Specifically, it was in the "determine searchwall when aiming at floor or
ceiling" part. Now, if prsectors[]->ceil.plane (or ->floor.plane) is NULL,
we set the searchwall to the sector's firstwall and return.
git-svn-id: https://svn.eduke32.com/eduke32@2610 1a8010ca-5511-0410-912e-c29ae57300e0
This was broken ever since updatesprite() had been forked off drawsprite() and
would cause static lights to never hit static sprites until one or the other
changed, after which the light would always get culled against outdated sprite
geometry.
git-svn-id: https://svn.eduke32.com/eduke32@2536 1a8010ca-5511-0410-912e-c29ae57300e0
Since updatesprite() only happens once for static sprites, avoiding
light culling if we hit it in a shadow pass means that sprite will
never be lit as long as that happens.
git-svn-id: https://svn.eduke32.com/eduke32@2535 1a8010ca-5511-0410-912e-c29ae57300e0
With VBOs enabled, the virtual pointer to the geometry of frame 0 of a model
is always NULL. This was confusing the polymer code responsible for deciding
whether to enable smooth animations or not, which meant that any animation
that would loop to frame 0 would do it as if animation interpolation was
disabled, which looks very choppy.
git-svn-id: https://svn.eduke32.com/eduke32@2443 1a8010ca-5511-0410-912e-c29ae57300e0
The following functions have const qualifiers attached for the 'destination'
arguments: kdfwrite, dfwrite in cache1d.c and copybuf{byte,reverse,} in the
pragmas source or headers. A couple uses of the latter ones were replaced
with calls to standard library functions.
git-svn-id: https://svn.eduke32.com/eduke32@2361 1a8010ca-5511-0410-912e-c29ae57300e0
Internally, updateanimation() now takes a third parameter 'lpal'.
git-svn-id: https://svn.eduke32.com/eduke32@2277 1a8010ca-5511-0410-912e-c29ae57300e0
The cause was a combination of many factors. First, Polymer requires the
start-drawing position to be inside the sector to draw (deviations lead to
incorrect drawing). This was violated by the game, because it interpolated
the current and next tic position without updating the sectnum, leading to
inconsistencies especially when passing through narrow sectors. Polymer
worked around it by doing an updatesector() each frame, however this works
poorly for maps with overlapping geometry such as SoS or RoR.
The solution to this is to add a new engine function,
updatesector_onlynextwalls(), which searches the sectors (via nextsector)
in breadth-first fashion instead of from 0 to numsectors-1, like updatesector
does when it fails for the *immediate* nextwall neighbors.
Internally, the breadt-first search helpers are moved into the engine.
git-svn-id: https://svn.eduke32.com/eduke32@2256 1a8010ca-5511-0410-912e-c29ae57300e0
Also, for debugging builds, show this message from verbosity level 1 on.
git-svn-id: https://svn.eduke32.com/eduke32@2250 1a8010ca-5511-0410-912e-c29ae57300e0
the SE ones douple, triple, ... after each save. Maphack lights are reloaded
with polymer_resetlights() (though I've had them disappearing after load, I
couldn't reproduce it afterwards).
- in debugging builds, print a message when polymer_resetlight is called on a
nonexistent one. The problem is not so much the call itself (it's guarded),
but the fact that resources on the caller side may not have been freed if it
thinks that those lights do in fact exist. Right now, it's not the case though.
- add 'loaded map hack' messages after some other successful loadmaphack calls
git-svn-id: https://svn.eduke32.com/eduke32@2236 1a8010ca-5511-0410-912e-c29ae57300e0
fashion. Among other things, this makes the FOV widening when shrunk
work. User CON code should always check the initial viewingrange value
when attempting to do FOV effects like zoom, because it is not guaranteed
to be 65536 due to different screen aspect ratios etc.
git-svn-id: https://svn.eduke32.com/eduke32@2078 1a8010ca-5511-0410-912e-c29ae57300e0
polymer_addlight(). I have no idea whether it's the 'right' thing,
but it fixes this:
Conditional jump or move depends on uninitialised value(s)
at polymer_updatelights (polymer.c:5090)
by polymer_drawrooms (polymer.c:938)
. . .
Uninitialised value was created by a stack allocation
at loadmaphack (engine.c:9482)
Fix another uninitialized access when rendering jpegs.
git-svn-id: https://svn.eduke32.com/eduke32@2069 1a8010ca-5511-0410-912e-c29ae57300e0
Getting the texture handles without requiring invalidation means we can
handle ART-animating tilenums without impacting performance too much.
git-svn-id: https://svn.eduke32.com/eduke32@2058 1a8010ca-5511-0410-912e-c29ae57300e0
This won't actually do anything until the a debug bit is added when
creating the context, but that requires support for GLX_create_context
which SDL doesn't support. I'll add support for WGL_create_context to
winlayer in a bit as a stopgap.
Also updates our local copy of glext.h with a fresh one from the
Khronos registry.
git-svn-id: https://svn.eduke32.com/eduke32@2056 1a8010ca-5511-0410-912e-c29ae57300e0
We need to Y-invert the projection texcoords when sampling
from the lightmap.
git-svn-id: https://svn.eduke32.com/eduke32@2055 1a8010ca-5511-0410-912e-c29ae57300e0
Sprites would stick with their old material on global palette change.
This goes through all prsprites and clobbers their CRCs so that an
alternate skin / tint / highpal can be applied accordingly.
git-svn-id: https://svn.eduke32.com/eduke32@2053 1a8010ca-5511-0410-912e-c29ae57300e0
The winding order of faces is different if only flipping about X or Y,
and we were unconditionally culling CW faces.
git-svn-id: https://svn.eduke32.com/eduke32@2052 1a8010ca-5511-0410-912e-c29ae57300e0
First, if we're in Polymer, don't precache tinted tiles that have a highpal
for that certain pal. Second, don't precache tiles for palnum 251 (the last
non-reserved one, which is used as a crosshair pal). Assuming that there are
no other tints, this cuts the initial precache time and cache size on disk in
half.
git-svn-id: https://svn.eduke32.com/eduke32@2045 1a8010ca-5511-0410-912e-c29ae57300e0
Keep track of which side is getting culled, swap around mirrors, swap
around model drawing.
git-svn-id: https://svn.eduke32.com/eduke32@2032 1a8010ca-5511-0410-912e-c29ae57300e0
is present (like in E4L1). Also sneak in a trivial change in cache1d.c
git-svn-id: https://svn.eduke32.com/eduke32@2010 1a8010ca-5511-0410-912e-c29ae57300e0
In cases where performing occlusion queries for multiple red walls
leading to the same sector, we would sometimes leak a number of
query objects. This was apparently harmless on NVIDIA drivers, but
would leak tons of memory on ATI.
git-svn-id: https://svn.eduke32.com/eduke32@1993 1a8010ca-5511-0410-912e-c29ae57300e0
The code was using the Polymer-cached version of the sector cstat to
determine the translucency of its diffuse material, meaning it would
never get it right on the first try; that's what you get for having
completely different variables named 's' and 'sec'.
git-svn-id: https://svn.eduke32.com/eduke32@1991 1a8010ca-5511-0410-912e-c29ae57300e0
We normally enable backface culling always, but turn it off for
two-sided sprites; we need to make sure it's re-enabled after that
as sector masks are drawn after.
git-svn-id: https://svn.eduke32.com/eduke32@1990 1a8010ca-5511-0410-912e-c29ae57300e0
Some of the ROR code was operating under the assumption that drawasks
always correspond to one call of polymer_displarooms(), but that's not
necessarily the case; external view draws all sectors from the top
polymer_drawrooms(), which was causing crashes. Just ignore the sector
mask queues in external view mode, we already don't draw masked walls.
git-svn-id: https://svn.eduke32.com/eduke32@1982 1a8010ca-5511-0410-912e-c29ae57300e0
Check all the bunch siblings if we get a floor/ceiling hit and walk
them if their own planes hit.
git-svn-id: https://svn.eduke32.com/eduke32@1981 1a8010ca-5511-0410-912e-c29ae57300e0
When walking the sectors in displayrooms, detect floor/sector masks to
place them in a mask queue. The queues are allocated on the heap, but
a stack pointer is maintained by the displayrooms to allow it to play
nice with mirror recursion.
Right now the masks are drawn after all of drawmasks is finished; that
means you can see sprites and wall masks through sector masks, but not
the opposite. There is a slight attempt at blending far to near, but
nothing too conclusive yet; that means that in some cases some sector
masks won't be visible through other sector masks, but it should work
for trivial cases.
The next step would be to copy all the drawmasks logic into Polymer
and have it (attempt to) properly interleave sector masks where they
belong in the blending order.
git-svn-id: https://svn.eduke32.com/eduke32@1980 1a8010ca-5511-0410-912e-c29ae57300e0
Previously they would get counted both as 1-way walls for immediate
drawing and as mask walls for delayed drawing.
git-svn-id: https://svn.eduke32.com/eduke32@1979 1a8010ca-5511-0410-912e-c29ae57300e0
Assigns the right alpha value to floor/ceiling planes if their
translucency bits are set and make sure that changing the cstat now
triggers an update of the material for editing purposes.
This doesn't do anything by itself, but will be needed to correctly
draw floor/ceiling masks.
git-svn-id: https://svn.eduke32.com/eduke32@1977 1a8010ca-5511-0410-912e-c29ae57300e0
When walking the sectors, put more drawing work between queuing and
reaping the occlusion queries to potentially make them more efficient.
git-svn-id: https://svn.eduke32.com/eduke32@1976 1a8010ca-5511-0410-912e-c29ae57300e0
Enumerate the vertical siblings of each walked sector and check them
against the current frustum. No occlusion queries yet, but that should
correct any missing sectors until a flaw in that logic is discovered.
git-svn-id: https://svn.eduke32.com/eduke32@1975 1a8010ca-5511-0410-912e-c29ae57300e0
Reorganize polymer.c a bit to match the order of prototypes from the
header, move macros from the middle into the header.
git-svn-id: https://svn.eduke32.com/eduke32@1974 1a8010ca-5511-0410-912e-c29ae57300e0
- fix an earlier 'oops': make highpals load correctly again
- maphack lights shouldn't be double-loaded now
- slight tweak to the loading screen fadein/out timing
- two bound checks
- make some variables static in engine.c
git-svn-id: https://svn.eduke32.com/eduke32@1969 1a8010ca-5511-0410-912e-c29ae57300e0
This fixes a ridiculous interaction problem between non-black fog and
additive lighting passes. Plagman is stupid.
git-svn-id: https://svn.eduke32.com/eduke32@1965 1a8010ca-5511-0410-912e-c29ae57300e0
* New shade/visibility calculation code, which is activated with 'r_usenewshading' (on by default), and is closer to the classic look. Also tweak the FOGSCALE macro to have approximately the same fog distance with all renderers.
* Mapster32: END modifier to RShift. If it's pressed when RShift is released, sprites which are in grayed out sectors are also selected; Make changing shade affect all highlighted sprites in 3D mode (when aiming at one of them).
* some debug code to watch out for suspicious glGenTexture/glDeleteTextures calls, not active.
git-svn-id: https://svn.eduke32.com/eduke32@1943 1a8010ca-5511-0410-912e-c29ae57300e0
First, it is now possible to disable the memcache (which is there to cache the texcache) by setting the new cvar 'r_memcache' to 0. Do this if you're constrained on memory or getting crashes when e.g. doing vidrestart often. Also, the memcache will disable itself (and free its storage, if it's there) the first time it fails to allocate.
Fix a strcpy with identical arguments in game.c and a couple of uninitialized mem accesses related to Polymer lights.
git-svn-id: https://svn.eduke32.com/eduke32@1941 1a8010ca-5511-0410-912e-c29ae57300e0
Engine stuff:
* Add 'r_shadescale_unbounded' cvar. When set to 0, OpenGL renderers should never
draw completely black objects (currently only implemented for Polymost)
Mapster32:
* Add 'r_shadescale' to config
* In 3D mode, make SPACE behave the same as holding down a mose button: the
currently pointed-at object is locked. Required some modification of a.m32
to play well (i.e. not reset SPACE). This is useful by itself but more so
in conjunction with the next point
* make Alt behave as a modifier with PGUP/PGDN: when aiming at a 2-sided wall,
move the other side's sector's ceiling or floor (only this is new).
* Auto-alignment of walls can be controlled in a finer grained fashion now:
When pressing '.', only the immediate neighbors get aligned. Use Ctrl-. for
the old behaviour.
* When inserting a point in 2D mode, auto-align the neighboring wall
git-svn-id: https://svn.eduke32.com/eduke32@1848 1a8010ca-5511-0410-912e-c29ae57300e0
