Skip to content
View Janus5G's full-sized avatar

Block or report Janus5G

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
.github/profile/README.md
Janus — optical datapath, spectral storage, open toolchains

Sponsor Profile Site Location

Independent systems architect and developer.
Programming languages · education and AI · Linux tools · distributed intelligence · experimental optical hardware.


How it began

This work did not begin with a grand roadmap. Curiosity about why glass-crystal data storage had seen so little visible progress inspired early ChromaPlex experiments: a programming language, an assembly layer, a compiler and the Cplex editor. Questions about heat led to ChromaPlex OS v2. Later, the curved lid of a container inspired the next line of optical thinking that became PRISME. An idea sparked by a video about AI-to-AI communication led to ChromaSpeechAI and eventually ChromaNeural.

Other projects emerged from quite different problems. ChromaLearn addresses pupils using AI to bypass learning; ChromaPress grew from the need for dependable, user-friendly Linux ISO tools; Project Pusher2 helps newcomers publish with GitHub. These projects deserve their own place, not a forced connection to optical computing.

Security · Functionality · Usability are guiding principles. Less unnecessary computation, less data traffic and better energy efficiency are objectives to measure—not performance claims that have already been established.

Explore by interest

Area Start here
Learning with AI ChromaLearn
Linux customization and development ChromaPress · Coding Agent
Beginner-friendly GitHub tools Project Pusher2
Programming languages and optical research ChromaPlex OS · PRISME
Distributed AI collaboration ChromaNeural

Architecture at a glance

PROGRAMMING AND SOFTWARE EXECUTION
   Cplex (Rust/Slint editor)
       ├─ Danish CPL → textual CPA → CPA assembler → CrystalSimulator
       └─ simple CPL → separate legacy bytecode assembler → VM
   Brainfuck source → CPA translator → CPA assembler → CrystalSimulator

ARCHITECTURAL CONNECTIONS
   ChromaPlex OS
      └─ ChromaPlex OS v2 (57 facets, moving light, multi-crystal model)
          └─ PRISME (R/G/B/V quaternary payload + UV check symbol)
              ├─ PRISME Binary Extension (standalone binary packaging)
              ├─ Q8 hardware engineering and optical research directions
              └─ ChromaSpeechAI concepts → ChromaNeural development
                                         └─ Cubic-4-inspired task division

Important distinction: these are connected ideas and software developments, not one verified end-to-end hardware stack. The CPA/CrystalSimulator path and the legacy bytecode/VM path are distinct. CPA is a working software assembly language, not yet a complete FPGA instruction processor on Q8 hardware. Physical storage, optical links and performance-per-watt require laboratory validation.


ChromaPlex — the foundation

ChromaPlex OS is an experimental language and simulation stack for colour-channel, voxel-addressed data structures and lossless exponent–remainder number representation.

Layer Verified role
CPL — ChromaPlex Language Higher-level source language for values, coordinates, colour channels, voxel storage, and output.
CPA — ChromaPlex Assembly Lower-level instruction language for registers, channel reads/writes, arithmetic, control flow, packing, and I/O.
Assembler + simulator Validates CPA, then executes it against sparse software-modelled 3D crystal voxels.
Representation Five logical channels: red, green, blue, violet, and UV; numeric values can use exponent–remainder pairs.

The implemented primary flow is:

CPL source → textual CPA → CPA assembler → CrystalSimulator

A separate Brainfuck → CPA translator is implemented in the ChromaPlex core. This is the verified compatibility input beyond CPL. No general translation layer for mainstream existing languages is claimed here.

Toolchain and authoring

chromaplex-os-compiler contains a separate CPL/CPA/VM toolchain with a limited CPL parser, bytecode assembler, virtual machine, simulated storage, and laser-control abstraction.

Cplex is the native Rust + Slint desktop editor for ChromaPlex/CPL. It uses a JSON bridge to bundled Python tooling and currently supports two CPL paths:

  • simple CPL → legacy assembly/bytecode path;
  • Danish CPL → CPA assembly plus a CHROMAPLEX_CPA_BUNDLE_V1 editor/simulator bundle.

Cplex is development tooling around ChromaPlex; it is not the language foundation itself.

ChromaPlex OS v2 — the next step

ChromaPlex OS v2 explores a 57-facet crystal model, multiple crystals, moving the light source instead of the crystal, software-modeled lens capture and write-once registration. The existing interactive software demonstration illustrates the concept; it does not establish physically tested glass storage. ChromaPlex OS v2 itself is unchanged by this profile update.


From ChromaPlex to PRISME

PRISME carries the colour-channel direction into a five-channel spectral byte mapping.

one byte
  │
  ├── R, G, B, V: four base-4 symbols = 8 payload bits
  └── UV: (R + G + B + V) mod 4 verification value

The four visible data channels represent all 256 byte values. UV is a verification channel, not a complete arbitrary multi-error ECC by itself. Higher-level error correction remains a separate layer.

Software evidence

Evidence What is supported Classification
All 256 byte values Encode → UV-check → decode in PRISME's published Python tests Reproducible software logic
Single-channel corruption 3,840 / 3,840 nonzero one-channel level changes detected by the modulo-4 check Exhaustive mathematical/software check
Reed–Solomon correction Public tests document the arithmetic of error/erasure limits Reference calculation, not integrated optical correction evidence
Physical optical operation Channel separation, noise, BER, write/read speed, power and durability Physical verification pending

The four R/G/B/V channels carry 8 payload bits (one byte) per modeled optical symbol; UV provides a separate 2-bit integrity value. Those are not 10 independent payload bits. UV detects single-channel level alterations but does not itself correct errors and may miss compensating multi-channel changes. Older precise long-run benchmark counts have been omitted from the profile because independent reproduction evidence for those runs is not available in the current public PRISME source tree.

A passive glass medium might retain information without standby electricity. That does not mean an operating reader, writer, server or datacenter uses zero power. Efficiency claims require measured, comparable baselines.


One foundation — multiple directions

Conventional-system interoperability

PRISME Binary Extension is a separate experimental repository for portable packaging for conventional binary systems.

It converts structured data or binary payloads into portable .prisme packages and can emit raw binary, manifests, and C/Rust-compatible byte arrays. The package format defines payload type, optional zlib compression, lengths, CRC-32, SHA-256, package ID, timestamp, and JSON manifest metadata.

Status: experimental specification and reference implementation; not production-validated. Its own rights-reserved license governs use, separate from the terms of other projects.

This extension is separate from PRISME's spectral mapping and is not a mandatory runtime dependency for ChromaPlex or PRISME.

Node and AI communication

ChromaSpeechAI is the color-symbol communication approach developed for AI-to-AI exchange and used in the ChromaNeural direction. Its optical proposal has four data colors, with seven shades each (28 base symbols); UV is reserved for verification/control, not a fifth data color. The representation was inspired by the PRISME work but is not identical to PRISME's four-level-per-channel byte mapping.

ChromaNeural is being developed to distribute tasks across AI nodes, combine and check results, and reuse useful verified knowledge. Its current public README identifies RC4 as withdrawn and RC5 as forthcoming. Internal development and reported tests are not the same as a complete published verification package.

Research area Status
ChromaSpeechAI network-based communication Implementation and LAN/WAN tests reported by the developer; publicly reproducible release evidence pending
Cubic-4-inspired task division, knowledge sharing and reuse Development underway; complete workflow verification pending
Optical lamp-and-camera communication Physical distinction of shades, timing, UV/control readout, BER and throughput remain to be measured
Fewer repeat queries and lower data/energy costs Hypotheses to benchmark, not established general savings

Hardware in the owen - with base module - 4 expansion cards and areana link for e-sport.

Independent projects — learning, Linux and accessible development

These projects were built to address different real-world problems. They belong alongside the optical work, rather than underneath it.

  • ChromaLearn — AI as a learning helper, not an answer shortcut. The published 0.4.5 evaluation release includes school roles and session-only student data handling within the app. Version 2 is under development; it must not be presented as already released.
  • ChromaPress — a Linux ISO customization, building and verification desktop workbench. The public v1.0.0a72 alpha includes evidence for a generated ISO booting in a VirtualBox live session; this is not proof of compatibility with every Linux installation or device.
  • Project Pusher2 — an approachable GUI for GitHub staging, commits and pushing. The public README describes v1.1.0 with opt-in automated releases for supported Electron projects.
  • ChromaPlex Coding Agent — a Linux coding/build environment (v0.4.0 alpha) with source integrity controls and target-aware security policy for desktop, web, embedded systems and CPL/CPA. Full GUI, target-device deployment and physical embedded validation remain outstanding.

A separate custom Linux and industrial device-programming direction is being developed to combine source-language tools, verification and deployment. Its proposed Raspberry Pi/ESP32/Arduino targets are not claimed to have been production-validated.


Project map

ChromaPlex, compilers and spectral research

Distributed AI, education and developer tools

All repositories retain their own licensing terms and validation boundaries. Software models and rendered concepts must not be presented as already-tested physical hardware.


Sponsorship supports public compilers, simulators, specifications and engineering documentation.
It does not sell equity, tokens or exclusive rights to public work.

github.com/sponsors/Janus5G

@Janus5G's activity is private