.)
• US 8,130,828 B2 (Microsoft), "Adjusting quantization to preserve non-zero AC coefficients," filed 2006‑04‑07 — deciding quantizer adjustments based on the rate/distortion effect of preserving vs. eliminating non-zero coefficients.

Motivation: block-skip/zero-block modes are the H.264-native way to signal "no residual," and they are already selected by RD cost; zeroing the quantized block is the coefficient-domain equivalent. Combining with Lim is thus motivated by the same "reduce rate where distortion impact is marginal" objective Lim itself states.

Caution: Because I could only confirm these two references through secondary citation contexts (not their full specifications) within the available steps, treat Combination C as secondary/optional; Combinations A and B are the robust theory.

5. Why the dependents fall with claim 1

6. Counterarguments / why the patent nonetheless issued

For balance, the strongest non-obviousness positions a patentee could press:

1. The cited art was considered and the examiner still allowed the claims over Lim and the Narroschke papers — arguing the Office did not find the efficiency-comparison taught.
2. The two-stage architecture (per-subblock compare, then carry forward only the winning subblock's cost into a macroblock-level aggregate compare) — see the '462's "kept for further consideration" language — is arguably a specific design not literally shown in a single reference.
3. The "second efficiency while all values are zeroes" step is a distinctive claim limitation; a patentee could argue Lim's threshold test is not the same as computing and comparing two efficiency values.
4. Teaching away is weak here, but the patent frames the prior heuristic as merely "unsatisfactory," not as something to be abandoned outright.

Against these: KSR does not require the prior art to show the exact two-stage flow, only a motivation with predictable results; the substitution of RD cost for the importance proxy, and the aggregation of per-block costs to a macroblock decision (Lim already aggregates), are predictable optimizations with a clear design incentive (better R–D performance at negligible complexity). Combination B in particular is strong because the inventors' own earlier papers already teach RD-cost comparison including a zero candidate.

Net assessment: Claims 1–7 and 11 present a strong prima facie § 103 case over Lim (JVT‑K049) in view of the H.264 Lagrangian RD-optimization art (Wiegand/Ostermann), and alternatively over Lim in view of the Narroschke 2005–2006 publications. This is a hypothetical analyst assessment, not a legal conclusion of invalidity.

7. Explicit uncertainty list

1. Content of several cited references not independently verified (tool-step limit): US 5,691,770 (Keesman), US 7,203,374 (Haaba), US 2002/0114392 (Sekiguchi), US 2006/0098733 (Matsumura), US 2007/0133891 (Jeong), WO 96/34495, US 2010/0220784 (Tanimoto). I asserted nothing about their specific teachings.
2. WO 2007/079964 A1 prior-art date: publication (Jul. 2007) post-dates the priority; whether it qualifies under § 102(e) as of its 2006 international filing date depends on language/designations I did not verify.
3. Combination C references (EP 1 761 069 A1; US 8,130,828) were identified through citation contexts, not full-text review, and are not on the '462's face.
4. Full dependent-claim wording (claims 8–10, 12+) was not retrieved; the dependent mapping for 2–7 rests on the earlier-generated summary and Patexia's excerpt.
5. Priority-family title discrepancy (US 2008/0225947 A1) remains flagged as a contradiction with the earlier summary; I did not fetch the document itself to settle it definitively.
6. The legal-status/expiration figures remain Google Patents' stated assumptions, not legal conclusions (carried over from the earlier section).