Skip to content

SpaceMouse: calibrate devices that are silent at rest, fix the pan Y direction, zoom on push/pull only - #983

Open
alexanderalber wants to merge 3 commits into
storytold:mainfrom
alexanderalber:spacemouse-silent-rest
Open

alexanderalber wants to merge 3 commits into
storytold:mainfrom
alexanderalber:spacemouse-silent-rest

Conversation

@alexanderalber

Copy link
Copy Markdown

Follow-up to #650, tested on real hardware (#623): classic SpaceNavigator 046d:c626, Windows 11, no 3DxWare.

Calibration never finished on a device that is silent at rest. The SpaceNavigator sends no reports while the cap rests. Measured over 40 s: [0 ×17, 38, 125 ×8, 16, 0 ×14] reports per second, all axes exactly 0 when it starts again. The rest calibration waits for 16 resting samples in 400 ms, so it finished only by chance, from a light touch, and used that touch as the bias. Now 300 ms without an axis report counts as rest at zero (Filter::quiet). Devices that stream at rest still calibrate through the existing window. My original measurement in #623 claimed the device streams at rest, which was wrong. Sorry.

Other fixes from the device test:

  • Pan Y was inverted: HID +Y points towards the user, the drawing's +Y away from them.
  • Twist no longer zooms: next to push/pull it felt odd on the device (I had suggested it in SpaceMouse (gaps.md H1): raw HID works natively on Windows without 3DxWare, measurements from a SpaceNavigator #623).
  • Zoom direction reversed: lifting the cap zooms in, pressing it zooms out. Having tried both on the device, this one is clearly more natural to me. "Reverse zoom direction" still flips it.
  • The 3D Mouse dialog no longer blocks navigation. Other dialogs still do.
  • Default pan sensitivity 0.8 → 1.2 (felt slow on the device; taste, drop if you disagree).
  • Help text "push/pull or twist to zoom" → "push or pull to zoom" in all nine catalogs (the twist clause removed, nothing else changed).
  • contributors/people.toml: my own entry (real name), as AGENTS.md describes.
  • docs/hardware-parity.md and docs/gaps.md H1: what Add raw-HID SpaceMouse 2D navigation #650 covers and what is left (macOS, web, orbit, 3DxWare coexistence). These rows overlap with the open audit PR docs: re-audit gaps.md and parity docs against main #819, which notes Add raw-HID SpaceMouse 2D navigation #650 as untested on a device; whichever merges second has a small conflict there, and this PR has the tested state.

Tests: spacemouse::motion::tests (silent calibration, reports postponing it, pan directions, twist not zooming, hostile reports). cargo xtask ci passes. Checked on the device: works right after start and after the calibrate button, pan in both directions, zoom in both directions, twist ignored, navigation with the 3D Mouse dialog open.

🤖 Generated with Claude Code

…m on push/pull only

Tested on a SpaceNavigator (046d:c626, Windows 11, no 3DxWare). It sends no
reports while the cap rests, so the rest calibration from storytold#650 waited for
resting samples that never came and finished only by chance. 300 ms without
an axis report now counts as rest at zero; devices that stream at rest keep
the window calibration.

Also from the device test: pan Y was inverted (HID +Y points towards the
user), twist no longer zooms, the 3D Mouse dialog no longer blocks the
device it configures, and the default pan sensitivity goes from 0.8 to 1.2.
gaps.md H1 and hardware-parity.md record what storytold#650 covers.

generated with Claude Code
Tried both directions on a SpaceNavigator; this one is clearly the more
natural. "Reverse zoom direction" still flips it.

generated with Claude Code
generated with Claude Code

This branch has not been deployed

No deployments
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant