zpe-xr 0.3.0

Deterministic XR hand-pose transport codec. ~24x compression, sub-mm fidelity, sub-ms latency, CPU-native.

ZPE-XR

Quick Start

The steps below are repository verification guidance for the live ZPE-XR workstream. They are not a public-release install claim and they do not imply runtime closure on Unity or Meta targets.

git clone https://github.com/Zer0pa/ZPE-XR.git zpe-xr
cd zpe-xr
python -m venv .venv
source .venv/bin/activate
python -m pip install "./code[dev]"
python ./executable/verify.py

Optional direct test replay:

python -m pytest ./code/tests -q

Current live repo:

• clone target: https://github.com/Zer0pa/ZPE-XR.git
• package metadata: code/pyproject.toml
• install/verify path: python ./executable/verify.py
• optional direct replay: python -m pytest ./code/tests -q

Not provided today:

• no PyPI release
• no public-release verdict
• no runtime-closure pass narrative

License boundary:

• Free use at or below the SAL v6.2 threshold; see LICENSE for exact terms.
• SPDX identifier: LicenseRef-Zer0pa-SAL-6.0.
• Commercial, hosted, or legal-interpretation questions route to [e-mail].
• The release posture remains PRIVATE_ONLY until the comparator and runtime gates are actually closed.

What This Is

24× hand-pose compression at sub-millimeter fidelity and sub-millisecond latency. Every encode is byte-identical. ContactPose multi-sequence benchmark validated.

ZPE-XR is a deterministic codec for two-hand joint streams — built for XR platform teams and spatial-computing infrastructure where generic compressors destroy the sub-millimeter precision that downstream physics, gesture recognition, and replay depend on. Package candidate, not yet released.

Question	Answer
What is this repo for?	A deterministic XR codec and evaluation harness for two-hand joint streams, packaging, and proof-surface adjudication.
What is the current strongest claim?	The outward-safe ContactPose benchmark lane is real and the package mechanics are real. Public release readiness and runtime closure are not.
What is not being claimed?	No Photon displacement claim, no exact-PRD-corpus closure claim, no Unity/Meta runtime closure claim, and no public-release readiness claim.
Where should an outsider acquire it?	`https://github.com/Zer0pa/ZPE-XR.git` is the live workstream repo that matches this package and docs surface.

Field	Value
Architecture	HAND_JOINT
Encoding	QUAT_PACK

Commercial Readiness

Field	Value
Verdict	PRIVATE_ONLY
Commit SHA	b0a08fa02fd1
Confidence	62%
Source	proofs/FINAL_STATUS.md

Evaluators: Modern comparator gate not yet closed. pip install -e . to evaluate. Contact [e-mail].

• Completeness basis: 5 closed claims / 8 tracked claims | RELEASE_READINESS_JSON

Key Metrics

Metric	Value	Baseline
COMPRESSION	23.90×	vs Ultraleap 8.465×
MPJPE	0.48	mm
LATENCY	0.06	ms
VS_ULTRALEAP	2.82×	—

Source: proofs/artifacts/2026-04-14_zpe_xr_live_014204/phase5_multi_sequence_benchmark.json | proofs/artifacts/2026-03-29_zpe_xr_phase6_mac_comparator_arm64/phase6_mac_comparator_benchmark.md

What We Prove

Auditable guarantees backed by committed proof artifacts. Start at AUDITOR_PLAYBOOK.md.

• 0.4786mm mean per-joint position error | ContactPose benchmark
• 23.898× compression vs raw | two-hand joint streams
• 0.0575ms encode+decode latency
• Local package install and tests pass

What We Don't Claim

• No modern comparator gate closure (0/5)
• No Unity or Meta XR runtime integration
• No release readiness
• No production headset deployment
• Rotation encoding — codec encodes joint positions only (3 floats/joint), not full pose (7 floats/joint)
• 8-primitive directional encoding — codec uses int16 quantization + int8 delta compression with CRC32 integrity, not directional primitives
• float16+zlib baseline beats ZPE-XR on 5/5 ContactPose sequences in the Phase 6 comparator matrix

Area	Gap	Impact
Public release	Modern comparator gate failed (0/5).	Blocks public-release language.
Runtime closure	XR-C007 remains PAUSED_EXTERNAL.	Blocks runtime-readiness and engine-integration claims.
Comparator displacement	Photon displacement remains secondary and open.	No incumbent-displacement claim is admissible.
Exact PRD corpus	ContactPose is outward-safe but not the exact PRD corpus.	Limits corpus and market-fit language.

Tests and Verification

Code	Check	Verdict
V_01	ContactPose benchmark lane	PASS
V_02	Package mechanics	PASS
V_03	Cold-start audit	PASS
V_04	Modern comparator gate	FAIL
V_05	XR-C007 runtime closure	INC
V_06	Public release readiness	FAIL

Proof Anchors

Path	State
proofs/FINAL_STATUS.md	VERIFIED
proofs/RELEASE_READINESS_REPORT.md	VERIFIED
proofs/artifacts/2026-03-21_zpe_xr_phase5_multi_sequence_161900Z/phase5_multi_sequence_benchmark.json	VERIFIED
proofs/artifacts/2026-03-21_zpe_xr_phase4_runpod_contactpose_124835Z/phase4_contactpose_benchmark.json	VERIFIED
proofs/artifacts/2026-03-21_zpe_xr_phase4_cold_start/phase4_cold_start_audit.json	VERIFIED
release_readiness.json	VERIFIED

Repo Shape

Field	Value
Proof Anchors	6
Modality Lanes	5
Authority Source	proofs/FINAL_STATUS.md

Ecosystem

• ZPE-Mocap — Human motion codec and replay.
• ZPE-Robotics — Robot motion codec and search.
• ZPE-IMC — Multimodal codec platform surface.

Observability: Comet dashboard (public)

Who This Is For

Ideal first buyer	XR platform team or spatial-computing infrastructure team
Pain	Hand-tracking pipelines generate high-throughput joint streams that are expensive to store, transmit, and replay without fidelity loss
Deployment	SDK — Python package candidate with evaluation harness
Family position	Product candidate in the Zer0pa deterministic encoding family. ZPE-IMC is the umbrella integration layer

---

Competitive Benchmarks

Phase 6 Mac benchmark. Source: proofs/artifacts/2026-03-29_zpe_xr_phase6_mac_comparator_arm64/

Tool	Ratio	MPJPE mm	Evidence
Photon Fusion	38.3×	—	doc-derived; rotations only
ZPE-XR	26.1×	0.879	measured local; full positions
Ultraleap VectorHand	8.5×	—	code-derived
float16+zlib baseline	4.3×	0.277	measured local; modern comparator

Photon exceeds ZPE-XR on ratio (narrower data semantics). float16+zlib baseline beats ZPE-XR on fidelity (lower MPJPE) on 5/5 ContactPose sequences despite lower compression ratio — this is why the modern comparator gate fails 0/5. Unity Netcode / Normcore: blocked. Modern comparator gate: 0/5 FAIL.

Go Next

If you need to...	Open this
Read the current verdict and claim boundary	proofs/FINAL_STATUS.md
Read the release decision and blockers	proofs/RELEASE_READINESS_REPORT.md
Understand package/runtime surfaces	docs/ARCHITECTURE.md
Orient quickly or answer common questions	docs/FAQ.md
Read audit limits and non-claims	PUBLIC_AUDIT_LIMITS.md
Follow the shortest honest audit path	AUDITOR_PLAYBOOK.md

Contributing, Security, Support

Contribution workflow: CONTRIBUTING.md	Security policy and reporting: SECURITY.md	Support routing: docs/SUPPORT.md
Questions and quick answers: docs/FAQ.md	Public audit boundary: PUBLIC_AUDIT_LIMITS.md