Valiantsin Dzerakh and Claude Opus 5.5
234a10f413
With sync_ahead an exit line gives the run total of the waits at a pool invalidate, as the shadow log lines stop at frame 1800 without TOTK_TRACE=perf
...
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com >
2026-11-18 18:55:13 +03:00
Valiantsin Dzerakh and Claude Opus 5.5
bc579c62e8
The qstamp patch set is switched off when sync_ahead is refused, so TOTK_SYNC_AHEAD=1 without shadow storage leaves the game's depth query check as it was
...
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com >
2026-11-18 18:55:12 +03:00
Valiantsin Dzerakh and Claude Opus 5.5
ee6d5a6183
Merge dev into task/pr1275
...
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com >
2026-11-18 18:54:52 +03:00
Valiantsin Dzerakh and Claude Opus 5.5
b9f104ef6b
Merge pull request #1276 : dlss_fixed_exposure creates DLSS without auto exposure, so distant fogged blocks in the Depths no longer change brightness for a frame (ipavlin98)
...
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com >
2026-11-18 18:47:52 +03:00
Valiantsin Dzerakh
792d5398ab
Merge branch 'pr1276-head' into task/pr1276
...
# Conflicts:
# docs/agents/symptoms.md
# runtime/cfg.c
2026-11-18 18:46:16 +03:00
Valiantsin Dzerakh and Claude Opus 5.5
65e6dc50d3
Merge pull request #1284 : on 1.4.3 DLAA's jitter goes into the game's camera instead of the viewport, so water seen at a low angle stops flickering (ipavlin98, #1126 )
...
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com >
2026-11-18 18:34:05 +03:00
Valiantsin Dzerakh and Claude Opus 5.5
c145f2bd5d
The camera jitter checks where each copy puts its matrix in the block: f_70000e31d4 takes the offsets from the shader's layout (x8), so only a projection at 0x70 and a view-projection at 0x30 take the offset, a projection whose view-projection lies elsewhere gets its centre terms back, and a block counts as given only once both matrices carry it ( #1126 )
...
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com >
2026-11-18 18:29:28 +03:00
Valiantsin Dzerakh and Claude Opus 5.5
902d91a94e
Merge pull request #1275 head into task/pr1275
...
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com >
2026-11-18 18:27:57 +03:00
Valiantsin Dzerakh and Claude Opus 5.5
0ae38dd34a
Merge pull request #1278 : a VRAM mirror copies from the shadow only after the shadow sync, so loads with sync_ahead and replay_workers no longer leave stretched slabs (ipavlin98)
...
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com >
2026-11-18 18:26:25 +03:00
Valiantsin Dzerakh
c1449b2a9b
Merge commit 'refs/pull/1284/head' of https://github.com/valentderah/totk-pc-recomp into task/pr1284
...
# Conflicts:
# docs/agents/symptoms.md
2026-11-18 18:22:56 +03:00
Valiantsin Dzerakh and Claude Opus 5.5
6ed99acddb
Merge pull request #1283 : water volume mask draws get a depth bias toward the viewer, so puddles stop flickering (ipavlin98, #1126 )
...
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com >
2026-11-18 18:20:25 +03:00
Valiantsin Dzerakh and Claude Opus 5.5
0350d91b72
The water volume mask bias is decided once per state change and only for LEQUAL and GEQUAL draws ( #1126 )
...
prepare_draw sets ex->volume_bias where the draw's dynamic state changes, so record_fixed reads a flag instead of
checking the pipeline state on every draw. A census of the matching draws in the field, the town and rain found two more
states with the same blend and depth signature, both tested GREATER: the volume's back faces into the same channel and
another shader's front faces into red. GREATER's equal case fails, and the bias toward passing turned the game's tie
the other way for them; they are now left as the game sets them.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com >
2026-11-18 18:18:22 +03:00
Valiantsin Dzerakh and Claude Opus 5.5
593eeeb8ed
The lock order comments name the new edge: mirror_sync and mirror_create take shadow_lock under mirror_lock
...
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com >
2026-11-18 18:08:44 +03:00
Valiantsin Dzerakh and Claude Opus 5.5
402d1fe8b0
Merge pull request #1278 head into task/pr1278
...
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com >
2026-11-18 18:07:48 +03:00
Valiantsin Dzerakh
a111ca77b3
Merge commit 'refs/pull/1283/head' of https://github.com/valentderah/totk-pc-recomp into task/pr1283
2026-11-18 18:06:22 +03:00
Valiantsin Dzerakh and Claude Opus 5.5
13ff0d7b52
Merge task/verfix: the setup recompiles when the release changes, so the menu shows the installed version ( #1078 )
...
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com >
2026-11-18 13:33:37 +03:00
Valiantsin Dzerakh and Claude Opus 5.5
d26b12b70c
Merge task/bgpad: gamepad input keeps working while the game window has no focus on Linux ( #971 )
...
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com >
2026-11-18 13:33:37 +03:00
Valiantsin Dzerakh and Claude Opus 5.5
cedc0a63ad
The setup's compile step runs again when the release string changes, so totk.exe carries the version of the release that built it ( #1078 )
...
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com >
2026-11-18 13:32:47 +03:00
Valiantsin Dzerakh and Claude Opus 5.5
a6ad6af274
Gamepad input keeps working while the game window has no focus: SDL's background joystick events hint is set before SDL starts ( #971 )
...
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com >
2026-11-18 13:30:52 +03:00
ipavlin98
85888bc9ba
Symptoms: the world behind the inventory streaked with DLAA, from the camera jitter's read-back taking the inventory camera's off-centre projection for a jitter
2026-11-18 10:50:58 +03:00
ipavlin98
04146289ac
The camera jitter's read-back takes a block's centre terms as an offset only when this code gave that block that offset: the inventory's camera has an off-centre projection of its own, which read as a jitter of 1152 pixels, so DLSS smeared the world behind the inventory into horizontal streaks
2026-11-18 10:42:23 +03:00
Valiantsin Dzerakh and Claude Opus 5.5
44450a15a0
Merge task/gyro: the gyro drift learning switch on the Controls tab ( #292 , #1267 )
...
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com >
2026-11-18 09:17:28 +03:00
Valiantsin Dzerakh and Claude Opus 5.5
18482e0af0
Gyro drift learning at rest gets its switch on the Controls tab, below the gyro calibration row
...
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com >
2026-11-18 09:17:28 +03:00
Valiantsin Dzerakh and Claude Opus 5.5
c3ec7fce0e
Merge task/gyro: gyro through Steam Input, levelled from the first reading, optional drift learning at rest ( #1282 , #292 , #1267 )
...
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com >
2026-11-18 09:12:06 +03:00
Valiantsin Dzerakh and Claude Opus 5.5
9f2a9d3e59
Gyro drift learning at rest is off by default: subtracting a learned bias made aiming feel worse in play before, so it stays a switch for players with a drifting controller
...
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com >
2026-11-18 09:12:00 +03:00
Valiantsin Dzerakh and Claude Opus 5.5
96a809044d
Steam Input motion: buttons move to Steam's virtual gamepad once it is up, a controller Steam hid keeps its own bias as the pad, and a sensor SDL could not turn on is logged with SDL's error ( #1282 )
...
1. scan_pads took a controller Steam hid when the virtual pad was not up yet and kept it for the session, so Steam's
remapping was lost. While the open pad is a hidden one, pad_physical looks again every RESCAN_MS and scan_pads
moves the buttons to the virtual pad; the hidden controller stays open as motion_extra, so motion goes on with
the same bias and orientation.
2. pad_identity marked " (under Steam)" only on motion_extra; a hidden controller that became the pad lost the mark
and could take the bias calibrated with Steam closed. The mark now follows steam_hidden_id.
3. After pull request #1024 : sensors_ready logs SDL_GetError when the pad has both sensors but enabling them failed,
instead of "has no gyro and accelerometer".
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com >
2026-11-18 09:08:52 +03:00
2dbdcf9a37
Read the gyro from the controller behind Steam Input's virtual gamepad when the game is started from Steam, from the patch in issue #1282
...
A game Steam starts gets SDL_GAMECONTROLLER_IGNORE_DEVICES with the controllers Steam reads itself and sees Steam's
virtual gamepad, which has no motion sensors. input.c steam_reveal keeps that list and clears the hint before
SDL_Init (only with motion on); scan_pads still takes the virtual pad for buttons and sticks, so Steam's remapping
holds, and motion_source opens the hidden controller for its gyro and accelerometer only. A bias belongs to the pad
as Steam presents it: " (Steam Input)" or " (under Steam)" in gyro_bias_pad. docs/controls.md, "Steam Input and
motion", says the Steam layout must bind the gyro to something or Steam zeroes it on the Steam Controller.
The patch as posted against 6151ac4 , with sensors_ready merged with #1274 's restart. Tested by the author on Linux
(Steam Controller) and the Steam Deck; not on Windows.
Co-authored-by: St5sz2Jk <St5sz2Jk@users.noreply.github.com >
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com >
2026-11-18 09:08:13 +03:00
45a0cd1f6c
Learn the gyro drift left after calibration while the controller lies still and subtract it, so the aim stops sliding on a table, from pull request #380 ( #292 , #1267 )
...
The calibration removes the bias measured once; a bias that moves as the controller warms up, or a controller never
calibrated, still turns the aim slowly. motion_fusion.c drift_track learns it from 256 ms windows in which the gyro
stays within 0.04 rad/s and the acceleration within 0.03 G peak to peak, after half a second of rest, with a time
constant of 0.25 s until the estimate settles and 30 s after. The estimate sits on top of the calibrated bias and is
cleared with a new controller, a new or loaded calibration and a change of gyro_calibration. New setting
gyro_auto_bias (live, default 1).
Taken from #380 : the rest window by peak to peak, the two time constants, the per-axis limit. Left out: the
0.006 rad/s dead zone, which ate slow aiming; the one global state shared by every fusion instance; the reset at
every StartSixAxisSensor; the 0.5 G acceleration gate, which let a controller in motion count as still. The first
estimate is limited to 0.1 rad/s from zero instead of 0.15; a larger bias is the calibration's job.
Also: a sample without a reading at a restart (#1274 ) no longer returns before the calibration's log and save.
--probe-services checks the drift with the switch off and on, after a calibration whose bias then moved, a slow
aim in the hand, and the levelling of #1274 with a bias on every axis.
Co-authored-by: airquark <97342227+airquark@users.noreply.github.com >
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com >
2026-11-18 09:06:30 +03:00
ipavlin98
d5d62ab406
The camera jitter is described as dlaa_cam_jitter, and the game's matrix copies skip it at once while DLSS is not running
2026-11-18 09:01:16 +03:00
ipavlin98
77cb787de7
The water volume mask's depth bias can be turned off with volume_mask_bias = 0 (or TOTK_VOLUME_MASK_BIAS), a test setting to compare against
2026-11-18 09:01:00 +03:00
ipavlin98
706144d673
Use the zero rate vector for the zero reading in the fusion start probe
2026-11-18 09:00:12 +03:00
ipavlin98
b155be1881
Level the six-axis orientation from the first real accelerometer reading and pull it toward gravity at 4/s while the controller is held still, so a bow aimed right after the game starts or after a load no longer turns sideways when the controller is tilted up or down
2026-11-18 09:00:12 +03:00
ipavlin98
0eab458030
Symptoms: water at a low angle flickering with DLAA from the viewport jitter the game's camera did not share; cfg knows DLAA_CAM_JITTER
2026-11-18 08:50:22 +03:00
ipavlin98
b5936a6aa9
The camera jitter moves on with the game's presents instead of at a camera block written again: the game fills two sets of blocks in turn, so each phase went to two frames, and each view-projection takes the offset its own block's projection took
2026-11-18 08:36:50 +03:00
ipavlin98
121b4fdf33
On 1.4.3 DLAA's jitter goes into the game's camera instead of the viewport: the projection's centre terms and the view-projection of each perspective camera with the scene's aspect take the offset where the game copies them into their uniform blocks, so the game's full-screen passes rebuild positions with the camera the scene was drawn with, and screen-space reflections on water seen at a low angle no longer flip between hit and miss with the jitter's phase
2026-11-18 08:13:26 +03:00
ipavlin98
cc84958cee
Give TotK's water volume mask draws a depth bias toward the viewer, so a puddle's surface, on which the volume's top face lies, counts as inside the volume in every frame instead of by rounding and the camera's sub-millimetre motion, which turned the puddle's reflection on and off in patches from frame to frame, more with DLAA
2026-11-18 05:41:05 +03:00
ipavlin98
07f9760971
Symptoms: stretched slabs over the ground in the Depths from a VRAM mirror copied from the shadow too early
2026-11-18 02:23:21 +03:00
ipavlin98
0f2cc1071a
A mirror's copies from the shadow wait for the shadow sync: the prologue seals the mirror queues before it syncs the shadow and copies only those, and a mirror sync touches its shadow pool, so with replay_workers and shadow_defer a recording thread no longer queues a copy that reads the shadow before the CPU's write reached it, which left terrain in the Depths as stretched slabs over the ground for the rest of the session
2026-11-18 02:03:53 +03:00
ipavlin98
efb896235a
DLSS without auto exposure is a setting, dlss_fixed_exposure, next to the other DLSS settings, off by default: 0 creates DLSS with AutoExposure as before, 1 without it, and a change recreates the features at once
2026-11-18 00:58:32 +03:00
ipavlin98
ae5fff3d46
With sync_ahead the game takes its depth query results when their frame number is one frame older than it expects, so in shrines the reflections of the lamps on the floor no longer go out on the frames where the GPU runs behind, which flickered at some camera angles
2026-11-18 00:57:06 +03:00
ipavlin98
68078948f5
With sync_ahead an invalidate of pool storage waits for the GPU write of a value the game was told is done when the range holds no older result, so the pixels TotK reads back once through a buffer per request are no longer zeros or an answer from the ring's last round, which in the Depths put Link in gloom on clean ground; buffers the game reads every frame still get the result of the frame before without a wait
2026-11-18 00:21:46 +03:00
ipavlin98
fd08b5cee3
DLSS is created without auto exposure, so large distant blocks behind the fog in the Depths no longer turn darker or lighter for a frame while the camera moves
2026-11-18 00:21:25 +03:00
Valiantsin Dzerakh and Claude Opus 5.5
6b0974f08d
Merge pull request #1188 : opt-in quick save and quick load on rebindable keys with no menu on the screen, game 1.4.3 (itaskaev2)
...
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com >
2026-11-17 20:17:57 +03:00
Valiantsin Dzerakh and Claude Opus 5.5
52d913c7e0
Quick save takes its request back with an atomic and, and the no-quick-save note names the key by its role, as it can be rebound ( #1188 )
...
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com >
2026-11-17 20:17:57 +03:00
Valiantsin Dzerakh and Claude Opus 5.5
358b7916d8
Merge task/hud: hud_width keeps the HUD within the middle 21:9 or 16:9 of a wide screen, from pull request #1262 (Marjaperse, ProSep, #643 )
...
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com >
2026-11-17 20:15:16 +03:00
Valiantsin Dzerakh and Claude Opus 5.5
7e2712e87e
Quick save logs once that it is 1.4.3 only on other game versions, and a quick load under way goes on while the settings menu opens, so the held picture always ends ( #1188 )
...
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com >
2026-11-17 20:15:04 +03:00
Valiantsin Dzerakh and Claude Opus 5.5
96afa531c9
Merge pull request #1131 : with frame-ahead on, fence write-backs no longer wait for the GPU about every ninth frame (ProSep)
...
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com >
2026-11-17 20:13:01 +03:00
Valiantsin Dzerakh and Claude Opus 5.5
6efa75f3dd
romfs_shapes_check.py links its harness with lld, which MSYS2 UCRT64 Clang needs for LTO objects
...
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com >
2026-11-17 20:12:56 +03:00
Valiantsin Dzerakh and Claude Opus 5.5
c781238d6d
The Shapes step keeps its old key when it writes no capped HUD folder, so an install with hud_width full does not prepare its shaped folders again ( #643 )
...
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com >
2026-11-17 20:12:56 +03:00
Valiantsin Dzerakh and Claude Opus 5.5
64e965078c
wb_eager moves to the diagnostics settings and out of totk.ini.example, as part of sync_ahead
...
It acts only with sync_ahead in effect and is on by default there, so a player who turns on sync_ahead gets it; 0 and check stay for comparing. docs/settings.md keeps the row.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com >
2026-11-17 20:12:19 +03:00