84870cf366
since tree_t only exists as a type that only allocates heap memory, we don't need to wrap it in a unique_ptr; this also allows the vectors to keep their memory on the next pass, which may improve performance for huge maps
...
removed tree_t parameter from functions that don't need it
pass reference to tree_t instead of ptr
use an enum instead of std::optional<bool> for the ternary value to store split type, since it's more explicit and obvious now what the three values do
2022-08-19 15:15:53 -04:00
69aad26097
move & document bevel boolean
...
don't report invalid winding more than once
ceil the extents, so we don't get weird values like "world extents calculated to 5082.29327318318"
2022-08-19 13:46:37 -04:00
057ae544c4
use references in places where we know it's non-null
2022-08-19 12:11:35 -04:00
39475e3213
use on_epsilon instead of 0.1, to match Q1
2022-08-19 09:44:11 -04:00
0c4888d5d1
remove leftover debug code
2022-08-19 09:42:11 -04:00
9596855c62
Merge branch 'brushbsp' of https://github.com/ericwa/ericw-tools into brushbsp
2022-08-19 07:15:51 -04:00
a5557bc87d
Revert "remove the Q1 bevel stuff"
...
This reverts commit f5f80479a3
2022-08-19 07:15:39 -04:00
6eab754ca3
qbsp: reset onnode at the beginning of BrushBSP
2022-08-18 22:26:33 -06:00
f57ecaf599
Q3 didn't need chop, we don't either!
2022-08-18 19:47:34 -04:00
424e3ee110
remove unused portal_t::face, unused
...
rename and re-type a few things to use twosided
2022-08-17 09:10:58 -04:00
9e764d8035
document some stuff
...
remove mapface_t::value and mapface_t::flags; they are duplicated by texinfo
add mapface_t::get_texinfo and face_t::get_texinfo so that you can fetch them easier
2022-08-17 07:30:31 -04:00
a2d6b63533
Merge branch 'brushbsp' of https://github.com/ericwa/ericw-tools into brushbsp
...
# Conflicts:
# qbsp/brush.cc
# qbsp/map.cc
2022-08-17 03:50:52 -04:00
f5f80479a3
remove the Q1 bevel stuff
2022-08-17 03:49:42 -04:00
fe43928802
fix build (unwanted static on TestExpandBrushes)
2022-08-17 01:18:11 -06:00
a2a7b987bc
revert last commit, q1 code is still producing better results with q1_rocks.map
2022-08-17 01:17:19 -06:00
688a033288
qbsp: switch back to qbsp3 brush bevels in q1 mode
...
now that the bug in the previous commit is fixed
2022-08-16 23:15:09 -06:00
6b1720e095
qbsp: fix bug from qbsp3's AddBrushBevels where edge bevels are wrongly rejected
...
fixes bad result in q1_hull_expansion_lip.map
2022-08-16 23:11:43 -06:00
95c728a5b6
fix legacy bevel bug
...
pushing up changes for Q1 hulls just in case we ever want these back
2022-08-16 13:52:30 -04:00
0c90a33f28
enable Q1-style hull expansion; these expanded brushes seem to look better. hopefully we can figure out a way to backport & optimize this code for mapbrush_t, but for now it's only used in Q1 hull expansions
2022-08-16 05:33:38 -04:00
f62136e04d
qbsp: wip fixing -expand feature
2022-08-16 01:36:09 -06:00
83cf63e77f
revert using detail on hull1/2 for now as it's breaking tests
2022-08-15 21:17:17 -06:00
92fdff1222
FreeTreePortals call missing
...
allow decompiling hulls (not hooked up to console yet)
2022-08-15 12:23:09 -04:00
027c45d608
Merge branch 'brushbsp' of https://github.com/ericwa/ericw-tools into brushbsp
2022-08-15 06:13:38 -04:00
f7b8f85ece
use an optional<uint8_t> to store hull number, which gets rid of special -1 collision hull number.
...
fixed no-hull always chopping even with chop off
2022-08-15 06:13:30 -04:00
20f5d73a3c
always use inside filling for auto
...
allow details in hulls
use std::optional for an optional-esque value
2022-08-15 04:53:36 -04:00
97665d15a9
fix slight mistype in Brush_LoadEntity
...
make ~settings_container virtual so light_t can be freed properly
reset visible to false on all brush sides
2022-08-15 03:37:38 -04:00
c2be045f65
qbsp: add WriteDebugTreePortalFile
2022-08-15 00:34:18 -06:00
854e1f48f5
qbsp: rewrite/fix hull and bmodel logging with new -loghulls and -logbmodels flags
2022-08-14 12:40:42 -06:00
2fe8ef2182
Merge branch 'brushbsp' of https://github.com/ericwa/ericw-tools into brushbsp
2022-08-14 07:33:49 -04:00
781a772d8a
move immediately instead of splicing after construction (probably compiles to the same code but whatever)
...
don't add all of the `result_portals_onnode` one at a time; just directly copy-elision the result.
2022-08-14 07:33:41 -04:00
e7adc1108a
qbsp: reset more settings between test cases
2022-08-13 20:38:57 -06:00
40b04b9518
use alloca instead of thread-local static
2022-08-13 20:30:51 -04:00
da5b40ce73
since buildportal_t is small and being stored in an std::list (which already stores them in nodes that require allocation & non-moving locations) we don't need to wrap them in unique_ptr; this saves a malloc/free per buildportal_t
2022-08-13 08:24:13 -04:00
2059835a47
use a more optimized version of SplitBrush for CheckPlaneAgainstVolume that doesn't require allocating/freeing a lot of memory
2022-08-13 06:39:30 -04:00
6d66e6d5cf
group together # of clipped faces, to reduce console spam of Q1 maps
...
always allow midplane split for hulls
2022-08-12 21:38:41 -04:00
06a116670b
winding_t: make non-copyable
2022-08-12 02:06:23 -06:00
ea38e5eeb7
fix missing move
...
fix clip_back causing extra copies
2022-08-12 02:03:00 -04:00
bf162c7a86
allow explicit calling of percent_clock so it can be simplified a bit
...
add more timers in a few spots
by default, display percent timers but don't display elapsed for non-verbose stuff (bmodels); this is so it still shows that it's "working" instead of just showing blank for large maps, give user an idea on progress for certain stuff
merge all "side not found"s into a single line
time `FillOutside`
time `ProcessMapBrushes`
time `Brush_LoadEntity`
2022-08-12 01:30:04 -04:00
4bf2ab338c
allowing logging::percent to handle indeterminate values
...
fix logging::percent getting stomped in multi-threaded operations
add new stats facility to make creating lists of single-use stats easier to manage
fix SplitBrush storing a bunch of extra stats for ChopBrushes
2022-08-11 20:33:12 -04:00
f851fac655
speed up compilation - we don't need to make a copy of node->volume going into SplitBrush now
2022-08-11 02:58:49 -04:00
f17ce185f6
Merge branch 'brushbsp' of https://github.com/ericwa/ericw-tools into brushbsp
2022-08-11 00:27:36 -04:00
0db1e0fc01
add chop control to settings
...
only chop if desired or if hull is -1 or > 0 since those can be chopped without affecting visuals
farewell, shared_lock
fix brush contents issue
2022-08-11 00:27:28 -04:00
1f3bf6edbc
qbsp: log BrushBSP function name
2022-08-10 21:45:08 -06:00
129dedb268
reset onnode between BrushBSP passes
2022-08-10 23:28:16 -04:00
4b4a1738fd
more partial revert
2022-08-10 23:06:37 -04:00
02e80645e6
partial revert of 097b564cc6
2022-08-10 22:43:42 -04:00
eb7c65362c
fix typo
...
assume non-sourced sides are non-visible
2022-08-10 22:27:40 -04:00
6ab037b03a
implement ChopBrushes
2022-08-10 22:12:23 -04:00
097b564cc6
remove bspbrush_t::original; it is no longer required since mapbrush_t is what the output is being written to
...
use some explicit types (bspbrush_t::ptr for pointer, bspbrush_t::container for container of pointers, and bspbrush_t::make_ptr to create one) so that we can swap it out with different pointers easier later
swap out bspbrush_t::ptr from unique_ptr to shared_ptr; it's not important yet, but it does not increase compilation time and will allow us to pass things to functions like SplitBrush without destroying the original as long as we still have a reference to it somewhere (important for ChopBrushes)
add get_positive_plane to mapface_t
instead of creating a whole copy of the bspbrush_t list, BrushBSP now just creates copies of the shared_ptrs which are then filtered down the tree; the originals are never modified by SplitBrushList and friends, only split into new brushes, so this is safe from what I can tell
portals now directly reference the mapface_ts
original_brushes now directly reference the mapbrush_ts
early exit in `BrushBSP` if brush list is empty (moved to top)
2022-08-10 16:02:16 -04:00
faf9173f10
remove file_order; we can infer this from the pointers now
...
remove unnecessary early microbrush check, use microbrush option value on later one
2022-08-10 02:22:32 -04:00
Jonathan
Eric Wasylishen