1
0
mirror of https://gitlab.com/OpenMW/openmw.git synced 2025-01-30 03:32:36 +00:00

26 Commits

Author SHA1 Message Date
Bret Curtis
d1fb854521 move most of the files from esm to esm3, keep common code in esm; this is make space for a future with esm4
esm typo

esm typo
2022-01-23 17:04:48 +01:00
wareya
20cbf941fb re-introduce short circuiting, but only under certain circumstances 2022-01-16 17:58:04 -05:00
wareya
49d2daee6a Movement solver tweaks
1) As much as I dislike it, upping the collision margin from 0.1 to 0.2
fixes bugs, particularly involving walking into upwards-slanted walls.

2) There were still some problems involving acute crevices/seams; they
were using the adjusted instead of unadjusted normal, and also they need
to bypass the don't-slide-upwards check to prevent (see #6379)

3) The move-away-from-what-we-just-hit code needs to run always, not
just on non-initial iterations. No idea why I did it this way before.

4) Force bullet to give actor boxes a tiny collision margin of 0.001
instead of the default 0.04. I can't tell whether this is actually
working or not, but it should reduce unexplained weirdness.

5) A piece of code that was meant to prevent bugs by short-circuiting
the movement solver if its direction changed more than 180 degrees
actually caused problems instead of preventing them, so I deleted it.
2021-11-05 14:56:55 -04:00
psi29a
c9f3c27dc7 Merge branch 'phys_perf_test' into 'master'
Help bullet optimize collisions with very complex collision meshes by making small collision tests if possible

See merge request OpenMW/openmw!1317
2021-11-03 11:49:44 +00:00
wareya
8c0102e1ee Optimize short-trace test to only be done when particularly helpful 2021-10-29 15:11:08 -04:00
fredzio
3750eb9cd8 Move Projectile simulation to the background thread. 2021-10-27 22:17:06 +02:00
Frederic Chardon
5009b66ef5 Use std::variant in the physics simulation for the different types of objects. For now only support only for actors. 2021-10-27 22:02:46 +02:00
elsid
6986feb81b
Initialize navigator max climb and max slope settings in makeSettingsFromSettingsManager
To avoid having multiple places to initialize them when they will be required
by multiple binaries.
2021-10-11 18:50:26 +02:00
fredzio
b4dd9e6b4d Avoid division by zero in movementsolver when an actor is immobile and
in a storm
2021-08-08 17:19:18 +02:00
fredzio
0c5cf6ec19 Store the btCollisionObject* of the object we're standing on instead of
MWWorld::Ptr:
- they are equivalent
- btCollisionObject* is readily available from the simulation, it saves
  a call to a mutex
- btCollisionObject* is smaller
2021-08-07 13:38:24 +02:00
fredzio
f68273c3c0 Remove Actor* from ActorFrameData 2021-08-07 13:38:24 +02:00
fredzio
b04c958410 Modify the way swimming is handled:
- compute the swimming state instead of storing it, it changes as part of the simulation and was not updated, so it was wrong anyway.
- store the swim level in ActorFrameData, it is constant per Actor so no need to compute it inside the simulation
2021-08-07 13:38:24 +02:00
fredzio
6e51a9a512 Simplify a bit the solver 2021-08-07 13:38:24 +02:00
fredzio
51514e44cc Handle jump as part of the simulation preparation (inside of
PhysicsSystem) instead of inside the simulaiton.
For mechanics, we don't care how the jump is handled, just that it will be.
2021-08-07 13:38:24 +02:00
fredzio
1bfaf353be Explicitely store all the potential states an Actor can have into the
ActActorFrameData structure. It makes it easier to reason about the
simulation (and hopefully simplify it).
Remove atomics from Actor class as a side effect.

Rename mFloatToSurface to mInert to make is explicit what it represent, not what it is used for
Store the Actor rotation (1 Vec2) instead of the whole ESM::Position (2 Vec3)
2021-08-07 13:38:24 +02:00
fredzio
9472728fa4 Do not generate data for immobile actors instead of early out from the solver 2021-08-07 13:38:24 +02:00
fredzio
4fa0972b2d Tone down actor's skip simulation flag to an optional skip collision
detection flag.
2021-05-01 14:22:30 +02:00
fredzio
ccd3cbc69a Use saved actor position instead of reading again RefData in unstuck. It
is a race condition to do so.
2021-03-26 23:49:31 +01:00
wareya
40265bf118 make unstucking slightly smarter (can turn itself off, also acts like flat ground) 2021-03-20 21:14:56 -04:00
fredzio
b78820de55 Ignore projectiles inside of MovementSolver::unstuck. It is normal for actors to be inside of a
projectile collision shape.
A side effect of moving actors outside of projectile collision shape is that if both the actor and the projectile are
near a wall, the actor could get moved outside of the world.
2021-02-07 18:32:03 +01:00
Andrei Kortunov
6a12c2b58b Fix GCC build warnings 2020-12-28 12:06:41 +04:00
wareya
18ef32ca82 values for this higher than sGroundOffset cause jittering on some surface; use safe-seeming value slightly less than sGroundOffset 2020-12-27 22:16:11 +00:00
Alexei Dobrohotov
22476281da Fix paralyze for swimming actors 2020-12-22 08:03:51 +03:00
fredzio
8e084dea2e Don't cache Ptr, it can be updated while the simulation is running. 2020-12-18 22:22:37 +01:00
fredzio
91b3926a49 We need to update the collision world after each step.
Change order of traversal simulation step to make it rare enough to be parallelizable

Before:
for actor in actors:
    repeat numstep:
        solve(actor)
After:
repeat numstep:
    for actor in actors:
        solve(actor)

Introduce struct ActorFrameData to pack all data that is necessary for
the solver
2020-10-15 06:41:22 +02:00
Capostrophic
19010ec045 Separate MovementSolver 2020-03-31 00:38:34 +03:00