The next steps are about making the decoder ecosystem more useful to more builders:

• Extend the supported display list (more sizes, timings, and panel types)
• Make the server encoder hardware accelerated so any GPU with OpenGL capabilities can run it
• Refine the PCB to fit into a proper demo device enclosure and add battery support
• Create a simple, stupid hardware keyboard

From my own experience, display integration is always the hardest part when trying to keep hardware flexible. Every panel behaves differently — signaling, timings, voltage levels, initialization sequences — and that’s usually where otherwise modular designs suddenly become rigid. A big focus of this roadmap is making that problem easier, both for me and for anyone building on top of this system.

• Expand the supported display catalog with verified timings and wiring notes
• Keep the decoder firmware update flow simple so builders can switch panels without redesigning the whole device

The current host encoder is fully software-rendered. It already performs fast enough for responsive use even without heavy optimization, SIMD tuning, or parallelization.

Despite that, the goal is to push efficiency further by adding a hardware-accelerated path. The target is an encoder that can offload work to any GPU with OpenGL support instead of relying on vendor-specific APIs.

• Keep the existing software encoder as a reliable baseline
• Add GPU acceleration for higher throughput and lower CPU load
• Ensure compatibility across different host systems and hardware

The intention is not to replace the current encoder but to extend it. Software rendering proves the architecture works; hardware acceleration expands performance headroom.

The current prototypes prove the concept. Next is turning it into a clean demo device: fewer hacks, cleaner routing, proper connectors, and something that can be assembled repeatedly.

• Refine PCB layout for compact integration
• Connector choices optimized for assembly and servicing
• Mechanical constraints aligned with a real enclosure

Touchscreens are fine for demos, but the demo device needs real keys. The plan is not a fancy keyboard. It’s a simple, reliable input module that makes the device usable without staring at a glass slab.

• Basic layout optimized for terminal and navigation
• Low part count, easy to build
• Designed as a module that others can replace or redesign