Files
DroidDeck/docs/development/astra-viture-stereo.md
T
maxjivi05 bdf876d47c Glasses: remember a 3D fallback per model, docks as display only, VR timing fixes
- A 3D attempt that ends at the 640x480 fallback is remembered for that glasses
  model, so Play SteamVR no longer blacks out the glasses after every restart;
  Retry clears it and "3D for VR games" stays reachable
- A VITURE Mobile Dock on its own no longer counts as glasses; through a dock
  Play SteamVR is hidden and the sheet shows the output and why there is no control
- HDR is probed on the glasses when the session starts there
- Glasses frames with a mismatched checksum are marked in the status frame log
- SteamVR layer and driver vblank counters stay monotonic across refresh changes;
  one-eye render scale applies from activation and on mono/3D switches
- Compositor binds announce the refresh clients were last told; the presenter
  retries a failed first swapchain
- Dead strings and helpers removed; docs updated
2026-10-14 11:30:22 -04:00

131 lines
8.5 KiB
Markdown
Raw Blame History

This file contains ambiguous Unicode characters
This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.
# Astra 2 and VITURE stereo
## Tested connection
2026-10-14: REDMAGIC Astra 2, model `NP06J`, Android 16,
`RedMagicOS11.5.16_NP06J_GB`, Snapdragon `SM8850P`; VITURE Pro XR
(`35ca:101d`), firmware `07.0.03.013_20250422`. The tablet reports a locked
bootloader and green verified boot. Tests used the ordinary ADB shell, without root.
2D negotiates **1920×1080 at 120 Hz**. The glasses accept stereo command
`0x0008 [32]` with reply `00 32`, but Android then exposes **only 640×480 at
60 Hz**. An accepted USB command is not proof of a working stereo signal.
In this state the glasses reject a 2D command (`01 31` or `01 34`) and their
mode query returns `01 00`. A physical cable reconnect restored 1920×1080/120
with this update; Android then requested USB permission again and control resumed.
On the fresh attempt, `0x030D [03]` arrived 3.1 seconds after the accepted stereo
command. The vendor SDK maps that report to 2D/120 (`0x34`); the subsequent mode
query was rejected. The report does not encode a fallback reason or establish a
fixed timeout. The earlier command acknowledgement cannot prove stereo retention.
## What was checked
| Route | Result on this firmware |
| --- | --- |
| Pro full SBS command `0x32` | Accepted by MCU; physical output falls to 640×480 |
| Android preferred mode 3840×1080/60 | Preference saved; physical supported modes remain 640×480 only; preference cleared after test |
| USB source/host → source/device | Request issued; actual port stayed source/host |
| USB data disable/enable | Data status changed and was restored; physical display remained 640×480 |
| `IUsbManager.resetUsbPort` through ADB | HAL callback reported success (`0`); physical display mode did not change |
| External-display power off/reset | Did not recover the signal; glasses still rejected 2D |
| Physical cable reconnect | Restored 1920×1080/120 and USB controls after the permission prompt |
| OEM mirror/3D controls, source review | Change virtual-display routing, not physical stereo timings |
| OEM resolution selection, source review | Chooses cached resolution candidates and prefers stale 16:9 mode |
| Pro startup mode / half SBS, SDK review | No supported Pro startup-mode setting or half-SBS input |
Public driver decompilations from the related REDMAGIC 11 Pro (`NX809J`, SM8850)
[retain an active mode override](https://github.com/Coding-BR/android_kernel_nubia_sm8850_qwjujube/blob/efee6f411184886b0fbdaf63fc9db36b3ed96036/decompiled/msm_drm/dp_display_get_modes.c#L37-L44)
across a glasses hotplug. Their [USB-PD disconnect path](https://github.com/Coding-BR/android_kernel_nubia_sm8850_qwjujube/blob/efee6f411184886b0fbdaf63fc9db36b3ed96036/decompiled/msm_drm/dp_display_usbpd_disconnect_cb.c#L127-L147)
clears that override; their simulated hotplug path does not. This is consistent
with the Astra's measured behavior, but does not prove its exact kernel
implementation. That driver also [validates saved user modes against EDID](https://github.com/Coding-BR/android_kernel_nubia_sm8850_qwjujube/blob/efee6f411184886b0fbdaf63fc9db36b3ed96036/decompiled/msm_drm/nubia_edid_modes.c#L226-L250),
so active overrides and saved preferences must not be treated as the same
mechanism. The independent [MagicDesk investigation](https://github.com/mekhontsev/magicdesk/blob/main/docs/xr-resolution-fix.md)
documents a related failure on its separately identified REDMAGIC firmware.
The [VITURE SDK matrix](https://www.viture.com/en-SG/developer/glasses-sdk/glasses)
specifies full SBS at 3840×1080/60 for Pro. Persistent startup mode is restricted
to Luma/Luma Pro. Among the Luma families supported here, 90 Hz stereo is an
Ultra capability. The connected Pro firmware matches both the N6PL entry in the
[public updater catalog](https://static.viture.com/dfu-util/js/vers.js) and the newest
N6P entry in the [public release-note service](https://cloud.viture.dev/api/v1/system/releasenote)
checked on 2026-10-14. This does not rule out unpublished, device-specific updates.
## Powered dock results
A VITURE Pro Mobile Dock (`35ca:3011`) was connected between the Astra and Pro XR.
Its official read-only P9 version query returned `2.01.03.0.03` with a valid
checksum. The Astra negotiated 1920×1080/120, but enumerated only the dock's
USB interfaces, not the glasses' `35ca:101d` control/IMU interfaces. A one-second
DroidDeck IMU sample returned zero packets.
DroidDeck treats the glasses as a display behind the dock: no USB permission
prompt, video at 1080p/120 on the glasses, no Play SteamVR, refresh, 3D or tracking
settings. The dock alone, without glasses, does not count as glasses.
VITURE's updater calls this dock P9 (app `35ca:3011`, bootloader `35ca:3010`). Its
HID messages cover dock versions, battery, inputs and LEDs; none sets a display
mode or forwards glasses commands or IMU data. `2.01.03.0.03` is public release
`1.03.0.03_20250709`; the latest, `1.03.0.06_20260409`, adds Beast support and
updater fixes only. The updater also reads a Lontium LT8668 in the dock, a
DisplayPort receiver with its own EDID handling, so the tablet most likely sees the
dock as its DisplayPort sink rather than the glasses.
[VITURE's Pro Dock specifications](https://www.viture.com/product/viture-pro-mobile-dock)
confirm 3840×1080/60 video support and exclude phone features requiring
communication with the glasses. The vendor SDK also treats the dock as a separate
accessory, not an IMU device. This dock may test the video path, but it does not
provide DroidDeck's current head-tracking path. Stereo negotiation and mode
retention during an upstream-only reconnect still require a physical test.
## No-root hardware design to test
A powered pass-through connection may let the tablet see the stereo EDID on a
fresh USB-PD connection while the glasses retain their stereo mode:
1. Power the adapter and glasses independently of the tablet connection.
2. Enable stereo on the glasses.
3. Disconnect only the tablet's upstream USB-C connection promptly after the
stereo acknowledgement, keeping the glasses powered. Compare with a delayed
disconnect to test whether the observed mode report affects retention.
4. Reconnect the tablet and check its physical supported modes before starting a game.
This requires an adapter that preserves glasses power and mode during the
upstream disconnect, passes 3840×1080/60 video, and forwards USB HID for controls
and head tracking. [VITURE's Adapter Ultra documentation](https://www.viture.com/academy/adapters/usb-c-xr-charging-adapter-ultra)
explicitly excludes wired glasses data on Android, so it does not satisfy
DroidDeck's current USB control/tracking path. Its power and video specifications
also do not promise mode retention across upstream disconnect. This design
remains **unverified**, not a purchasing recommendation. First verifying stereo
retention through the adapter on a compatible host would provide a useful control.
An intermediary presenting the stereo EDID before first host
connection is another possible design; it must pass the actual stereo timing
and preserve HID access, not merely advertise a wider desktop.
VITURE advertises full Android functionality for its separate
[USB-C to Glasses and Charging Adapter](https://www.viture.com/en-MY/product/usb-c-to-glasses-and-charging-adapter).
That makes it a candidate for the USB-data experiment, not a confirmed fix.
First check USB enumeration, brightness controls and live IMU data through the
adapter; then test power/mode retention and the stereo timing.
Success requires Android to expose and select 3840×1080/60, different left/right
eye images to reach the lenses, live head tracking, and repeatable return to 2D.
A wide screenshot or `wm size` override alone does not meet those conditions.
## App safeguards
DroidDeck waits up to 20 seconds for physical stereo confirmation. When the
attempt ends at a low-resolution fallback, DroidDeck remembers it for that glasses
model on this device, so Play SteamVR stops requesting 3D after a restart; **Retry** in the
glasses settings clears it. A display that stays missing, such as an unanswered
cast prompt, fails only that attempt. Automatic retries do not loop.
USB mode reports are tracked separately from the physical display. When a session
enabled 3D, stopping it requests a return to 2D, including during negotiation;
this firmware can still require a cable reconnect. Output timing follows
the active glasses display rather than the tablet's requested refresh rate.
Keep hardware experiments separate from game loads. End the session, force-stop
DroidDeck, and reopen its home screen after each game test. Read current HAL
temperatures rather than cached thermal-service samples.