Sync the kernel copy with the warp 2/3 surround fold
build / windows (push) Canceled after 0s
build / release (push) Canceled after 0s

The kernel copy follows the new kernel: the OAMD parser and the speaker
controls accept warp modes 2 and 3, and the renderer pans the 5.1-family and
4.0 layouts in an expanded surround layout before folding the rear pair into
the target. Trim selection, balance and automatic compensation stay with the
final target layout, and the folded result feeds the existing ramp and mix.

Builds Release-Static|x64 and |Win32. layout_table_test and
channel_swap_test still pass; the render harness now renders a warp-3
stream instead of reporting the unimplemented mode.
This commit is contained in:
TheM14 committed 2026-11-03 13:36:40 +08:00
1 parent d35f6abe7a
commit 5b41868e94
4 files changed
+200 -18

No files matched your search

+5 -2
View File
@@ -131,8 +131,11 @@ the three trim levels and the two front-back balances. trim_bypass is indexed
by object slot (1..15); slot 0 carries the LFE and is unused.
warp_mode 0 leaves the floor Y coordinate unchanged; mode 1 maps it through
clamp(2Y, 0, 1). warp modes 2 and 3 have no speaker implementation and are
rejected by the caller that builds this structure.
clamp(2Y, 0, 1). Modes 2 and 3 render the 5.1-family and 4.0 layouts through an
expanded surround layout (7.1-family, or a rear pair added to 4.0 without a
centre or LFE) and fold the rear pair into the target surround pair: mode 2
uses the PLIIx coefficients (sqrt(3)/2 and 1/2), mode 3 the Standard (Lo/Ro)
coefficients (1 and 0). Layouts without a single surround pair render directly.
trim_mode selects where the gain comes from: automatic applies the layout
position compensation, disabled applies no trim, and explicit applies the