gzdoom-gles/src/v_palette.cpp

893 lines
21 KiB
C++
Raw Normal View History

/*
** v_palette.cpp
** Automatic colormap generation for "colored lights", etc.
**
**---------------------------------------------------------------------------
** Copyright 1998-2006 Randy Heit
** All rights reserved.
**
** Redistribution and use in source and binary forms, with or without
** modification, are permitted provided that the following conditions
** are met:
**
** 1. Redistributions of source code must retain the above copyright
** notice, this list of conditions and the following disclaimer.
** 2. Redistributions in binary form must reproduce the above copyright
** notice, this list of conditions and the following disclaimer in the
** documentation and/or other materials provided with the distribution.
** 3. The name of the author may not be used to endorse or promote products
** derived from this software without specific prior written permission.
**
** THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
** IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
** OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
** IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
** INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
** NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
** DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
** THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
** (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
** THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
**---------------------------------------------------------------------------
**
*/
#include <stddef.h>
#include <string.h>
#include <math.h>
#ifdef _WIN32
#include <io.h>
#else
#include <unistd.h>
#define O_BINARY 0
#endif
#include <fcntl.h>
#include "templates.h"
#include "v_video.h"
#include "i_system.h"
#include "r_main.h" // For lighting constants
#include "w_wad.h"
#include "i_video.h"
#include "c_dispatch.h"
#include "g_level.h"
#include "st_stuff.h"
#include "gi.h"
#include "x86.h"
#include "colormatcher.h"
#include "v_palette.h"
extern "C" {
FDynamicColormap NormalLight;
}
FPalette GPalette;
- Updated lempar.c to v1.31. - Added .txt files to the list of types (wad, zip, and pk3) that can be loaded without listing them after -file. - Fonts that are created by the ACS setfont command to wrap a texture now support animated textures. - FON2 fonts can now use their full palette for CR_UNTRANSLATED when drawn with the hardware 2D path instead of being restricted to the game palette. - Fixed: Toggling vid_vsync would reset the displayed fullscreen gamma to 1 on a Radeon 9000. - Added back the off-by-one palette handling, but in a much more limited scope than before. The skipped entry is assumed to always be at 248, and it is assumed that all Shader Model 1.4 cards suffer from this. That's because all SM1.4 cards are based on variants of the ATI R200 core, and the RV250 in a Radeon 9000 craps up like this. I see no reason to assume that other flavors of the R200 are any different. (Interesting note: With the Radeon 9000, D3DTADDRESS_CLAMP is an invalid address mode when using the debug Direct3D 9 runtime, but it works perfectly fine with the retail Direct3D 9 runtime.) (Insight: The R200 probably uses bytes for all its math inside pixel shaders. That would explain perfectly why I can't use constants greater than 1 with PS1.4 and why it can't do an exact mapping to every entry in the color palette. - Fixed: The software shaded drawer did not work for 2D, because its selected "color"map was replaced with the identitymap before being used. - Fixed: I cannot use Printf to output messages before the framebuffer was completely setup, meaning that Shader Model 1.4 cards could not change resolution. - I have decided to let remap palettes specify variable alpha values for their colors. D3DFB no longer forces them to 255. - Updated re2c to version 0.12.3. - Fixed: A_Wander used threshold as a timer, when it should have used reactiontime. - Fixed: A_CustomRailgun would not fire at all for actors without a target when the aim parameter was disabled. - Made the warp command work in multiplayer, again courtesy of Karate Chris. - Fixed: Trying to spawn a bot while not in a game made for a crashing time. (Patch courtesy of Karate Chris.) - Removed some floating point math from hu_scores.cpp that somebody's GCC gave warnings for (not mine, though). - Fixed: The SBarInfo drawbar command crashed if the sprite image was unavailable. - Fixed: FString::operator=(const char *) did not release its old buffer when being assigned to the null string. - The scanner no longer has an upper limit on the length of strings it accepts, though short strings will be faster than long ones. - Moved all the text scanning functions into a class. Mainly, this means that multiple script scanner states can be stored without being forced to do so recursively. I think I might be taking advantage of that in the near future. Possibly. Maybe. - Removed some potential buffer overflows from the decal parser. - Applied Blzut3's SBARINFO update #9: * Fixed: When using even length values in drawnumber it would cap to a 98 value instead of a 99 as intended. * The SBarInfo parser can now accept negatives for coordinates. This doesn't allow much right now, but later I plan to add better fullscreen hud support in which the negatives will be more useful. This also cleans up the source a bit since all calls for (x, y) coordinates are with the function getCoordinates(). - Added support for stencilling actors. - Added support for non-black colors specified with DTA_ColorOverlay to the software renderer. - Fixed: The inverse, gold, red, and green fixed colormaps each allocated space for 32 different colormaps, even though each only used the first one. - Added two new blending flags to make reverse subtract blending more useful: STYLEF_InvertSource and STYLEF_InvertOverlay. These invert the color that gets blended with the background, since that seems like a good idea for reverse subtraction. They also work with the other two blending operations. - Added subtract and reverse subtract blending operations to the renderer. Since the ERenderStyle enumeration was getting rather unwieldy, I converted it into a new FRenderStyle structure that lets each parameter of the blending equation be set separately. This simplified the set up for the blend quite a bit, and it means a number of new combinations are available by setting the parameters properly. SVN r710 (trunk)
2008-01-25 23:57:44 +00:00
BYTE InverseColormap[256];
BYTE GoldColormap[256];
// [BC] New Skulltag colormaps.
- Updated lempar.c to v1.31. - Added .txt files to the list of types (wad, zip, and pk3) that can be loaded without listing them after -file. - Fonts that are created by the ACS setfont command to wrap a texture now support animated textures. - FON2 fonts can now use their full palette for CR_UNTRANSLATED when drawn with the hardware 2D path instead of being restricted to the game palette. - Fixed: Toggling vid_vsync would reset the displayed fullscreen gamma to 1 on a Radeon 9000. - Added back the off-by-one palette handling, but in a much more limited scope than before. The skipped entry is assumed to always be at 248, and it is assumed that all Shader Model 1.4 cards suffer from this. That's because all SM1.4 cards are based on variants of the ATI R200 core, and the RV250 in a Radeon 9000 craps up like this. I see no reason to assume that other flavors of the R200 are any different. (Interesting note: With the Radeon 9000, D3DTADDRESS_CLAMP is an invalid address mode when using the debug Direct3D 9 runtime, but it works perfectly fine with the retail Direct3D 9 runtime.) (Insight: The R200 probably uses bytes for all its math inside pixel shaders. That would explain perfectly why I can't use constants greater than 1 with PS1.4 and why it can't do an exact mapping to every entry in the color palette. - Fixed: The software shaded drawer did not work for 2D, because its selected "color"map was replaced with the identitymap before being used. - Fixed: I cannot use Printf to output messages before the framebuffer was completely setup, meaning that Shader Model 1.4 cards could not change resolution. - I have decided to let remap palettes specify variable alpha values for their colors. D3DFB no longer forces them to 255. - Updated re2c to version 0.12.3. - Fixed: A_Wander used threshold as a timer, when it should have used reactiontime. - Fixed: A_CustomRailgun would not fire at all for actors without a target when the aim parameter was disabled. - Made the warp command work in multiplayer, again courtesy of Karate Chris. - Fixed: Trying to spawn a bot while not in a game made for a crashing time. (Patch courtesy of Karate Chris.) - Removed some floating point math from hu_scores.cpp that somebody's GCC gave warnings for (not mine, though). - Fixed: The SBarInfo drawbar command crashed if the sprite image was unavailable. - Fixed: FString::operator=(const char *) did not release its old buffer when being assigned to the null string. - The scanner no longer has an upper limit on the length of strings it accepts, though short strings will be faster than long ones. - Moved all the text scanning functions into a class. Mainly, this means that multiple script scanner states can be stored without being forced to do so recursively. I think I might be taking advantage of that in the near future. Possibly. Maybe. - Removed some potential buffer overflows from the decal parser. - Applied Blzut3's SBARINFO update #9: * Fixed: When using even length values in drawnumber it would cap to a 98 value instead of a 99 as intended. * The SBarInfo parser can now accept negatives for coordinates. This doesn't allow much right now, but later I plan to add better fullscreen hud support in which the negatives will be more useful. This also cleans up the source a bit since all calls for (x, y) coordinates are with the function getCoordinates(). - Added support for stencilling actors. - Added support for non-black colors specified with DTA_ColorOverlay to the software renderer. - Fixed: The inverse, gold, red, and green fixed colormaps each allocated space for 32 different colormaps, even though each only used the first one. - Added two new blending flags to make reverse subtract blending more useful: STYLEF_InvertSource and STYLEF_InvertOverlay. These invert the color that gets blended with the background, since that seems like a good idea for reverse subtraction. They also work with the other two blending operations. - Added subtract and reverse subtract blending operations to the renderer. Since the ERenderStyle enumeration was getting rather unwieldy, I converted it into a new FRenderStyle structure that lets each parameter of the blending equation be set separately. This simplified the set up for the blend quite a bit, and it means a number of new combinations are available by setting the parameters properly. SVN r710 (trunk)
2008-01-25 23:57:44 +00:00
BYTE RedColormap[256];
BYTE GreenColormap[256];
BYTE BlueColormap[256];
BYTE DesaturateColormap[31][256];
static void FreeSpecialLights();;
FColorMatcher ColorMatcher;
/* Current color blending values */
int BlendR, BlendG, BlendB, BlendA;
static int STACK_ARGS sortforremap (const void *a, const void *b);
static int STACK_ARGS sortforremap2 (const void *a, const void *b);
/**************************/
/* Gamma correction stuff */
/**************************/
BYTE newgamma[256];
CUSTOM_CVAR (Float, Gamma, 1.f, CVAR_ARCHIVE|CVAR_GLOBALCONFIG)
{
if (self == 0.f)
{ // Gamma values of 0 are illegal.
self = 1.f;
return;
}
if (screen != NULL)
{
screen->SetGamma (self);
}
}
/****************************/
/* Palette management stuff */
/****************************/
- Ported vlinetallasm4 to AMD64 assembly. Even with the increased number of registers AMD64 provides, this routine still needs to be written as self- modifying code for maximum performance. The additional registers do allow for further optimization over the x86 version by allowing all four pixels to be in flight at the same time. The end result is that AMD64 ASM is about 2.18 times faster than AMD64 C and about 1.06 times faster than x86 ASM. (For further comparison, AMD64 C and x86 C are practically the same for this function.) Should I port any more assembly to AMD64, mvlineasm4 is the most likely candidate, but it's not used enough at this point to bother. Also, this may or may not work with Linux at the moment, since it doesn't have the eh_handler metadata. Win64 is easier, since I just need to structure the function prologue and epilogue properly and use some assembler directives/macros to automatically generate the metadata. And that brings up another point: You need YASM to assemble the AMD64 code, because NASM doesn't support the Win64 metadata directives. - Added an SSE version of DoBlending. This is strictly C intrinsics. VC++ still throws around unneccessary register moves. GCC seems to be pretty close to optimal, requiring only about 2 cycles/color. They're both faster than my hand-written MMX routine, so I don't need to feel bad about not hand-optimizing this for x64 builds. - Removed an extra instruction from DoBlending_MMX, transposed two instructions, and unrolled it once, shaving off about 80 cycles from the time required to blend 256 palette entries. Why? Because I tried writing a C version of the routine using compiler intrinsics and was appalled by all the extra movq's VC++ added to the code. GCC was better, but still generated extra instructions. I only wanted a C version because I can't use inline assembly with VC++'s x64 compiler, and x64 assembly is a bit of a pain. (It's a pain because Linux and Windows have different calling conventions, and you need to maintain extra metadata for functions.) So, the assembly version stays and the C version stays out. - Removed all the pixel doubling r_detail modes, since the one platform they were intended to assist (486) actually sees very little benefit from them. - Rewrote CheckMMX in C and renamed it to CheckCPU. - Fixed: CPUID function 0x80000005 is specified to return detailed L1 cache only for AMD processors, so we must not use it on other architectures, or we end up overwriting the L1 cache line size with 0 or some other number we don't actually understand. SVN r1134 (trunk)
2008-08-09 03:13:43 +00:00
extern "C" BYTE BestColor_MMX (DWORD rgb, const DWORD *pal);
int BestColor (const uint32 *pal_in, int r, int g, int b, int first, int num)
{
- Ported vlinetallasm4 to AMD64 assembly. Even with the increased number of registers AMD64 provides, this routine still needs to be written as self- modifying code for maximum performance. The additional registers do allow for further optimization over the x86 version by allowing all four pixels to be in flight at the same time. The end result is that AMD64 ASM is about 2.18 times faster than AMD64 C and about 1.06 times faster than x86 ASM. (For further comparison, AMD64 C and x86 C are practically the same for this function.) Should I port any more assembly to AMD64, mvlineasm4 is the most likely candidate, but it's not used enough at this point to bother. Also, this may or may not work with Linux at the moment, since it doesn't have the eh_handler metadata. Win64 is easier, since I just need to structure the function prologue and epilogue properly and use some assembler directives/macros to automatically generate the metadata. And that brings up another point: You need YASM to assemble the AMD64 code, because NASM doesn't support the Win64 metadata directives. - Added an SSE version of DoBlending. This is strictly C intrinsics. VC++ still throws around unneccessary register moves. GCC seems to be pretty close to optimal, requiring only about 2 cycles/color. They're both faster than my hand-written MMX routine, so I don't need to feel bad about not hand-optimizing this for x64 builds. - Removed an extra instruction from DoBlending_MMX, transposed two instructions, and unrolled it once, shaving off about 80 cycles from the time required to blend 256 palette entries. Why? Because I tried writing a C version of the routine using compiler intrinsics and was appalled by all the extra movq's VC++ added to the code. GCC was better, but still generated extra instructions. I only wanted a C version because I can't use inline assembly with VC++'s x64 compiler, and x64 assembly is a bit of a pain. (It's a pain because Linux and Windows have different calling conventions, and you need to maintain extra metadata for functions.) So, the assembly version stays and the C version stays out. - Removed all the pixel doubling r_detail modes, since the one platform they were intended to assist (486) actually sees very little benefit from them. - Rewrote CheckMMX in C and renamed it to CheckCPU. - Fixed: CPUID function 0x80000005 is specified to return detailed L1 cache only for AMD processors, so we must not use it on other architectures, or we end up overwriting the L1 cache line size with 0 or some other number we don't actually understand. SVN r1134 (trunk)
2008-08-09 03:13:43 +00:00
#ifdef X86_ASM
if (CPU.bMMX)
{
int pre = 256 - num - first;
return BestColor_MMX (((first+pre)<<24)|(r<<16)|(g<<8)|b, pal_in-pre) - pre;
}
#endif
const PalEntry *pal = (const PalEntry *)pal_in;
int bestcolor = first;
int bestdist = 257*257+257*257+257*257;
for (int color = first; color < num; color++)
{
- Ported vlinetallasm4 to AMD64 assembly. Even with the increased number of registers AMD64 provides, this routine still needs to be written as self- modifying code for maximum performance. The additional registers do allow for further optimization over the x86 version by allowing all four pixels to be in flight at the same time. The end result is that AMD64 ASM is about 2.18 times faster than AMD64 C and about 1.06 times faster than x86 ASM. (For further comparison, AMD64 C and x86 C are practically the same for this function.) Should I port any more assembly to AMD64, mvlineasm4 is the most likely candidate, but it's not used enough at this point to bother. Also, this may or may not work with Linux at the moment, since it doesn't have the eh_handler metadata. Win64 is easier, since I just need to structure the function prologue and epilogue properly and use some assembler directives/macros to automatically generate the metadata. And that brings up another point: You need YASM to assemble the AMD64 code, because NASM doesn't support the Win64 metadata directives. - Added an SSE version of DoBlending. This is strictly C intrinsics. VC++ still throws around unneccessary register moves. GCC seems to be pretty close to optimal, requiring only about 2 cycles/color. They're both faster than my hand-written MMX routine, so I don't need to feel bad about not hand-optimizing this for x64 builds. - Removed an extra instruction from DoBlending_MMX, transposed two instructions, and unrolled it once, shaving off about 80 cycles from the time required to blend 256 palette entries. Why? Because I tried writing a C version of the routine using compiler intrinsics and was appalled by all the extra movq's VC++ added to the code. GCC was better, but still generated extra instructions. I only wanted a C version because I can't use inline assembly with VC++'s x64 compiler, and x64 assembly is a bit of a pain. (It's a pain because Linux and Windows have different calling conventions, and you need to maintain extra metadata for functions.) So, the assembly version stays and the C version stays out. - Removed all the pixel doubling r_detail modes, since the one platform they were intended to assist (486) actually sees very little benefit from them. - Rewrote CheckMMX in C and renamed it to CheckCPU. - Fixed: CPUID function 0x80000005 is specified to return detailed L1 cache only for AMD processors, so we must not use it on other architectures, or we end up overwriting the L1 cache line size with 0 or some other number we don't actually understand. SVN r1134 (trunk)
2008-08-09 03:13:43 +00:00
int x = r - pal[color].r;
int y = g - pal[color].g;
int z = b - pal[color].b;
int dist = x*x + y*y + z*z;
if (dist < bestdist)
{
if (dist == 0)
return color;
bestdist = dist;
bestcolor = color;
}
}
return bestcolor;
}
FPalette::FPalette ()
{
}
FPalette::FPalette (const BYTE *colors)
{
SetPalette (colors);
}
void FPalette::SetPalette (const BYTE *colors)
{
for (int i = 0; i < 256; i++, colors += 3)
{
BaseColors[i] = PalEntry (colors[0], colors[1], colors[2]);
Remap[i] = i;
}
// Find white and black from the original palette so that they can be
// used to make an educated guess of the translucency % for a BOOM
// translucency map.
WhiteIndex = BestColor ((DWORD *)BaseColors, 255, 255, 255);
BlackIndex = BestColor ((DWORD *)BaseColors, 0, 0, 0);
}
// In ZDoom's new texture system, color 0 is used as the transparent color.
// But color 0 is also a valid color for Doom engine graphics. What to do?
// Simple. The default palette for every game has at least one duplicate
// color, so find a duplicate pair of palette entries, make one of them a
// duplicate of color 0, and remap every graphic so that it uses that entry
// instead of entry 0.
void FPalette::MakeGoodRemap ()
{
PalEntry color0 = BaseColors[0];
int i;
// First try for an exact match of color 0. Only Hexen does not have one.
for (i = 1; i < 256; ++i)
{
if (BaseColors[i] == color0)
{
Remap[0] = i;
break;
}
}
// If there is no duplicate of color 0, find the first set of duplicate
// colors and make one of them a duplicate of color 0. In Hexen's PLAYPAL
// colors 209 and 229 are the only duplicates, but we cannot assume
// anything because the player might be using a custom PLAYPAL where those
// entries are not duplicates.
if (Remap[0] == 0)
{
PalEntry sortcopy[256];
for (i = 0; i < 256; ++i)
{
sortcopy[i] = BaseColors[i] | (i << 24);
}
qsort (sortcopy, 256, 4, sortforremap);
for (i = 255; i > 0; --i)
{
if ((sortcopy[i] & 0xFFFFFF) == (sortcopy[i-1] & 0xFFFFFF))
{
int new0 = sortcopy[i].a;
int dup = sortcopy[i-1].a;
if (new0 > dup)
{
// Make the lower-numbered entry a copy of color 0. (Just because.)
swap (new0, dup);
}
Remap[0] = new0;
Remap[new0] = dup;
BaseColors[new0] = color0;
break;
}
}
}
// If there were no duplicates, InitPalette() will remap color 0 to the
// closest matching color. Hopefully nobody will use a palette where all
// 256 entries are different. :-)
}
static int STACK_ARGS sortforremap (const void *a, const void *b)
{
return (*(const DWORD *)a & 0xFFFFFF) - (*(const DWORD *)b & 0xFFFFFF);
}
struct RemappingWork
{
DWORD Color;
BYTE Foreign; // 0 = local palette, 1 = foreign palette
BYTE PalEntry; // Entry # in the palette
BYTE Pad[2];
};
void FPalette::MakeRemap (const DWORD *colors, BYTE *remap, const BYTE *useful, int numcolors) const
{
RemappingWork workspace[255+256];
int i, j, k;
// Fill in workspace with the colors from the passed palette and this palette.
// By sorting this array, we can quickly find exact matches so that we can
// minimize the time spent calling BestColor for near matches.
for (i = 1; i < 256; ++i)
{
workspace[i-1].Color = DWORD(BaseColors[i]) & 0xFFFFFF;
workspace[i-1].Foreign = 0;
workspace[i-1].PalEntry = i;
}
for (i = k = 0, j = 255; i < numcolors; ++i)
{
if (useful == NULL || useful[i] != 0)
{
workspace[j].Color = colors[i] & 0xFFFFFF;
workspace[j].Foreign = 1;
workspace[j].PalEntry = i;
++j;
++k;
}
else
{
remap[i] = 0;
}
}
qsort (workspace, j, sizeof(RemappingWork), sortforremap2);
// Find exact matches
--j;
for (i = 0; i < j; ++i)
{
if (workspace[i].Foreign)
{
if (!workspace[i+1].Foreign && workspace[i].Color == workspace[i+1].Color)
{
remap[workspace[i].PalEntry] = workspace[i+1].PalEntry;
workspace[i].Foreign = 2;
++i;
--k;
}
}
}
// Find near matches
if (k > 0)
{
for (i = 0; i <= j; ++i)
{
if (workspace[i].Foreign == 1)
{
remap[workspace[i].PalEntry] = BestColor ((DWORD *)BaseColors,
RPART(workspace[i].Color), GPART(workspace[i].Color), BPART(workspace[i].Color),
1, 255);
}
}
}
}
static int STACK_ARGS sortforremap2 (const void *a, const void *b)
{
const RemappingWork *ap = (const RemappingWork *)a;
const RemappingWork *bp = (const RemappingWork *)b;
if (ap->Color == bp->Color)
{
return bp->Foreign - ap->Foreign;
}
else
{
return ap->Color - bp->Color;
}
}
static bool FixBuildPalette (BYTE *opal, int lump, bool blood)
{
if (Wads.LumpLength (lump) < 768)
{
return false;
}
FMemLump data = Wads.ReadLump (lump);
const BYTE *ipal = (const BYTE *)data.GetMem();
// Reverse the palette because BUILD used entry 255 as
// transparent, but we use 0 as transparent.
2006-05-16 02:50:18 +00:00
for (int c = 0; c < 768; c += 3)
{
if (!blood)
{
opal[c] = (ipal[765-c] << 2) | (ipal[765-c] >> 4);
opal[c+1] = (ipal[766-c] << 2) | (ipal[766-c] >> 4);
opal[c+2] = (ipal[767-c] << 2) | (ipal[767-c] >> 4);
}
else
{
opal[c] = ipal[765-c];
opal[c+1] = ipal[766-c];
opal[c+2] = ipal[767-c];
}
}
return true;
}
void InitPalette ()
{
BYTE pal[768];
BYTE *shade;
int c;
bool usingBuild = false;
int lump;
atterm (FreeSpecialLights);
if ((lump = Wads.CheckNumForFullName ("palette.dat")) >= 0 && Wads.LumpLength (lump) >= 768)
{
usingBuild = FixBuildPalette (pal, lump, false);
}
else if ((lump = Wads.CheckNumForFullName ("blood.pal")) >= 0 && Wads.LumpLength (lump) >= 768)
{
usingBuild = FixBuildPalette (pal, lump, true);
}
if (!usingBuild)
{
FWadLump palump = Wads.OpenLumpName ("PLAYPAL");
palump.Read (pal, 768);
}
GPalette.SetPalette (pal);
GPalette.MakeGoodRemap ();
ColorMatcher.SetPalette ((DWORD *)GPalette.BaseColors);
// The BUILD engine already has a transparent color, so it doesn't need any remapping.
if (!usingBuild)
{
if (GPalette.Remap[0] == 0)
{ // No duplicates, so settle for something close to color 0
GPalette.Remap[0] = BestColor ((DWORD *)GPalette.BaseColors,
GPalette.BaseColors[0].r, GPalette.BaseColors[0].g, GPalette.BaseColors[0].b, 1, 255);
}
}
NormalLight.Color = PalEntry (255, 255, 255);
NormalLight.Fade = 0;
// NormalLight.Maps is set by R_InitColormaps()
// build special maps (e.g. invulnerability)
May 3, 2006 (Changes by Graf Zahl) - Removed doom.x, heretic.x and strife.x from the SVN repository. These are generated files. - Fixed: A_PainDie has to check whether a valid target exists before calling IsFriend. - Fixed: FDecalLib::FindAnimator needs a signed counter to work properly. May 1, 2006 (Changes by Graf Zahl) - Added support for game specific pickup messages, if only to be able to define Raven's invulnerability item in DECORATE. - Removed A_TreeDeath because it is no longer used. - Fixed: When picking up a PowerupGiver for an active powerup the blend color and the duration were transferred to a temorary item and never took effect. They have to be trnasferred to the newly created powerup item before trying to give it to the player, not afterward. - Made the colormap of the InvulnerabilitySphere item specific. The base power class still needs to have its color adjusted per game though and since Raven's invulnerability item is used in both Hexen and Heretic it can't define its own colormap/blend. - Separated the invulnerability colormaps from the game being played and made them item specific. They can also be specified as regular blend colors in DECORATE now. - Converted a_hereticarmor.cpp and most of a_doomartifacts.cpp, a_hereticartifacts.cpp and a_heretickeys.cpp to DECORATE. - Changed the Soulsphere to be a real health item with the Dehacked modifications made in d_dehacked.cpp as for most other items which need to be adjusted. - Added IF_BIGPOWERUP flag to AInventory to expose the RESPAWN_SUPER dmflag to DECORATE. Also removed the now obsolete ShouldRespawn methods from AInvulnerabilitySphere and ABlurSphere. - Converted a_splashes.cpp to DECORATE. - Converted most of a_debris.cpp to DECORATE. SVN r73 (trunk)
2006-05-03 14:54:48 +00:00
int intensity;
- Updated lempar.c to v1.31. - Added .txt files to the list of types (wad, zip, and pk3) that can be loaded without listing them after -file. - Fonts that are created by the ACS setfont command to wrap a texture now support animated textures. - FON2 fonts can now use their full palette for CR_UNTRANSLATED when drawn with the hardware 2D path instead of being restricted to the game palette. - Fixed: Toggling vid_vsync would reset the displayed fullscreen gamma to 1 on a Radeon 9000. - Added back the off-by-one palette handling, but in a much more limited scope than before. The skipped entry is assumed to always be at 248, and it is assumed that all Shader Model 1.4 cards suffer from this. That's because all SM1.4 cards are based on variants of the ATI R200 core, and the RV250 in a Radeon 9000 craps up like this. I see no reason to assume that other flavors of the R200 are any different. (Interesting note: With the Radeon 9000, D3DTADDRESS_CLAMP is an invalid address mode when using the debug Direct3D 9 runtime, but it works perfectly fine with the retail Direct3D 9 runtime.) (Insight: The R200 probably uses bytes for all its math inside pixel shaders. That would explain perfectly why I can't use constants greater than 1 with PS1.4 and why it can't do an exact mapping to every entry in the color palette. - Fixed: The software shaded drawer did not work for 2D, because its selected "color"map was replaced with the identitymap before being used. - Fixed: I cannot use Printf to output messages before the framebuffer was completely setup, meaning that Shader Model 1.4 cards could not change resolution. - I have decided to let remap palettes specify variable alpha values for their colors. D3DFB no longer forces them to 255. - Updated re2c to version 0.12.3. - Fixed: A_Wander used threshold as a timer, when it should have used reactiontime. - Fixed: A_CustomRailgun would not fire at all for actors without a target when the aim parameter was disabled. - Made the warp command work in multiplayer, again courtesy of Karate Chris. - Fixed: Trying to spawn a bot while not in a game made for a crashing time. (Patch courtesy of Karate Chris.) - Removed some floating point math from hu_scores.cpp that somebody's GCC gave warnings for (not mine, though). - Fixed: The SBarInfo drawbar command crashed if the sprite image was unavailable. - Fixed: FString::operator=(const char *) did not release its old buffer when being assigned to the null string. - The scanner no longer has an upper limit on the length of strings it accepts, though short strings will be faster than long ones. - Moved all the text scanning functions into a class. Mainly, this means that multiple script scanner states can be stored without being forced to do so recursively. I think I might be taking advantage of that in the near future. Possibly. Maybe. - Removed some potential buffer overflows from the decal parser. - Applied Blzut3's SBARINFO update #9: * Fixed: When using even length values in drawnumber it would cap to a 98 value instead of a 99 as intended. * The SBarInfo parser can now accept negatives for coordinates. This doesn't allow much right now, but later I plan to add better fullscreen hud support in which the negatives will be more useful. This also cleans up the source a bit since all calls for (x, y) coordinates are with the function getCoordinates(). - Added support for stencilling actors. - Added support for non-black colors specified with DTA_ColorOverlay to the software renderer. - Fixed: The inverse, gold, red, and green fixed colormaps each allocated space for 32 different colormaps, even though each only used the first one. - Added two new blending flags to make reverse subtract blending more useful: STYLEF_InvertSource and STYLEF_InvertOverlay. These invert the color that gets blended with the background, since that seems like a good idea for reverse subtraction. They also work with the other two blending operations. - Added subtract and reverse subtract blending operations to the renderer. Since the ERenderStyle enumeration was getting rather unwieldy, I converted it into a new FRenderStyle structure that lets each parameter of the blending equation be set separately. This simplified the set up for the blend quite a bit, and it means a number of new combinations are available by setting the parameters properly. SVN r710 (trunk)
2008-01-25 23:57:44 +00:00
// Doom invulnerability is an inverted grayscale.
May 3, 2006 (Changes by Graf Zahl) - Removed doom.x, heretic.x and strife.x from the SVN repository. These are generated files. - Fixed: A_PainDie has to check whether a valid target exists before calling IsFriend. - Fixed: FDecalLib::FindAnimator needs a signed counter to work properly. May 1, 2006 (Changes by Graf Zahl) - Added support for game specific pickup messages, if only to be able to define Raven's invulnerability item in DECORATE. - Removed A_TreeDeath because it is no longer used. - Fixed: When picking up a PowerupGiver for an active powerup the blend color and the duration were transferred to a temorary item and never took effect. They have to be trnasferred to the newly created powerup item before trying to give it to the player, not afterward. - Made the colormap of the InvulnerabilitySphere item specific. The base power class still needs to have its color adjusted per game though and since Raven's invulnerability item is used in both Hexen and Heretic it can't define its own colormap/blend. - Separated the invulnerability colormaps from the game being played and made them item specific. They can also be specified as regular blend colors in DECORATE now. - Converted a_hereticarmor.cpp and most of a_doomartifacts.cpp, a_hereticartifacts.cpp and a_heretickeys.cpp to DECORATE. - Changed the Soulsphere to be a real health item with the Dehacked modifications made in d_dehacked.cpp as for most other items which need to be adjusted. - Added IF_BIGPOWERUP flag to AInventory to expose the RESPAWN_SUPER dmflag to DECORATE. Also removed the now obsolete ShouldRespawn methods from AInvulnerabilitySphere and ABlurSphere. - Converted a_splashes.cpp to DECORATE. - Converted most of a_debris.cpp to DECORATE. SVN r73 (trunk)
2006-05-03 14:54:48 +00:00
// Strife uses it when firing the Sigil
shade = InverseColormap;
for (c = 0; c < 256; c++)
{
intensity = (65535 -
(GPalette.BaseColors[c].r * 77 +
GPalette.BaseColors[c].g * 143 +
GPalette.BaseColors[c].b * 37)) >> 8;
- Updated lempar.c to v1.31. - Added .txt files to the list of types (wad, zip, and pk3) that can be loaded without listing them after -file. - Fonts that are created by the ACS setfont command to wrap a texture now support animated textures. - FON2 fonts can now use their full palette for CR_UNTRANSLATED when drawn with the hardware 2D path instead of being restricted to the game palette. - Fixed: Toggling vid_vsync would reset the displayed fullscreen gamma to 1 on a Radeon 9000. - Added back the off-by-one palette handling, but in a much more limited scope than before. The skipped entry is assumed to always be at 248, and it is assumed that all Shader Model 1.4 cards suffer from this. That's because all SM1.4 cards are based on variants of the ATI R200 core, and the RV250 in a Radeon 9000 craps up like this. I see no reason to assume that other flavors of the R200 are any different. (Interesting note: With the Radeon 9000, D3DTADDRESS_CLAMP is an invalid address mode when using the debug Direct3D 9 runtime, but it works perfectly fine with the retail Direct3D 9 runtime.) (Insight: The R200 probably uses bytes for all its math inside pixel shaders. That would explain perfectly why I can't use constants greater than 1 with PS1.4 and why it can't do an exact mapping to every entry in the color palette. - Fixed: The software shaded drawer did not work for 2D, because its selected "color"map was replaced with the identitymap before being used. - Fixed: I cannot use Printf to output messages before the framebuffer was completely setup, meaning that Shader Model 1.4 cards could not change resolution. - I have decided to let remap palettes specify variable alpha values for their colors. D3DFB no longer forces them to 255. - Updated re2c to version 0.12.3. - Fixed: A_Wander used threshold as a timer, when it should have used reactiontime. - Fixed: A_CustomRailgun would not fire at all for actors without a target when the aim parameter was disabled. - Made the warp command work in multiplayer, again courtesy of Karate Chris. - Fixed: Trying to spawn a bot while not in a game made for a crashing time. (Patch courtesy of Karate Chris.) - Removed some floating point math from hu_scores.cpp that somebody's GCC gave warnings for (not mine, though). - Fixed: The SBarInfo drawbar command crashed if the sprite image was unavailable. - Fixed: FString::operator=(const char *) did not release its old buffer when being assigned to the null string. - The scanner no longer has an upper limit on the length of strings it accepts, though short strings will be faster than long ones. - Moved all the text scanning functions into a class. Mainly, this means that multiple script scanner states can be stored without being forced to do so recursively. I think I might be taking advantage of that in the near future. Possibly. Maybe. - Removed some potential buffer overflows from the decal parser. - Applied Blzut3's SBARINFO update #9: * Fixed: When using even length values in drawnumber it would cap to a 98 value instead of a 99 as intended. * The SBarInfo parser can now accept negatives for coordinates. This doesn't allow much right now, but later I plan to add better fullscreen hud support in which the negatives will be more useful. This also cleans up the source a bit since all calls for (x, y) coordinates are with the function getCoordinates(). - Added support for stencilling actors. - Added support for non-black colors specified with DTA_ColorOverlay to the software renderer. - Fixed: The inverse, gold, red, and green fixed colormaps each allocated space for 32 different colormaps, even though each only used the first one. - Added two new blending flags to make reverse subtract blending more useful: STYLEF_InvertSource and STYLEF_InvertOverlay. These invert the color that gets blended with the background, since that seems like a good idea for reverse subtraction. They also work with the other two blending operations. - Added subtract and reverse subtract blending operations to the renderer. Since the ERenderStyle enumeration was getting rather unwieldy, I converted it into a new FRenderStyle structure that lets each parameter of the blending equation be set separately. This simplified the set up for the blend quite a bit, and it means a number of new combinations are available by setting the parameters properly. SVN r710 (trunk)
2008-01-25 23:57:44 +00:00
shade[c] = ColorMatcher.Pick (intensity, intensity, intensity);
}
- Updated lempar.c to v1.31. - Added .txt files to the list of types (wad, zip, and pk3) that can be loaded without listing them after -file. - Fonts that are created by the ACS setfont command to wrap a texture now support animated textures. - FON2 fonts can now use their full palette for CR_UNTRANSLATED when drawn with the hardware 2D path instead of being restricted to the game palette. - Fixed: Toggling vid_vsync would reset the displayed fullscreen gamma to 1 on a Radeon 9000. - Added back the off-by-one palette handling, but in a much more limited scope than before. The skipped entry is assumed to always be at 248, and it is assumed that all Shader Model 1.4 cards suffer from this. That's because all SM1.4 cards are based on variants of the ATI R200 core, and the RV250 in a Radeon 9000 craps up like this. I see no reason to assume that other flavors of the R200 are any different. (Interesting note: With the Radeon 9000, D3DTADDRESS_CLAMP is an invalid address mode when using the debug Direct3D 9 runtime, but it works perfectly fine with the retail Direct3D 9 runtime.) (Insight: The R200 probably uses bytes for all its math inside pixel shaders. That would explain perfectly why I can't use constants greater than 1 with PS1.4 and why it can't do an exact mapping to every entry in the color palette. - Fixed: The software shaded drawer did not work for 2D, because its selected "color"map was replaced with the identitymap before being used. - Fixed: I cannot use Printf to output messages before the framebuffer was completely setup, meaning that Shader Model 1.4 cards could not change resolution. - I have decided to let remap palettes specify variable alpha values for their colors. D3DFB no longer forces them to 255. - Updated re2c to version 0.12.3. - Fixed: A_Wander used threshold as a timer, when it should have used reactiontime. - Fixed: A_CustomRailgun would not fire at all for actors without a target when the aim parameter was disabled. - Made the warp command work in multiplayer, again courtesy of Karate Chris. - Fixed: Trying to spawn a bot while not in a game made for a crashing time. (Patch courtesy of Karate Chris.) - Removed some floating point math from hu_scores.cpp that somebody's GCC gave warnings for (not mine, though). - Fixed: The SBarInfo drawbar command crashed if the sprite image was unavailable. - Fixed: FString::operator=(const char *) did not release its old buffer when being assigned to the null string. - The scanner no longer has an upper limit on the length of strings it accepts, though short strings will be faster than long ones. - Moved all the text scanning functions into a class. Mainly, this means that multiple script scanner states can be stored without being forced to do so recursively. I think I might be taking advantage of that in the near future. Possibly. Maybe. - Removed some potential buffer overflows from the decal parser. - Applied Blzut3's SBARINFO update #9: * Fixed: When using even length values in drawnumber it would cap to a 98 value instead of a 99 as intended. * The SBarInfo parser can now accept negatives for coordinates. This doesn't allow much right now, but later I plan to add better fullscreen hud support in which the negatives will be more useful. This also cleans up the source a bit since all calls for (x, y) coordinates are with the function getCoordinates(). - Added support for stencilling actors. - Added support for non-black colors specified with DTA_ColorOverlay to the software renderer. - Fixed: The inverse, gold, red, and green fixed colormaps each allocated space for 32 different colormaps, even though each only used the first one. - Added two new blending flags to make reverse subtract blending more useful: STYLEF_InvertSource and STYLEF_InvertOverlay. These invert the color that gets blended with the background, since that seems like a good idea for reverse subtraction. They also work with the other two blending operations. - Added subtract and reverse subtract blending operations to the renderer. Since the ERenderStyle enumeration was getting rather unwieldy, I converted it into a new FRenderStyle structure that lets each parameter of the blending equation be set separately. This simplified the set up for the blend quite a bit, and it means a number of new combinations are available by setting the parameters properly. SVN r710 (trunk)
2008-01-25 23:57:44 +00:00
// Heretic invulnerability is a golden shade.
May 3, 2006 (Changes by Graf Zahl) - Removed doom.x, heretic.x and strife.x from the SVN repository. These are generated files. - Fixed: A_PainDie has to check whether a valid target exists before calling IsFriend. - Fixed: FDecalLib::FindAnimator needs a signed counter to work properly. May 1, 2006 (Changes by Graf Zahl) - Added support for game specific pickup messages, if only to be able to define Raven's invulnerability item in DECORATE. - Removed A_TreeDeath because it is no longer used. - Fixed: When picking up a PowerupGiver for an active powerup the blend color and the duration were transferred to a temorary item and never took effect. They have to be trnasferred to the newly created powerup item before trying to give it to the player, not afterward. - Made the colormap of the InvulnerabilitySphere item specific. The base power class still needs to have its color adjusted per game though and since Raven's invulnerability item is used in both Hexen and Heretic it can't define its own colormap/blend. - Separated the invulnerability colormaps from the game being played and made them item specific. They can also be specified as regular blend colors in DECORATE now. - Converted a_hereticarmor.cpp and most of a_doomartifacts.cpp, a_hereticartifacts.cpp and a_heretickeys.cpp to DECORATE. - Changed the Soulsphere to be a real health item with the Dehacked modifications made in d_dehacked.cpp as for most other items which need to be adjusted. - Added IF_BIGPOWERUP flag to AInventory to expose the RESPAWN_SUPER dmflag to DECORATE. Also removed the now obsolete ShouldRespawn methods from AInvulnerabilitySphere and ABlurSphere. - Converted a_splashes.cpp to DECORATE. - Converted most of a_debris.cpp to DECORATE. SVN r73 (trunk)
2006-05-03 14:54:48 +00:00
shade = GoldColormap;
for (c = 0; c < 256; c++)
{
intensity = GPalette.BaseColors[c].r * 77 +
GPalette.BaseColors[c].g * 143 +
GPalette.BaseColors[c].b * 37;
- Updated lempar.c to v1.31. - Added .txt files to the list of types (wad, zip, and pk3) that can be loaded without listing them after -file. - Fonts that are created by the ACS setfont command to wrap a texture now support animated textures. - FON2 fonts can now use their full palette for CR_UNTRANSLATED when drawn with the hardware 2D path instead of being restricted to the game palette. - Fixed: Toggling vid_vsync would reset the displayed fullscreen gamma to 1 on a Radeon 9000. - Added back the off-by-one palette handling, but in a much more limited scope than before. The skipped entry is assumed to always be at 248, and it is assumed that all Shader Model 1.4 cards suffer from this. That's because all SM1.4 cards are based on variants of the ATI R200 core, and the RV250 in a Radeon 9000 craps up like this. I see no reason to assume that other flavors of the R200 are any different. (Interesting note: With the Radeon 9000, D3DTADDRESS_CLAMP is an invalid address mode when using the debug Direct3D 9 runtime, but it works perfectly fine with the retail Direct3D 9 runtime.) (Insight: The R200 probably uses bytes for all its math inside pixel shaders. That would explain perfectly why I can't use constants greater than 1 with PS1.4 and why it can't do an exact mapping to every entry in the color palette. - Fixed: The software shaded drawer did not work for 2D, because its selected "color"map was replaced with the identitymap before being used. - Fixed: I cannot use Printf to output messages before the framebuffer was completely setup, meaning that Shader Model 1.4 cards could not change resolution. - I have decided to let remap palettes specify variable alpha values for their colors. D3DFB no longer forces them to 255. - Updated re2c to version 0.12.3. - Fixed: A_Wander used threshold as a timer, when it should have used reactiontime. - Fixed: A_CustomRailgun would not fire at all for actors without a target when the aim parameter was disabled. - Made the warp command work in multiplayer, again courtesy of Karate Chris. - Fixed: Trying to spawn a bot while not in a game made for a crashing time. (Patch courtesy of Karate Chris.) - Removed some floating point math from hu_scores.cpp that somebody's GCC gave warnings for (not mine, though). - Fixed: The SBarInfo drawbar command crashed if the sprite image was unavailable. - Fixed: FString::operator=(const char *) did not release its old buffer when being assigned to the null string. - The scanner no longer has an upper limit on the length of strings it accepts, though short strings will be faster than long ones. - Moved all the text scanning functions into a class. Mainly, this means that multiple script scanner states can be stored without being forced to do so recursively. I think I might be taking advantage of that in the near future. Possibly. Maybe. - Removed some potential buffer overflows from the decal parser. - Applied Blzut3's SBARINFO update #9: * Fixed: When using even length values in drawnumber it would cap to a 98 value instead of a 99 as intended. * The SBarInfo parser can now accept negatives for coordinates. This doesn't allow much right now, but later I plan to add better fullscreen hud support in which the negatives will be more useful. This also cleans up the source a bit since all calls for (x, y) coordinates are with the function getCoordinates(). - Added support for stencilling actors. - Added support for non-black colors specified with DTA_ColorOverlay to the software renderer. - Fixed: The inverse, gold, red, and green fixed colormaps each allocated space for 32 different colormaps, even though each only used the first one. - Added two new blending flags to make reverse subtract blending more useful: STYLEF_InvertSource and STYLEF_InvertOverlay. These invert the color that gets blended with the background, since that seems like a good idea for reverse subtraction. They also work with the other two blending operations. - Added subtract and reverse subtract blending operations to the renderer. Since the ERenderStyle enumeration was getting rather unwieldy, I converted it into a new FRenderStyle structure that lets each parameter of the blending equation be set separately. This simplified the set up for the blend quite a bit, and it means a number of new combinations are available by setting the parameters properly. SVN r710 (trunk)
2008-01-25 23:57:44 +00:00
shade[c] = ColorMatcher.Pick (
May 3, 2006 (Changes by Graf Zahl) - Removed doom.x, heretic.x and strife.x from the SVN repository. These are generated files. - Fixed: A_PainDie has to check whether a valid target exists before calling IsFriend. - Fixed: FDecalLib::FindAnimator needs a signed counter to work properly. May 1, 2006 (Changes by Graf Zahl) - Added support for game specific pickup messages, if only to be able to define Raven's invulnerability item in DECORATE. - Removed A_TreeDeath because it is no longer used. - Fixed: When picking up a PowerupGiver for an active powerup the blend color and the duration were transferred to a temorary item and never took effect. They have to be trnasferred to the newly created powerup item before trying to give it to the player, not afterward. - Made the colormap of the InvulnerabilitySphere item specific. The base power class still needs to have its color adjusted per game though and since Raven's invulnerability item is used in both Hexen and Heretic it can't define its own colormap/blend. - Separated the invulnerability colormaps from the game being played and made them item specific. They can also be specified as regular blend colors in DECORATE now. - Converted a_hereticarmor.cpp and most of a_doomartifacts.cpp, a_hereticartifacts.cpp and a_heretickeys.cpp to DECORATE. - Changed the Soulsphere to be a real health item with the Dehacked modifications made in d_dehacked.cpp as for most other items which need to be adjusted. - Added IF_BIGPOWERUP flag to AInventory to expose the RESPAWN_SUPER dmflag to DECORATE. Also removed the now obsolete ShouldRespawn methods from AInvulnerabilitySphere and ABlurSphere. - Converted a_splashes.cpp to DECORATE. - Converted most of a_debris.cpp to DECORATE. SVN r73 (trunk)
2006-05-03 14:54:48 +00:00
MIN (255, (intensity+intensity/2)>>8), intensity>>8, 0);
}
// [BC] Build the Doomsphere colormap. It is red!
shade = RedColormap;
for (c = 0; c < 256; c++)
{
intensity = ((GPalette.BaseColors[c].r * 77 +
GPalette.BaseColors[c].g * 143 +
GPalette.BaseColors[c].b * 37));
- Updated lempar.c to v1.31. - Added .txt files to the list of types (wad, zip, and pk3) that can be loaded without listing them after -file. - Fonts that are created by the ACS setfont command to wrap a texture now support animated textures. - FON2 fonts can now use their full palette for CR_UNTRANSLATED when drawn with the hardware 2D path instead of being restricted to the game palette. - Fixed: Toggling vid_vsync would reset the displayed fullscreen gamma to 1 on a Radeon 9000. - Added back the off-by-one palette handling, but in a much more limited scope than before. The skipped entry is assumed to always be at 248, and it is assumed that all Shader Model 1.4 cards suffer from this. That's because all SM1.4 cards are based on variants of the ATI R200 core, and the RV250 in a Radeon 9000 craps up like this. I see no reason to assume that other flavors of the R200 are any different. (Interesting note: With the Radeon 9000, D3DTADDRESS_CLAMP is an invalid address mode when using the debug Direct3D 9 runtime, but it works perfectly fine with the retail Direct3D 9 runtime.) (Insight: The R200 probably uses bytes for all its math inside pixel shaders. That would explain perfectly why I can't use constants greater than 1 with PS1.4 and why it can't do an exact mapping to every entry in the color palette. - Fixed: The software shaded drawer did not work for 2D, because its selected "color"map was replaced with the identitymap before being used. - Fixed: I cannot use Printf to output messages before the framebuffer was completely setup, meaning that Shader Model 1.4 cards could not change resolution. - I have decided to let remap palettes specify variable alpha values for their colors. D3DFB no longer forces them to 255. - Updated re2c to version 0.12.3. - Fixed: A_Wander used threshold as a timer, when it should have used reactiontime. - Fixed: A_CustomRailgun would not fire at all for actors without a target when the aim parameter was disabled. - Made the warp command work in multiplayer, again courtesy of Karate Chris. - Fixed: Trying to spawn a bot while not in a game made for a crashing time. (Patch courtesy of Karate Chris.) - Removed some floating point math from hu_scores.cpp that somebody's GCC gave warnings for (not mine, though). - Fixed: The SBarInfo drawbar command crashed if the sprite image was unavailable. - Fixed: FString::operator=(const char *) did not release its old buffer when being assigned to the null string. - The scanner no longer has an upper limit on the length of strings it accepts, though short strings will be faster than long ones. - Moved all the text scanning functions into a class. Mainly, this means that multiple script scanner states can be stored without being forced to do so recursively. I think I might be taking advantage of that in the near future. Possibly. Maybe. - Removed some potential buffer overflows from the decal parser. - Applied Blzut3's SBARINFO update #9: * Fixed: When using even length values in drawnumber it would cap to a 98 value instead of a 99 as intended. * The SBarInfo parser can now accept negatives for coordinates. This doesn't allow much right now, but later I plan to add better fullscreen hud support in which the negatives will be more useful. This also cleans up the source a bit since all calls for (x, y) coordinates are with the function getCoordinates(). - Added support for stencilling actors. - Added support for non-black colors specified with DTA_ColorOverlay to the software renderer. - Fixed: The inverse, gold, red, and green fixed colormaps each allocated space for 32 different colormaps, even though each only used the first one. - Added two new blending flags to make reverse subtract blending more useful: STYLEF_InvertSource and STYLEF_InvertOverlay. These invert the color that gets blended with the background, since that seems like a good idea for reverse subtraction. They also work with the other two blending operations. - Added subtract and reverse subtract blending operations to the renderer. Since the ERenderStyle enumeration was getting rather unwieldy, I converted it into a new FRenderStyle structure that lets each parameter of the blending equation be set separately. This simplified the set up for the blend quite a bit, and it means a number of new combinations are available by setting the parameters properly. SVN r710 (trunk)
2008-01-25 23:57:44 +00:00
shade[c] = ColorMatcher.Pick (
MIN( 255, ( intensity + ( intensity / 2 )) >> 8 ), 0, 0 );
}
// [BC] Build the Guardsphere colormap. It's a greenish-white kind of thing.
shade = GreenColormap;
for (c = 0; c < 256; c++)
{
intensity = GPalette.BaseColors[c].r * 77 +
GPalette.BaseColors[c].g * 143 +
GPalette.BaseColors[c].b * 37;
- Updated lempar.c to v1.31. - Added .txt files to the list of types (wad, zip, and pk3) that can be loaded without listing them after -file. - Fonts that are created by the ACS setfont command to wrap a texture now support animated textures. - FON2 fonts can now use their full palette for CR_UNTRANSLATED when drawn with the hardware 2D path instead of being restricted to the game palette. - Fixed: Toggling vid_vsync would reset the displayed fullscreen gamma to 1 on a Radeon 9000. - Added back the off-by-one palette handling, but in a much more limited scope than before. The skipped entry is assumed to always be at 248, and it is assumed that all Shader Model 1.4 cards suffer from this. That's because all SM1.4 cards are based on variants of the ATI R200 core, and the RV250 in a Radeon 9000 craps up like this. I see no reason to assume that other flavors of the R200 are any different. (Interesting note: With the Radeon 9000, D3DTADDRESS_CLAMP is an invalid address mode when using the debug Direct3D 9 runtime, but it works perfectly fine with the retail Direct3D 9 runtime.) (Insight: The R200 probably uses bytes for all its math inside pixel shaders. That would explain perfectly why I can't use constants greater than 1 with PS1.4 and why it can't do an exact mapping to every entry in the color palette. - Fixed: The software shaded drawer did not work for 2D, because its selected "color"map was replaced with the identitymap before being used. - Fixed: I cannot use Printf to output messages before the framebuffer was completely setup, meaning that Shader Model 1.4 cards could not change resolution. - I have decided to let remap palettes specify variable alpha values for their colors. D3DFB no longer forces them to 255. - Updated re2c to version 0.12.3. - Fixed: A_Wander used threshold as a timer, when it should have used reactiontime. - Fixed: A_CustomRailgun would not fire at all for actors without a target when the aim parameter was disabled. - Made the warp command work in multiplayer, again courtesy of Karate Chris. - Fixed: Trying to spawn a bot while not in a game made for a crashing time. (Patch courtesy of Karate Chris.) - Removed some floating point math from hu_scores.cpp that somebody's GCC gave warnings for (not mine, though). - Fixed: The SBarInfo drawbar command crashed if the sprite image was unavailable. - Fixed: FString::operator=(const char *) did not release its old buffer when being assigned to the null string. - The scanner no longer has an upper limit on the length of strings it accepts, though short strings will be faster than long ones. - Moved all the text scanning functions into a class. Mainly, this means that multiple script scanner states can be stored without being forced to do so recursively. I think I might be taking advantage of that in the near future. Possibly. Maybe. - Removed some potential buffer overflows from the decal parser. - Applied Blzut3's SBARINFO update #9: * Fixed: When using even length values in drawnumber it would cap to a 98 value instead of a 99 as intended. * The SBarInfo parser can now accept negatives for coordinates. This doesn't allow much right now, but later I plan to add better fullscreen hud support in which the negatives will be more useful. This also cleans up the source a bit since all calls for (x, y) coordinates are with the function getCoordinates(). - Added support for stencilling actors. - Added support for non-black colors specified with DTA_ColorOverlay to the software renderer. - Fixed: The inverse, gold, red, and green fixed colormaps each allocated space for 32 different colormaps, even though each only used the first one. - Added two new blending flags to make reverse subtract blending more useful: STYLEF_InvertSource and STYLEF_InvertOverlay. These invert the color that gets blended with the background, since that seems like a good idea for reverse subtraction. They also work with the other two blending operations. - Added subtract and reverse subtract blending operations to the renderer. Since the ERenderStyle enumeration was getting rather unwieldy, I converted it into a new FRenderStyle structure that lets each parameter of the blending equation be set separately. This simplified the set up for the blend quite a bit, and it means a number of new combinations are available by setting the parameters properly. SVN r710 (trunk)
2008-01-25 23:57:44 +00:00
shade[c] = ColorMatcher.Pick (
MIN( 255, ( intensity + ( intensity / 2 )) >> 8 ),
MIN( 255, ( intensity + ( intensity / 2 )) >> 8 ),
intensity>>8 );
}
// Build a blue colormap.
shade = BlueColormap;
for (c = 0; c < 256; c++)
{
intensity = ((GPalette.BaseColors[c].r * 77 +
GPalette.BaseColors[c].g * 143 +
GPalette.BaseColors[c].b * 37));
shade[c] = ColorMatcher.Pick (
MIN( 255, ( intensity + ( intensity / 2 )) >> 8 ), 0, 0 );
}
// desaturated colormaps
for(int m = 0; m < 31; m++)
{
shade = DesaturateColormap[m];
for (c = 0; c < 256; c++)
{
intensity = (GPalette.BaseColors[c].r * 77 +
GPalette.BaseColors[c].g * 143 +
GPalette.BaseColors[c].b * 37) / 255;
int r = (GPalette.BaseColors[c].r * (31-m) + intensity *m) / 31;
int g = (GPalette.BaseColors[c].g * (31-m) + intensity *m) / 31;
int b = (GPalette.BaseColors[c].b * (31-m) + intensity *m) / 31;
shade[c] = ColorMatcher.Pick (r, g, b);
}
}
}
- Ported vlinetallasm4 to AMD64 assembly. Even with the increased number of registers AMD64 provides, this routine still needs to be written as self- modifying code for maximum performance. The additional registers do allow for further optimization over the x86 version by allowing all four pixels to be in flight at the same time. The end result is that AMD64 ASM is about 2.18 times faster than AMD64 C and about 1.06 times faster than x86 ASM. (For further comparison, AMD64 C and x86 C are practically the same for this function.) Should I port any more assembly to AMD64, mvlineasm4 is the most likely candidate, but it's not used enough at this point to bother. Also, this may or may not work with Linux at the moment, since it doesn't have the eh_handler metadata. Win64 is easier, since I just need to structure the function prologue and epilogue properly and use some assembler directives/macros to automatically generate the metadata. And that brings up another point: You need YASM to assemble the AMD64 code, because NASM doesn't support the Win64 metadata directives. - Added an SSE version of DoBlending. This is strictly C intrinsics. VC++ still throws around unneccessary register moves. GCC seems to be pretty close to optimal, requiring only about 2 cycles/color. They're both faster than my hand-written MMX routine, so I don't need to feel bad about not hand-optimizing this for x64 builds. - Removed an extra instruction from DoBlending_MMX, transposed two instructions, and unrolled it once, shaving off about 80 cycles from the time required to blend 256 palette entries. Why? Because I tried writing a C version of the routine using compiler intrinsics and was appalled by all the extra movq's VC++ added to the code. GCC was better, but still generated extra instructions. I only wanted a C version because I can't use inline assembly with VC++'s x64 compiler, and x64 assembly is a bit of a pain. (It's a pain because Linux and Windows have different calling conventions, and you need to maintain extra metadata for functions.) So, the assembly version stays and the C version stays out. - Removed all the pixel doubling r_detail modes, since the one platform they were intended to assist (486) actually sees very little benefit from them. - Rewrote CheckMMX in C and renamed it to CheckCPU. - Fixed: CPUID function 0x80000005 is specified to return detailed L1 cache only for AMD processors, so we must not use it on other architectures, or we end up overwriting the L1 cache line size with 0 or some other number we don't actually understand. SVN r1134 (trunk)
2008-08-09 03:13:43 +00:00
extern "C" void STACK_ARGS DoBlending_MMX (const PalEntry *from, PalEntry *to, int count, int r, int g, int b, int a);
extern void DoBlending_SSE2 (const PalEntry *from, PalEntry *to, int count, int r, int g, int b, int a);
void DoBlending (const PalEntry *from, PalEntry *to, int count, int r, int g, int b, int a)
{
if (a == 0)
{
if (from != to)
{
memcpy (to, from, count * sizeof(DWORD));
}
}
else if (a == 256)
{
DWORD t = MAKERGB(r,g,b);
int i;
for (i = 0; i < count; i++)
{
to[i] = t;
}
}
- Ported vlinetallasm4 to AMD64 assembly. Even with the increased number of registers AMD64 provides, this routine still needs to be written as self- modifying code for maximum performance. The additional registers do allow for further optimization over the x86 version by allowing all four pixels to be in flight at the same time. The end result is that AMD64 ASM is about 2.18 times faster than AMD64 C and about 1.06 times faster than x86 ASM. (For further comparison, AMD64 C and x86 C are practically the same for this function.) Should I port any more assembly to AMD64, mvlineasm4 is the most likely candidate, but it's not used enough at this point to bother. Also, this may or may not work with Linux at the moment, since it doesn't have the eh_handler metadata. Win64 is easier, since I just need to structure the function prologue and epilogue properly and use some assembler directives/macros to automatically generate the metadata. And that brings up another point: You need YASM to assemble the AMD64 code, because NASM doesn't support the Win64 metadata directives. - Added an SSE version of DoBlending. This is strictly C intrinsics. VC++ still throws around unneccessary register moves. GCC seems to be pretty close to optimal, requiring only about 2 cycles/color. They're both faster than my hand-written MMX routine, so I don't need to feel bad about not hand-optimizing this for x64 builds. - Removed an extra instruction from DoBlending_MMX, transposed two instructions, and unrolled it once, shaving off about 80 cycles from the time required to blend 256 palette entries. Why? Because I tried writing a C version of the routine using compiler intrinsics and was appalled by all the extra movq's VC++ added to the code. GCC was better, but still generated extra instructions. I only wanted a C version because I can't use inline assembly with VC++'s x64 compiler, and x64 assembly is a bit of a pain. (It's a pain because Linux and Windows have different calling conventions, and you need to maintain extra metadata for functions.) So, the assembly version stays and the C version stays out. - Removed all the pixel doubling r_detail modes, since the one platform they were intended to assist (486) actually sees very little benefit from them. - Rewrote CheckMMX in C and renamed it to CheckCPU. - Fixed: CPUID function 0x80000005 is specified to return detailed L1 cache only for AMD processors, so we must not use it on other architectures, or we end up overwriting the L1 cache line size with 0 or some other number we don't actually understand. SVN r1134 (trunk)
2008-08-09 03:13:43 +00:00
else if (CPU.bSSE2)
{
- Ported vlinetallasm4 to AMD64 assembly. Even with the increased number of registers AMD64 provides, this routine still needs to be written as self- modifying code for maximum performance. The additional registers do allow for further optimization over the x86 version by allowing all four pixels to be in flight at the same time. The end result is that AMD64 ASM is about 2.18 times faster than AMD64 C and about 1.06 times faster than x86 ASM. (For further comparison, AMD64 C and x86 C are practically the same for this function.) Should I port any more assembly to AMD64, mvlineasm4 is the most likely candidate, but it's not used enough at this point to bother. Also, this may or may not work with Linux at the moment, since it doesn't have the eh_handler metadata. Win64 is easier, since I just need to structure the function prologue and epilogue properly and use some assembler directives/macros to automatically generate the metadata. And that brings up another point: You need YASM to assemble the AMD64 code, because NASM doesn't support the Win64 metadata directives. - Added an SSE version of DoBlending. This is strictly C intrinsics. VC++ still throws around unneccessary register moves. GCC seems to be pretty close to optimal, requiring only about 2 cycles/color. They're both faster than my hand-written MMX routine, so I don't need to feel bad about not hand-optimizing this for x64 builds. - Removed an extra instruction from DoBlending_MMX, transposed two instructions, and unrolled it once, shaving off about 80 cycles from the time required to blend 256 palette entries. Why? Because I tried writing a C version of the routine using compiler intrinsics and was appalled by all the extra movq's VC++ added to the code. GCC was better, but still generated extra instructions. I only wanted a C version because I can't use inline assembly with VC++'s x64 compiler, and x64 assembly is a bit of a pain. (It's a pain because Linux and Windows have different calling conventions, and you need to maintain extra metadata for functions.) So, the assembly version stays and the C version stays out. - Removed all the pixel doubling r_detail modes, since the one platform they were intended to assist (486) actually sees very little benefit from them. - Rewrote CheckMMX in C and renamed it to CheckCPU. - Fixed: CPUID function 0x80000005 is specified to return detailed L1 cache only for AMD processors, so we must not use it on other architectures, or we end up overwriting the L1 cache line size with 0 or some other number we don't actually understand. SVN r1134 (trunk)
2008-08-09 03:13:43 +00:00
if (count >= 4)
{
int not3count = count & ~3;
DoBlending_SSE2 (from, to, not3count, r, g, b, a);
count &= 3;
if (count <= 0)
{
return;
}
from += not3count;
to += not3count;
}
}
- Ported vlinetallasm4 to AMD64 assembly. Even with the increased number of registers AMD64 provides, this routine still needs to be written as self- modifying code for maximum performance. The additional registers do allow for further optimization over the x86 version by allowing all four pixels to be in flight at the same time. The end result is that AMD64 ASM is about 2.18 times faster than AMD64 C and about 1.06 times faster than x86 ASM. (For further comparison, AMD64 C and x86 C are practically the same for this function.) Should I port any more assembly to AMD64, mvlineasm4 is the most likely candidate, but it's not used enough at this point to bother. Also, this may or may not work with Linux at the moment, since it doesn't have the eh_handler metadata. Win64 is easier, since I just need to structure the function prologue and epilogue properly and use some assembler directives/macros to automatically generate the metadata. And that brings up another point: You need YASM to assemble the AMD64 code, because NASM doesn't support the Win64 metadata directives. - Added an SSE version of DoBlending. This is strictly C intrinsics. VC++ still throws around unneccessary register moves. GCC seems to be pretty close to optimal, requiring only about 2 cycles/color. They're both faster than my hand-written MMX routine, so I don't need to feel bad about not hand-optimizing this for x64 builds. - Removed an extra instruction from DoBlending_MMX, transposed two instructions, and unrolled it once, shaving off about 80 cycles from the time required to blend 256 palette entries. Why? Because I tried writing a C version of the routine using compiler intrinsics and was appalled by all the extra movq's VC++ added to the code. GCC was better, but still generated extra instructions. I only wanted a C version because I can't use inline assembly with VC++'s x64 compiler, and x64 assembly is a bit of a pain. (It's a pain because Linux and Windows have different calling conventions, and you need to maintain extra metadata for functions.) So, the assembly version stays and the C version stays out. - Removed all the pixel doubling r_detail modes, since the one platform they were intended to assist (486) actually sees very little benefit from them. - Rewrote CheckMMX in C and renamed it to CheckCPU. - Fixed: CPUID function 0x80000005 is specified to return detailed L1 cache only for AMD processors, so we must not use it on other architectures, or we end up overwriting the L1 cache line size with 0 or some other number we don't actually understand. SVN r1134 (trunk)
2008-08-09 03:13:43 +00:00
#ifdef X86_ASM
else if (CPU.bMMX)
{
- Ported vlinetallasm4 to AMD64 assembly. Even with the increased number of registers AMD64 provides, this routine still needs to be written as self- modifying code for maximum performance. The additional registers do allow for further optimization over the x86 version by allowing all four pixels to be in flight at the same time. The end result is that AMD64 ASM is about 2.18 times faster than AMD64 C and about 1.06 times faster than x86 ASM. (For further comparison, AMD64 C and x86 C are practically the same for this function.) Should I port any more assembly to AMD64, mvlineasm4 is the most likely candidate, but it's not used enough at this point to bother. Also, this may or may not work with Linux at the moment, since it doesn't have the eh_handler metadata. Win64 is easier, since I just need to structure the function prologue and epilogue properly and use some assembler directives/macros to automatically generate the metadata. And that brings up another point: You need YASM to assemble the AMD64 code, because NASM doesn't support the Win64 metadata directives. - Added an SSE version of DoBlending. This is strictly C intrinsics. VC++ still throws around unneccessary register moves. GCC seems to be pretty close to optimal, requiring only about 2 cycles/color. They're both faster than my hand-written MMX routine, so I don't need to feel bad about not hand-optimizing this for x64 builds. - Removed an extra instruction from DoBlending_MMX, transposed two instructions, and unrolled it once, shaving off about 80 cycles from the time required to blend 256 palette entries. Why? Because I tried writing a C version of the routine using compiler intrinsics and was appalled by all the extra movq's VC++ added to the code. GCC was better, but still generated extra instructions. I only wanted a C version because I can't use inline assembly with VC++'s x64 compiler, and x64 assembly is a bit of a pain. (It's a pain because Linux and Windows have different calling conventions, and you need to maintain extra metadata for functions.) So, the assembly version stays and the C version stays out. - Removed all the pixel doubling r_detail modes, since the one platform they were intended to assist (486) actually sees very little benefit from them. - Rewrote CheckMMX in C and renamed it to CheckCPU. - Fixed: CPUID function 0x80000005 is specified to return detailed L1 cache only for AMD processors, so we must not use it on other architectures, or we end up overwriting the L1 cache line size with 0 or some other number we don't actually understand. SVN r1134 (trunk)
2008-08-09 03:13:43 +00:00
if (count >= 4)
{
- Ported vlinetallasm4 to AMD64 assembly. Even with the increased number of registers AMD64 provides, this routine still needs to be written as self- modifying code for maximum performance. The additional registers do allow for further optimization over the x86 version by allowing all four pixels to be in flight at the same time. The end result is that AMD64 ASM is about 2.18 times faster than AMD64 C and about 1.06 times faster than x86 ASM. (For further comparison, AMD64 C and x86 C are practically the same for this function.) Should I port any more assembly to AMD64, mvlineasm4 is the most likely candidate, but it's not used enough at this point to bother. Also, this may or may not work with Linux at the moment, since it doesn't have the eh_handler metadata. Win64 is easier, since I just need to structure the function prologue and epilogue properly and use some assembler directives/macros to automatically generate the metadata. And that brings up another point: You need YASM to assemble the AMD64 code, because NASM doesn't support the Win64 metadata directives. - Added an SSE version of DoBlending. This is strictly C intrinsics. VC++ still throws around unneccessary register moves. GCC seems to be pretty close to optimal, requiring only about 2 cycles/color. They're both faster than my hand-written MMX routine, so I don't need to feel bad about not hand-optimizing this for x64 builds. - Removed an extra instruction from DoBlending_MMX, transposed two instructions, and unrolled it once, shaving off about 80 cycles from the time required to blend 256 palette entries. Why? Because I tried writing a C version of the routine using compiler intrinsics and was appalled by all the extra movq's VC++ added to the code. GCC was better, but still generated extra instructions. I only wanted a C version because I can't use inline assembly with VC++'s x64 compiler, and x64 assembly is a bit of a pain. (It's a pain because Linux and Windows have different calling conventions, and you need to maintain extra metadata for functions.) So, the assembly version stays and the C version stays out. - Removed all the pixel doubling r_detail modes, since the one platform they were intended to assist (486) actually sees very little benefit from them. - Rewrote CheckMMX in C and renamed it to CheckCPU. - Fixed: CPUID function 0x80000005 is specified to return detailed L1 cache only for AMD processors, so we must not use it on other architectures, or we end up overwriting the L1 cache line size with 0 or some other number we don't actually understand. SVN r1134 (trunk)
2008-08-09 03:13:43 +00:00
int not3count = count & ~3;
DoBlending_MMX (from, to, not3count, r, g, b, a);
count &= 3;
if (count <= 0)
{
return;
}
from += not3count;
to += not3count;
}
}
- Ported vlinetallasm4 to AMD64 assembly. Even with the increased number of registers AMD64 provides, this routine still needs to be written as self- modifying code for maximum performance. The additional registers do allow for further optimization over the x86 version by allowing all four pixels to be in flight at the same time. The end result is that AMD64 ASM is about 2.18 times faster than AMD64 C and about 1.06 times faster than x86 ASM. (For further comparison, AMD64 C and x86 C are practically the same for this function.) Should I port any more assembly to AMD64, mvlineasm4 is the most likely candidate, but it's not used enough at this point to bother. Also, this may or may not work with Linux at the moment, since it doesn't have the eh_handler metadata. Win64 is easier, since I just need to structure the function prologue and epilogue properly and use some assembler directives/macros to automatically generate the metadata. And that brings up another point: You need YASM to assemble the AMD64 code, because NASM doesn't support the Win64 metadata directives. - Added an SSE version of DoBlending. This is strictly C intrinsics. VC++ still throws around unneccessary register moves. GCC seems to be pretty close to optimal, requiring only about 2 cycles/color. They're both faster than my hand-written MMX routine, so I don't need to feel bad about not hand-optimizing this for x64 builds. - Removed an extra instruction from DoBlending_MMX, transposed two instructions, and unrolled it once, shaving off about 80 cycles from the time required to blend 256 palette entries. Why? Because I tried writing a C version of the routine using compiler intrinsics and was appalled by all the extra movq's VC++ added to the code. GCC was better, but still generated extra instructions. I only wanted a C version because I can't use inline assembly with VC++'s x64 compiler, and x64 assembly is a bit of a pain. (It's a pain because Linux and Windows have different calling conventions, and you need to maintain extra metadata for functions.) So, the assembly version stays and the C version stays out. - Removed all the pixel doubling r_detail modes, since the one platform they were intended to assist (486) actually sees very little benefit from them. - Rewrote CheckMMX in C and renamed it to CheckCPU. - Fixed: CPUID function 0x80000005 is specified to return detailed L1 cache only for AMD processors, so we must not use it on other architectures, or we end up overwriting the L1 cache line size with 0 or some other number we don't actually understand. SVN r1134 (trunk)
2008-08-09 03:13:43 +00:00
#endif
int i, ia;
ia = 256 - a;
r *= a;
g *= a;
b *= a;
- Ported vlinetallasm4 to AMD64 assembly. Even with the increased number of registers AMD64 provides, this routine still needs to be written as self- modifying code for maximum performance. The additional registers do allow for further optimization over the x86 version by allowing all four pixels to be in flight at the same time. The end result is that AMD64 ASM is about 2.18 times faster than AMD64 C and about 1.06 times faster than x86 ASM. (For further comparison, AMD64 C and x86 C are practically the same for this function.) Should I port any more assembly to AMD64, mvlineasm4 is the most likely candidate, but it's not used enough at this point to bother. Also, this may or may not work with Linux at the moment, since it doesn't have the eh_handler metadata. Win64 is easier, since I just need to structure the function prologue and epilogue properly and use some assembler directives/macros to automatically generate the metadata. And that brings up another point: You need YASM to assemble the AMD64 code, because NASM doesn't support the Win64 metadata directives. - Added an SSE version of DoBlending. This is strictly C intrinsics. VC++ still throws around unneccessary register moves. GCC seems to be pretty close to optimal, requiring only about 2 cycles/color. They're both faster than my hand-written MMX routine, so I don't need to feel bad about not hand-optimizing this for x64 builds. - Removed an extra instruction from DoBlending_MMX, transposed two instructions, and unrolled it once, shaving off about 80 cycles from the time required to blend 256 palette entries. Why? Because I tried writing a C version of the routine using compiler intrinsics and was appalled by all the extra movq's VC++ added to the code. GCC was better, but still generated extra instructions. I only wanted a C version because I can't use inline assembly with VC++'s x64 compiler, and x64 assembly is a bit of a pain. (It's a pain because Linux and Windows have different calling conventions, and you need to maintain extra metadata for functions.) So, the assembly version stays and the C version stays out. - Removed all the pixel doubling r_detail modes, since the one platform they were intended to assist (486) actually sees very little benefit from them. - Rewrote CheckMMX in C and renamed it to CheckCPU. - Fixed: CPUID function 0x80000005 is specified to return detailed L1 cache only for AMD processors, so we must not use it on other architectures, or we end up overwriting the L1 cache line size with 0 or some other number we don't actually understand. SVN r1134 (trunk)
2008-08-09 03:13:43 +00:00
for (i = count; i > 0; i--, to++, from++)
{
to->r = (r + from->r * ia) >> 8;
to->g = (g + from->g * ia) >> 8;
to->b = (b + from->b * ia) >> 8;
}
}
void V_SetBlend (int blendr, int blendg, int blendb, int blenda)
{
// Don't do anything if the new blend is the same as the old
if (((blenda|BlendA) == 0) ||
(blendr == BlendR &&
blendg == BlendG &&
blendb == BlendB &&
blenda == BlendA))
return;
V_ForceBlend (blendr, blendg, blendb, blenda);
}
void V_ForceBlend (int blendr, int blendg, int blendb, int blenda)
{
BlendR = blendr;
BlendG = blendg;
BlendB = blendb;
BlendA = blenda;
screen->SetFlash (PalEntry (BlendR, BlendG, BlendB), BlendA);
}
CCMD (testblend)
{
FString colorstring;
int color;
float amt;
if (argv.argc() < 3)
{
Printf ("testblend <color> <amount>\n");
}
else
{
if ( !(colorstring = V_GetColorStringByName (argv[1])).IsEmpty() )
{
color = V_GetColorFromString (NULL, colorstring);
}
else
{
color = V_GetColorFromString (NULL, argv[1]);
}
amt = (float)atof (argv[2]);
if (amt > 1.0f)
amt = 1.0f;
else if (amt < 0.0f)
amt = 0.0f;
BaseBlendR = RPART(color);
BaseBlendG = GPART(color);
BaseBlendB = BPART(color);
BaseBlendA = amt;
}
}
CCMD (testfade)
{
FString colorstring;
DWORD color;
if (argv.argc() < 2)
{
Printf ("testfade <color>\n");
}
else
{
if ( !(colorstring = V_GetColorStringByName (argv[1])).IsEmpty() )
{
color = V_GetColorFromString (NULL, colorstring);
}
else
{
color = V_GetColorFromString (NULL, argv[1]);
}
level.fadeto = color;
NormalLight.ChangeFade (color);
}
}
/****** Colorspace Conversion Functions ******/
// Code from http://www.cs.rit.edu/~yxv4997/t_convert.html
// r,g,b values are from 0 to 1
// h = [0,360], s = [0,1], v = [0,1]
// if s == 0, then h = -1 (undefined)
// Green Doom guy colors:
// RGB - 0: { .46 1 .429 } 7: { .254 .571 .206 } 15: { .0317 .0794 .0159 }
// HSV - 0: { 116.743 .571 1 } 7: { 112.110 .639 .571 } 15: { 105.071 .800 .0794 }
void RGBtoHSV (float r, float g, float b, float *h, float *s, float *v)
{
float min, max, delta, foo;
if (r == g && g == b)
{
*h = 0;
*s = 0;
*v = r;
return;
}
foo = r < g ? r : g;
min = (foo < b) ? foo : b;
foo = r > g ? r : g;
max = (foo > b) ? foo : b;
*v = max; // v
delta = max - min;
*s = delta / max; // s
if (r == max)
*h = (g - b) / delta; // between yellow & magenta
else if (g == max)
*h = 2 + (b - r) / delta; // between cyan & yellow
else
*h = 4 + (r - g) / delta; // between magenta & cyan
*h *= 60; // degrees
if (*h < 0)
*h += 360;
}
void HSVtoRGB (float *r, float *g, float *b, float h, float s, float v)
{
int i;
float f, p, q, t;
if (s == 0)
{ // achromatic (grey)
*r = *g = *b = v;
return;
}
h /= 60; // sector 0 to 5
i = (int)floor (h);
f = h - i; // factorial part of h
p = v * (1 - s);
q = v * (1 - s * f);
t = v * (1 - s * (1 - f));
switch (i)
{
case 0: *r = v; *g = t; *b = p; break;
case 1: *r = q; *g = v; *b = p; break;
case 2: *r = p; *g = v; *b = t; break;
case 3: *r = p; *g = q; *b = v; break;
case 4: *r = t; *g = p; *b = v; break;
default: *r = v; *g = p; *b = q; break;
}
}
/****** Colored Lighting Stuffs ******/
FDynamicColormap *GetSpecialLights (PalEntry color, PalEntry fade, int desaturate)
{
FDynamicColormap *colormap;
// If this colormap has already been created, just return it
for (colormap = &NormalLight; colormap != NULL; colormap = colormap->Next)
{
if (color == colormap->Color &&
fade == colormap->Fade &&
desaturate == colormap->Desaturate)
{
return colormap;
}
}
// Not found. Create it.
colormap = new FDynamicColormap;
colormap->Next = NormalLight.Next;
colormap->Color = color;
colormap->Fade = fade;
colormap->Desaturate = desaturate;
NormalLight.Next = colormap;
if (screen->UsesColormap())
{
colormap->Maps = new BYTE[NUMCOLORMAPS*256];
colormap->BuildLights ();
}
else colormap->Maps = NULL;
return colormap;
}
// Free all lights created with GetSpecialLights
static void FreeSpecialLights()
{
FDynamicColormap *colormap, *next;
for (colormap = NormalLight.Next; colormap != NULL; colormap = next)
{
next = colormap->Next;
delete[] colormap->Maps;
delete colormap;
}
}
// Builds NUMCOLORMAPS colormaps lit with the specified color
void FDynamicColormap::BuildLights ()
{
int l, c;
int lr, lg, lb, ld, ild;
PalEntry colors[256], basecolors[256];
BYTE *shade;
if (Maps == NULL)
return;
// Scale light to the range 0-256, so we can avoid
// dividing by 255 in the bottom loop.
lr = Color.r*256/255;
lg = Color.g*256/255;
lb = Color.b*256/255;
ld = Desaturate*256/255;
if (ld < 0) // No negative desaturations, please.
{
ld = -ld;
}
ild = 256-ld;
if (ld == 0)
{
memcpy (basecolors, GPalette.BaseColors, sizeof(basecolors));
}
else
{
// Desaturate the palette before lighting it.
for (c = 0; c < 256; c++)
{
int r = GPalette.BaseColors[c].r;
int g = GPalette.BaseColors[c].g;
int b = GPalette.BaseColors[c].b;
int intensity = ((r * 77 + g * 143 + b * 37) >> 8) * ld;
basecolors[c].r = (r*ild + intensity) >> 8;
basecolors[c].g = (g*ild + intensity) >> 8;
basecolors[c].b = (b*ild + intensity) >> 8;
basecolors[c].a = 0;
}
}
// build normal (but colored) light mappings
for (l = 0; l < NUMCOLORMAPS; l++)
{
DoBlending (basecolors, colors, 256,
Fade.r, Fade.g, Fade.b, l * (256 / NUMCOLORMAPS));
shade = Maps + 256*l;
if ((DWORD)Color == MAKERGB(255,255,255))
{ // White light, so we can just pick the colors directly
for (c = 0; c < 256; c++)
{
*shade++ = ColorMatcher.Pick (colors[c].r, colors[c].g, colors[c].b);
}
}
else
- Updated lempar.c to v1.31. - Added .txt files to the list of types (wad, zip, and pk3) that can be loaded without listing them after -file. - Fonts that are created by the ACS setfont command to wrap a texture now support animated textures. - FON2 fonts can now use their full palette for CR_UNTRANSLATED when drawn with the hardware 2D path instead of being restricted to the game palette. - Fixed: Toggling vid_vsync would reset the displayed fullscreen gamma to 1 on a Radeon 9000. - Added back the off-by-one palette handling, but in a much more limited scope than before. The skipped entry is assumed to always be at 248, and it is assumed that all Shader Model 1.4 cards suffer from this. That's because all SM1.4 cards are based on variants of the ATI R200 core, and the RV250 in a Radeon 9000 craps up like this. I see no reason to assume that other flavors of the R200 are any different. (Interesting note: With the Radeon 9000, D3DTADDRESS_CLAMP is an invalid address mode when using the debug Direct3D 9 runtime, but it works perfectly fine with the retail Direct3D 9 runtime.) (Insight: The R200 probably uses bytes for all its math inside pixel shaders. That would explain perfectly why I can't use constants greater than 1 with PS1.4 and why it can't do an exact mapping to every entry in the color palette. - Fixed: The software shaded drawer did not work for 2D, because its selected "color"map was replaced with the identitymap before being used. - Fixed: I cannot use Printf to output messages before the framebuffer was completely setup, meaning that Shader Model 1.4 cards could not change resolution. - I have decided to let remap palettes specify variable alpha values for their colors. D3DFB no longer forces them to 255. - Updated re2c to version 0.12.3. - Fixed: A_Wander used threshold as a timer, when it should have used reactiontime. - Fixed: A_CustomRailgun would not fire at all for actors without a target when the aim parameter was disabled. - Made the warp command work in multiplayer, again courtesy of Karate Chris. - Fixed: Trying to spawn a bot while not in a game made for a crashing time. (Patch courtesy of Karate Chris.) - Removed some floating point math from hu_scores.cpp that somebody's GCC gave warnings for (not mine, though). - Fixed: The SBarInfo drawbar command crashed if the sprite image was unavailable. - Fixed: FString::operator=(const char *) did not release its old buffer when being assigned to the null string. - The scanner no longer has an upper limit on the length of strings it accepts, though short strings will be faster than long ones. - Moved all the text scanning functions into a class. Mainly, this means that multiple script scanner states can be stored without being forced to do so recursively. I think I might be taking advantage of that in the near future. Possibly. Maybe. - Removed some potential buffer overflows from the decal parser. - Applied Blzut3's SBARINFO update #9: * Fixed: When using even length values in drawnumber it would cap to a 98 value instead of a 99 as intended. * The SBarInfo parser can now accept negatives for coordinates. This doesn't allow much right now, but later I plan to add better fullscreen hud support in which the negatives will be more useful. This also cleans up the source a bit since all calls for (x, y) coordinates are with the function getCoordinates(). - Added support for stencilling actors. - Added support for non-black colors specified with DTA_ColorOverlay to the software renderer. - Fixed: The inverse, gold, red, and green fixed colormaps each allocated space for 32 different colormaps, even though each only used the first one. - Added two new blending flags to make reverse subtract blending more useful: STYLEF_InvertSource and STYLEF_InvertOverlay. These invert the color that gets blended with the background, since that seems like a good idea for reverse subtraction. They also work with the other two blending operations. - Added subtract and reverse subtract blending operations to the renderer. Since the ERenderStyle enumeration was getting rather unwieldy, I converted it into a new FRenderStyle structure that lets each parameter of the blending equation be set separately. This simplified the set up for the blend quite a bit, and it means a number of new combinations are available by setting the parameters properly. SVN r710 (trunk)
2008-01-25 23:57:44 +00:00
{ // Colored light, so do the (slightly) slower thing
for (c = 0; c < 256; c++)
{
*shade++ = ColorMatcher.Pick (
(colors[c].r*lr)>>8,
(colors[c].g*lg)>>8,
(colors[c].b*lb)>>8);
}
}
}
}
void FDynamicColormap::ChangeColor (PalEntry lightcolor, int desaturate)
{
if (lightcolor != Color || desaturate != Desaturate)
{
Color = lightcolor;
// [BB] desaturate must be in [0,255]
if( desaturate > 255 )
desaturate = 255;
else if ( desaturate < 0 )
desaturate = 0;
Desaturate = desaturate;
if (Maps) BuildLights ();
}
}
void FDynamicColormap::ChangeFade (PalEntry fadecolor)
{
if (fadecolor != Fade)
{
Fade = fadecolor;
if (Maps) BuildLights ();
}
}
void FDynamicColormap::ChangeColorFade (PalEntry lightcolor, PalEntry fadecolor)
{
if (lightcolor != Color || fadecolor != Fade)
{
Color = lightcolor;
Fade = fadecolor;
if (Maps) BuildLights ();
}
}
void FDynamicColormap::RebuildAllLights()
{
if (screen->UsesColormap())
{
FDynamicColormap *cm;
for (cm = &NormalLight; cm != NULL; cm = cm->Next)
{
if (cm->Maps == NULL)
{
cm->Maps = new BYTE[NUMCOLORMAPS*256];
cm->BuildLights ();
}
}
}
}
CCMD (testcolor)
{
FString colorstring;
DWORD color;
int desaturate;
if (argv.argc() < 2)
{
Printf ("testcolor <color> [desaturation]\n");
}
else
{
if ( !(colorstring = V_GetColorStringByName (argv[1])).IsEmpty() )
{
color = V_GetColorFromString (NULL, colorstring);
}
else
{
color = V_GetColorFromString (NULL, argv[1]);
}
if (argv.argc() > 2)
{
desaturate = atoi (argv[2]);
}
else
{
desaturate = NormalLight.Desaturate;
}
NormalLight.ChangeColor (color, desaturate);
}
}