Repository navigation
Conversation
Held upright, as the console is in handheld, a turn to the side reads mostly on the pad's roll axis, so games that aim with yaw alone moved the camera far less side to side than up and down. The yaw sent is now the turn about the vertical the accelerometer finds (player space, with the Gyro Wiki's 1.41 relaxation). Pitch, roll and gravity stay the pad's own. Every 10 s a moving pad logs the angle it read against the angle the host was given, so drift on the way can be measured. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Relink, the game this was tried with, already turns the gyro about the player's vertical itself. The yaw this title rewrote put the same turn on two axes, so the game's estimate of up tipped and the aim sank while turning back to the left. The gyro goes out as the pad reads it again. The 10 s summary now also pairs the gyro's pitch between two steady poses with the tilt gravity measures between them: their ratio is the gyro's scale, which hid names no unit for. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The scale check found no pose in a handheld run: its steady limits (5 deg/s, 3% of 1 g) were tighter than a held console, and it only looked at tilts about x. The 10 s summary now gives each wire axis's turn, net and loss, the mean gravity, how much of the window was calm and near 1 g, and, between steady poses (10 deg/s, 8%), the gyro's response to each axis gravity saw turn about: (1, 0, 0) for x when hid's unit and axes are right. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Why
Streaming to Relink (native gyro, host pad
joyconpair), left/right aimed less than up/down; on a PC with the Joy-Cons wired directly it felt right.The first commit turned the yaw about the player's vertical. On the console the aim then sank while turning back to the left: Relink already does that conversion itself (
host/hos_motion.h: an up vector from a complementary filter, yaw = gyro · up, pitch pulled toward gravity's tilt). The rewritten yaw put the same turn on two axes, which tipped Relink's up and switched off its pitch pull past 15° of roll. The second commit sends the pad's own gyro again.Because Relink pulls pitch toward gravity and nothing anchors yaw, a gyro scale under 1 would show only as weak left/right. hid names no unit for the gyro (
pad_map::motionassumes turns per second, unverified).What
Every 10 s a moving pad logs, per wire axis (x right, y up out of the face, z toward the player):
x→(1.00, 0.00, 0.00)when hid's unit and axes are right. A turn about gravity itself is unseen, so upright the left/right axis has no check; up/down and the wheel do.Console
4df9f6c(player space), Relink in handheld: the sending loses at most 0.5° of yaw per 10 s window while turning 200–450°; the aim sank turning left, hence the revert.d40e7d6, Relink in handheld: loss under 0.1° per window; the x-only scale check found no pose (limits too tight for a held console). Large pitch totals beside small yaw ones while turning left/right, unexplained: the turn upright reads on z, which that build did not log.700e534: pending, a guided run (upright left/right, up/down, wheel; then flat).Test
cargo fmt --check,cargo test -p pf-nx-app, workspace clippy-D warningsin the host image.