This method of interpolation aims to correctly interpolate the render
transformation matrices as best as possible without doing an entirely
new render pass. This is accomplished by deconstructing the matrices
into its composition of transformations, interpolating those simpler
transformations individually, and then reconstructing the
transformation matrix with the interpolated transformations.
Local head rotation is now no longer copied to all other players
C-Up mode head rotation now sends to the other players through
the player packet so others can see the head move around
m->particleFlags are now synced to the other players now
Fixed the following audio bugs:
Bug: Rom-hacks sequences don't seem to be affected by volume scaling and
muting
Fix: Force the BGM sequences to follow the vanilla behavior:
Volume can't go higher than default volume
Volume is reduced to 31% when the game is paused
Audio is stopped when the game is paused outside the Castle levels
Bug: (Pointed out by Draco) Mario's voice clips are not replaced by the
player's character's in the following instances: fall to death
barrier, "here we go" in the ending cutscene, "let's a go"
after selecting a star, "okey dokey" after starting the game.
Fix: The first two ones now call
play_character_sound(m, CHAR_SOUND_...) instead of
play_sound(SOUND_MARIO_..., pos). The last two ones couldn't be
fixed the same way for two reasons: First, the corresponding sounds
were not referenced in the sound table, second, the sound played is
always cut-off after a few frames (due to how sm64 resets the sound
banks after loading a level).
Added SOUND_*_LETS_A_GO and SOUND_*_OKEY_DOKEY sounds for each playable
character as Bass samples.
Character Bass sounds work the same way as vanilla sounds (i.e. can be
played with play_character_sound), but they cannot be prematurely stopped
by sm64 sound banks shenanigans.
This fixes the cut-off for both the star select and the castle grounds
entry, plays the sound corresponding to the player's character, and doesn't
need to extend or edit the sound table.
DynOS can detect texture duplicates when generating a bin or lvl file.
When a duplicate is detected, the name of the original texture node is
written instead of the whole PNG data, decreasing significantly the
resulting file size.
