Replace almost all texture uses with calls to the new system.
Add some anistropic filtering to terrain textures.
Let Atlas load terrain texture previews partly-asynchronously by
polling.
Fix inefficient texture colour determination for minimap.
Remove unused global g_TerrainModified.
Change GUI texcoord computation to be less efficient but to cope with
dynamic texture changes.
Fix GUI renderer effects leaving bogus colour state.
This was SVN commit r8099.
this snowballed into a massive search+destroy of the hodgepodge of
mostly equivalent types we had in use (int, uint, unsigned, unsigned
int, i32, u32, ulong, uintN).
it is more efficient to use 64-bit types in 64-bit mode, so the
preferred default is size_t (for anything remotely resembling a size or
index). tile coordinates are ssize_t to allow more efficient conversion
to/from floating point. flags are int because we almost never need more
than 15 distinct bits, bit test/set is not slower and int is fastest to
type. finally, some data that is pretty much directly passed to OpenGL
is now typed accordingly.
after several hours, the code now requires fewer casts and less
guesswork.
other changes:
- unit and player IDs now have an "invalid id" constant in the
respective class to avoid casting and -1
- fix some endian/64-bit bugs in the map (un)packing. added a
convenience function to write/read a size_t.
- ia32: change CPUID interface to allow passing in ecx (required for
cache topology detection, which I need at work). remove some unneeded
functions from asm, replace with intrinsics where possible.
This was SVN commit r5942.
- The server will no longer send a new turn until the previous one has
been "acknowledged". (TODO: this may create lag between turns in its
current form; investigate and possibly allow executing two turns while
the third is being negotiated).
- Mutexes are now being used in NetServer and NetClient to ensure
commands go into the right batches.
- Commented out some orders in the GUI code that should not be there and
are not 100% working anyway (they were part of the follow/formation
system).
- Units that spawn or are created by scripts now have net-safe position
and orientation.
- Added a debug flag that can be turned on to print details about when
commands are received and executed (DEBUG_SYNCHRONIZATION). This is
especially useful if you diff the stdouts of two running games. There
should be no differences if everything is in synch.
This was SVN commit r5463.
. the massive renaming undertaking: camelCase functions -> PascalCase.
. add some cppdoc.
. minor additional renaming improvements: e.g. GetIsClosed -> IsClosed
. in entity code, replace constructs like "pvec = new vector; return
pvec; use *pvec; delete pvec" with a simple stack variable passed as
output parameter (avoid unnecessary dynamic allocs)
. timer: simpler handling of raw ticks vs normal timer (less #if)
This was SVN commit r5017.
GameView: pushed near clipping plane out a bit (to get more precision),
rounded far plane to power of two for no real reason.
This was SVN commit r4896.
* CMeshManager: Changed to check for .pmd and .dae files and convert
and cache as appropriate.
* CModelDef: Fixed misinterpreted doc comments in.
* lib:
* Fixed init/shutdown sequences to support multiple VFS-using tests
correctly.
* Fixed most reported memory leaks from the new leak-reporting test
system.
* Fixed error when trying to dump debug data about fonts after
unloading them.
* Added sphere dae/pmd data for tests.
* Added output buffering to DAE->PMD converter.
* Added precompiled COLLADA converter DLLs.
* Removed old redundant conversion script.
This was SVN commit r4709.
* Skinning is done in a way that works when there's more than one bone
influencing a vertex.
* PMDs now store vertexes in world-space instead of bind-space. (The
loader converts the old-version PMDs so they still work.)
* Moved SkinPoint, SkinNormal into CModelDef so it could use them when
loading the old PMDs.
* Made the FastNormals approach non-optional, so the inverse-transpose
bone matrices could be removed. Changed the explanation of why it's a
valid approach.
* Quaternion: Made GetInverse assume that the quaternions have unit
length (which they do when they're representing 3D rotations).
* lib: Added support for DDS files that aren't a multiple of 4x4 (most
useful for 1x1, 2x2, etc that are still powers of two).
* Actor Viewer: Added white terrain texture to the minimal test mod, so
shadows are visible. Changed default so walk/run animations don't move
the unit along the ground.
* Removed some redundant repetition in doc comments.
* Removed some unnecessary #includes.
This was SVN commit r4696.
- u_anim_name disambiguation
- fix implementation of finite (use our fpclassify instead of compiler's
routine)
- รค -> ae
- workaround for MAP_ANONYMOUS
- fix GLint in ogl.cpp
- add include for SIZE_MAX in string_s
- avoid PIC clobbered error and speed up rdtsc a bit
- add include for stat
This was SVN commit r4170.
This is a huge patch, including:
* add a LitRenderModifier abstract base class for RenderModifiers with
shadow+light
* add LitRenderModifiers for all types of models
* add STREAM_TEXGENTOUV1 to request generation of shadow map texcoords
for models
* create facilities to pass the texture matrix from the
RenderModifier (fragment stage) to the ModelRenderer (vertex stage)
* split ambient and diffuse terms of lighting until further down in the
pipeline; this is necessary since shadowed regions receive only
ambient light
* small improvement in how RenderPathVertexShader scales to a greater
number of vertex shaders
This was SVN commit r3690.
- explicitly split ambient and diffuse
- add renderer.shadowZBias to bias depths in the depth texture
and avoid flickering artifacts caused by Z fighting
- cantabrian_generated uses new light values so that shadows aren't
pitch
black
This was SVN commit r3508.
Add an "instancing" model renderer to improve rendering of
non-transparent,
unanimated models. This renderer is used when the vertexshader path is
rendering path is used.
This was SVN commit r3052.
transpose. This assumes orthogonal transformations (which we have
exclusively,
as far as I can tell), but is significantly faster, because calculating
the
inverse is slow.
This was SVN commit r3041.
Performance impact:
+ Player color rendered models are batched like normal models
- Transparent sorted models never use the vertex shader path
This was SVN commit r3009.