[AlxcBTW]

What

The captured mouse (ToggleMouse) polled SDL_GetRelativeMouseState every 33 ms and tilted the right stick by clamp(distance * sensitivity + 16, 64, 128): any movement tilted the stick at least halfway, the tilt depended on the poll interval, and nothing could be tuned except MouseSensitivity (#4009).

Mouse motion is now accumulated from SDL_MOUSEMOTION events and turned into a speed in counts per second, so the same hand speed gives the same tilt at any frame rate. The keyboard and mouse logic moves out of PadInput into Pad::KeyboardMouse (libScePad/include/KeyboardMouse.hpp), which takes its time as an argument and so can be tested offline; PadInput keeps the SDL calls.

• Speed is smoothed with an exponential filter, capped at MouseMaxSpeed, shaped by MouseCurve and lifted by MouseDeadzoneOffset, so slow movement clears the game's own deadzone without the old half-tilt jump.
• The stick is centered MouseDecayMs after the last movement.
• Six new settings next to MouseSensitivity: MouseVerticalScale, MouseCurve, MouseDeadzoneOffset, MouseMaxSpeed, MouseSmoothingMs, MouseDecayMs. Invalid or repeated values stop initialization with file and line, like the existing settings.
• MouseSensitivity keeps its range (above 0, at most 10) and name. At 1, 3000 counts per second is a full tilt. The old behaviour is approximated by MouseDeadzoneOffset = 0.5, MouseSmoothingMs = 0, MouseDecayMs = 33; the defaults change how the mouse feels, deliberately.
• Capture sets SDL_HINT_MOUSE_RELATIVE_MODE_WARP=0 and SDL_HINT_MOUSE_RELATIVE_SYSTEM_SCALE=0, so the pointer speed of the operating system does not apply.
• Losing focus lets go of held keys and the wheel press and releases the capture.

docs/user/INPUT_MAPPING.md gains a "Mouse as the right stick" section and the settings table.

New test pad_keyboard_mouse (tests/KeyboardMouse.cpp, tests/PadSimulation.hpp) feeds timed events into Pad::KeyboardMouse and samples the pad state at 33, 8 and 1 ms polls: steady deflection follows the curve, the stick centers within the decay time, a flick above full speed does not extend the tail, integrated deflection agrees across poll rates, plus the key, wheel and focus-loss paths. pad_input_mapping gains parsing and rejection cases for the new settings.

Tested

Not tested in a title; no game was run. Windows 11 build node (WinLibs GCC 15.2.0, Release, BUILD_TESTING=ON): full build and the non-GPU ctest set, including pad_input_mapping and the new pad_keyboard_mouse. tools/check_conventions.py --base origin/main passes.

Checklist

• Based on current main; no other open PR implements the same functions (feat(input): add native SDL input configuration with ImGui #674 edits bindings through an ImGui editor, feat(input): add browser-based input mapping configurator #3456 is a browser editor for the ini, feat(libSceVideoOut): capture the host cursor for guest mice and scale their deltas #4217 captures the host cursor for guest mice; none changes the analog conversion)
• One topic per PR; follow-ups go in a new PR
• No comments in code except technical debt
• Unimplemented paths throw (NotImplemented_nid_no_patch); silent stubs are listed in TechnicalDebt (none added)
• No notes, investigation .md files or images added to the repository
• New third-party code is a submodule built from source (none)
• Generic behaviour, not specific to one title
• Depends on: none
• Unblocks: none
• AI-assisted: yes (Claude Code)

🤖 Generated with Claude Code

[github-actions]

Other open pull requests touch the same code. Merging one will break the other, or both implement the same thing. If this one builds on them, list them in Depends on.

Pull request	Conflicting files (lines on main)	Same exports	Same new files
#4756	core/libs/prx/libSceVideoOut/src/PadInput.cpp L226-260

[xXPBSXx]

Windows test: head d7ca998dd merged onto current main ff8fe4dbc, together with 6 other open pull requests in one build (the merges were conflict-free); a failure would have been bisected per PR. Windows 10, MinGW GCC (WinLibs), Ninja Release, -j2: full build 0 errors; ctest for the tests this change adds or touches: pad_input_mapping passed, pad_keyboard_mouse passed; python tools/check_conventions.py --base upstream/main passed for the combined tree. I did not run the new tests against main without the change.

AI-assisted (Claude Code).

[adriadam10]

Checked on current main: merged as is, the build failed because PadInput::MouseModeEnabled() (added by #4217) still returned the removed mouseEnabled. I pushed one commit that makes it return keyboardMouse.MouseMode(), rebased on main.

• pad_input_mapping and pad_keyboard_mouse pass; the new test does not build without the change.
• Thanks @xXPBSXx for the Windows run.

Merging once CI is green.

[AlxcBTW]

Thanks @adriadam10. That's the right fix: MouseModeEnabled() came in with #4217 after my last rebase, and keyboardMouse.MouseMode() is the state it should report now.