You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
 
 
 
 
 

510 lines
15 KiB

/*==LICENSE==*
CyanWorlds.com Engine - MMOG client, server and tools
Copyright (C) 2011 Cyan Worlds, Inc.
This program is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>.
Additional permissions under GNU GPL version 3 section 7
If you modify this Program, or any covered work, by linking or
combining it with any of RAD Game Tools Bink SDK, Autodesk 3ds Max SDK,
NVIDIA PhysX SDK, Microsoft DirectX SDK, OpenSSL library, Independent
JPEG Group JPEG library, Microsoft Windows Media SDK, or Apple QuickTime SDK
(or a modified version of those libraries),
containing parts covered by the terms of the Bink SDK EULA, 3ds Max EULA,
PhysX SDK EULA, DirectX SDK EULA, OpenSSL and SSLeay licenses, IJG
JPEG Library README, Windows Media SDK EULA, or QuickTime SDK EULA, the
licensors of this Program grant you additional
permission to convey the resulting work. Corresponding Source for a
non-source form of such a combination shall include the source code for
the parts of OpenSSL and IJG JPEG Library used as well as that of the covered
work.
You can contact Cyan Worlds, Inc. by email legal@cyan.com
or by snail mail at:
Cyan Worlds, Inc.
14617 N Newport Hwy
Mead, WA 99021
*==LICENSE==*/
#include "HeadSpin.h"
#include "hsInterp.h"
#include "plTransform/hsAffineParts.h"
#include "hsColorRGBA.h"
#include "hsPoint2.h"
//
///////////////////////////////////////////////////////
// linear interpolation
///////////////////////////////////////////////////////
//
void hsInterp::LinInterp(float k1, float k2, float t, float* result)
{
*result = k1 + t * (k2 - k1);
}
void hsInterp::LinInterp(const hsScalarTriple* k1, const hsScalarTriple* k2, float t,
hsScalarTriple* result)
{
if (t==0.0)
*result = *k1;
else
if (t==1.0)
*result = *k2;
else
{
LinInterp(k1->fX, k2->fX, t, &result->fX);
LinInterp(k1->fY, k2->fY, t, &result->fY);
LinInterp(k1->fZ, k2->fZ, t, &result->fZ);
}
}
void hsInterp::LinInterp(const hsColorRGBA* k1, const hsColorRGBA* k2, float t,
hsColorRGBA* result, uint32_t flags)
{
if (t==0.0)
{
// copy
result->r = k1->r;
result->g = k1->g;
result->b = k1->b;
if (!(flags & kIgnoreAlpha))
result->a = k1->a;
return;
}
if (t==1.0)
{
result->r = k2->r;
result->g = k2->g;
result->b = k2->b;
if (!(flags & kIgnoreAlpha))
result->a = k2->a;
return;
}
LinInterp(k1->r, k2->r, t, &result->r);
LinInterp(k1->g, k2->g, t, &result->g);
LinInterp(k1->b, k2->b, t, &result->b);
if (!(flags & kIgnoreAlpha))
LinInterp(k1->a, k2->a, t, &result->a);
}
void hsInterp::LinInterp(const hsMatrix33* k1, const hsMatrix33* k2, float t,
hsMatrix33* result, uint32_t flags)
{
if (t==0.0)
{
// copy
result->fMap[0][0] = k1->fMap[0][0];
result->fMap[0][1] = k1->fMap[0][1];
result->fMap[0][2] = k1->fMap[0][2];
result->fMap[1][0] = k1->fMap[1][0];
result->fMap[1][1] = k1->fMap[1][1];
result->fMap[1][2] = k1->fMap[1][2];
if (!(flags & kIgnoreLastMatRow))
{
result->fMap[2][0] = k1->fMap[2][0];
result->fMap[2][1] = k1->fMap[2][1];
result->fMap[2][2] = k1->fMap[2][2];
}
return;
}
if (t==1.0)
{
// copy
result->fMap[0][0] = k2->fMap[0][0];
result->fMap[0][1] = k2->fMap[0][1];
result->fMap[0][2] = k2->fMap[0][2];
result->fMap[1][0] = k2->fMap[1][0];
result->fMap[1][1] = k2->fMap[1][1];
result->fMap[1][2] = k2->fMap[1][2];
if (!(flags & kIgnoreLastMatRow))
{
result->fMap[2][0] = k2->fMap[2][0];
result->fMap[2][1] = k2->fMap[2][1];
result->fMap[2][2] = k2->fMap[2][2];
}
return;
}
LinInterp(k1->fMap[0][0], k2->fMap[0][0], t, &result->fMap[0][0]);
LinInterp(k1->fMap[0][1], k2->fMap[0][1], t, &result->fMap[0][1]);
LinInterp(k1->fMap[0][2], k2->fMap[0][2], t, &result->fMap[0][2]);
LinInterp(k1->fMap[1][0], k2->fMap[1][0], t, &result->fMap[1][0]);
LinInterp(k1->fMap[1][1], k2->fMap[1][1], t, &result->fMap[1][1]);
LinInterp(k1->fMap[1][2], k2->fMap[1][2], t, &result->fMap[1][2]);
if (!(flags & kIgnoreLastMatRow))
{
LinInterp(k1->fMap[2][0], k2->fMap[2][0], t, &result->fMap[2][0]);
LinInterp(k1->fMap[2][1], k2->fMap[2][1], t, &result->fMap[2][1]);
LinInterp(k1->fMap[2][2], k2->fMap[2][2], t, &result->fMap[2][2]);
}
}
//
//
void hsInterp::LinInterp(const hsMatrix44* mat1, const hsMatrix44* mat2, float t,
hsMatrix44* out, uint32_t flags)
{
if (flags == 0)
{
if( 0 == t )
{
*out = *mat1;
return;
}
if( 1.f == t )
{
*out = *mat2;
return;
}
}
if( flags & kIgnorePartsScale )
{
if (!(flags & kIgnorePartsRot))
{
// interp rotation with quats
hsQuat q1, q2, qOut;
q1.SetFromMatrix(mat1);
q2.SetFromMatrix(mat2);
LinInterp(&q1, &q2, t, &qOut);
qOut.Normalize();
qOut.MakeMatrix(out);
}
else
out->Reset();
#if 1
hsAssert(mat2->fMap[3][0]==0 && mat2->fMap[3][1]==0 && mat2->fMap[3][2]==0 && mat2->fMap[3][3]==1,
"matrix prob?");
#else
// copy
for(int i=0; i<3; i++)
out->fMap[3][i] = mat2->fMap[3][i];
#endif
if (!(flags & kIgnorePartsPos))
{
// interp translation
hsPoint3 p1,p2,pOut;
mat1->GetTranslate(&p1);
mat2->GetTranslate(&p2);
LinInterp(&p1, &p2, t, &pOut);
out->SetTranslate(&pOut);
out->NotIdentity(); // in case no rot
}
}
else
{
// Complete decomp and parts interp
gemAffineParts gemParts1, gemParts2;
hsAffineParts parts1, parts2, partsOut;
decomp_affine(mat1->fMap, &gemParts1);
AP_SET(parts1, gemParts1);
decomp_affine(mat2->fMap, &gemParts2);
AP_SET(parts2, gemParts2);
LinInterp(&parts1, &parts2, t, &partsOut, flags); // flags will be parsed here
partsOut.ComposeMatrix(out);
}
}
void hsInterp::LinInterp(const hsQuat* k1, const hsQuat* k2, float t, hsQuat* result)
{
if (t==0.0)
*result = *k1;
else
if (t==1.0)
*result = *k2;
else
{
result->SetFromSlerp(*k1, *k2, t);
}
}
void hsInterp::LinInterp(const hsScaleValue* k1, const hsScaleValue* k2, float t,
hsScaleValue* result)
{
LinInterp(&k1->fS, &k2->fS, t, &result->fS); // Stretch rotation
LinInterp(&k1->fQ, &k2->fQ, t, &result->fQ); // Stretch factor
}
void hsInterp::LinInterp(const hsAffineParts* k1, const hsAffineParts* k2, float t,
hsAffineParts* result, uint32_t flags)
{
if (t==0.0)
{
// copy
if (!(flags & kIgnorePartsPos))
result->fT = k1->fT;
if (!(flags & kIgnorePartsRot))
result->fQ = k1->fQ;
if (!(flags & kIgnorePartsScale))
{
// same as preserveScale
result->fU = k1->fU;
result->fK = k1->fK;
}
result->fF = k1->fF;
return;
}
if (flags & kPreservePartsScale)
{
result->fU = k1->fU; // just copy scale from 1st key
result->fK = k1->fK;
}
if (t==1.0)
{
// copy
if (!(flags & kIgnorePartsPos))
result->fT = k2->fT;
if (!(flags & kIgnorePartsRot))
result->fQ = k2->fQ;
if (!(flags & (kIgnorePartsScale | kPreservePartsScale)))
{
result->fU = k2->fU;
result->fK = k2->fK;
}
result->fF = k2->fF;
return;
}
if(k1->fF!=k2->fF)
hsStatusMessageF("WARNING: Inequality in affine parts flip value.");
// hsAssert(k1->fF==k2->fF, "inequality in affine parts flip value");
if (!(flags & kIgnorePartsPos))
LinInterp(&k1->fT, &k2->fT, t, &result->fT); // Translation
if (!(flags & kIgnorePartsRot))
{
LinInterp(&k1->fQ, &k2->fQ, t, &result->fQ); // Essential rotation
}
if (!(flags & (kIgnorePartsScale | kPreservePartsScale)))
{
LinInterp(&k1->fU, &k2->fU, t, &result->fU); // Stretch rotation
LinInterp(&k1->fK, &k2->fK, t, &result->fK); // Stretch factor
}
#if 0
if (!(flags & kIgnorePartsDet))
LinInterp(k1->fF, k2->fF, t, &result->fF); // Flip rot var
#else
result->fF = k1->fF;
#endif
}
//
///////////////////////////////////////////////////////
// Key interpolation
///////////////////////////////////////////////////////
//
void hsInterp::BezScalarEval(const float value1, const float outTan,
const float value2, const float inTan,
const float t, const float tanScale, float *result)
{
#if 0
// If the tangents were what you'd expect them to be... Hermite splines, than this code
// would make sense. But no, Max likes to store them in a scaled form based on the
// time of each frame. If we ever optimize this further, we could do the scaling on export,
// but I need this to work right now before all the artists hate me too much.
const float t2 = t * t;
const float t3 = t2 * t;
const float term1 = 2 * t3 - 3 * t2;
*result = ((term1 + 1) * value1) +
(-term1 * value2) +
((t3 - 2 * t2 + 1) * outTan) +
((t3 - t2) * inTan);
#else
const float oneMinusT = (1.0f - t);
const float tSq = t * t;
const float oneMinusTSq = oneMinusT * oneMinusT;
*result = (oneMinusT * oneMinusTSq * value1) +
(3.f * t * oneMinusTSq * (value1 + outTan * tanScale)) +
(3.f * tSq * oneMinusT * (value2 + inTan * tanScale)) +
(tSq * t * value2);
#endif
}
void hsInterp::BezInterp(const hsBezPoint3Key* k1, const hsBezPoint3Key* k2, const float t, hsScalarTriple* result)
{
float scale = (k2->fFrame - k1->fFrame) * MAX_TICKS_PER_FRAME / 3.f;
BezScalarEval(k1->fValue.fX, k1->fOutTan.fX, k2->fValue.fX, k2->fInTan.fX, t, scale, &result->fX);
BezScalarEval(k1->fValue.fY, k1->fOutTan.fY, k2->fValue.fY, k2->fInTan.fY, t, scale, &result->fY);
BezScalarEval(k1->fValue.fZ, k1->fOutTan.fZ, k2->fValue.fZ, k2->fInTan.fZ, t, scale, &result->fZ);
}
void hsInterp::BezInterp(const hsBezScalarKey* k1, const hsBezScalarKey* k2, const float t, float* result)
{
float scale = (k2->fFrame - k1->fFrame) * MAX_TICKS_PER_FRAME / 3.f;
BezScalarEval(k1->fValue, k1->fOutTan, k2->fValue, k2->fInTan, t, scale, result);
}
void hsInterp::BezInterp(const hsBezScaleKey* k1, const hsBezScaleKey* k2, const float t, hsScaleValue* result)
{
float scale = (k2->fFrame - k1->fFrame) * MAX_TICKS_PER_FRAME / 3.f;
BezScalarEval(k1->fValue.fS.fX, k1->fOutTan.fX, k2->fValue.fS.fX, k2->fInTan.fX, t, scale, &result->fS.fX);
BezScalarEval(k1->fValue.fS.fY, k1->fOutTan.fY, k2->fValue.fS.fY, k2->fInTan.fY, t, scale, &result->fS.fY);
BezScalarEval(k1->fValue.fS.fZ, k1->fOutTan.fZ, k2->fValue.fS.fZ, k2->fInTan.fZ, t, scale, &result->fS.fZ);
// Slerp scale axis
LinInterp(&k1->fValue.fQ, &k2->fValue.fQ, t, &result->fQ);
}
//
// Get an element from an array of unknown type
//
static inline hsKeyFrame* GetKey(int32_t i, void *keys, int32_t size)
{
return (hsKeyFrame*) ((char*)keys + size * i);
}
//
// STATIC
// Given a list of keys, and a time, fills in the 2 boundary keys and
// a fraction (p=0-1) indicating where the time falls between them.
// Returns the index of the first key which can be passed in as a hint (lastKeyIdx)
// for the next search.
//
void hsInterp::GetBoundaryKeyFrames(float time, uint32_t numKeys, void *keys, uint32_t size,
hsKeyFrame **kF1, hsKeyFrame **kF2, uint32_t *lastKeyIdx, float *p, bool forwards)
{
hsAssert(numKeys>1, "Must have more than 1 keyframe");
int k1, k2;
float frame = time * MAX_FRAMES_PER_SEC;
// boundary case, past end
if (frame > GetKey(numKeys-1, keys, size)->fFrame)
{
k1=k2=numKeys-1;
(*kF2) = GetKey(k1, keys, size);
(*kF1) = (*kF2);
*p = 0.0;
goto ret;
}
hsKeyFrame *key1, *key2;
// boundary case, before start
if (frame < (key1=GetKey(0, keys, size))->fFrame)
{
k1=k2=0;
(*kF1) = GetKey(k1, keys, size);
(*kF2) = (*kF1);
*p = 0.0;
goto ret;
}
// prime loop
int i;
i = 1;
if (*lastKeyIdx > 0 && *lastKeyIdx < numKeys - 1)
{
// new starting point for search
if (forwards)
key1 = GetKey(*lastKeyIdx, keys, size);
else
key2 = GetKey(*lastKeyIdx + 1, keys, size);
i = *lastKeyIdx + 1;
}
else if (!forwards)
{
key2 = GetKey(1, keys, size);
}
// search pairs of keys
int count;
if (forwards)
{
for (count = 1; count <= numKeys; count++, i++)
{
if (i >= numKeys)
{
key1 = GetKey(0, keys, size);
i = 1;
count++;
}
key2 = GetKey(i, keys, size);
if (frame <= key2->fFrame && frame >= key1->fFrame)
{
k2=i;
k1=i-1;
(*kF2) = key2;
(*kF1) = key1;
*p = (time - (*kF1)->fFrame / MAX_FRAMES_PER_SEC) / (((*kF2)->fFrame - (*kF1)->fFrame) / MAX_FRAMES_PER_SEC);
goto ret;
}
key1=key2;
}
}
else
{
for (count = 1; count <= numKeys; count++, i--)
{
if (i < 1)
{
i = numKeys - 1;
key2 = GetKey(i, keys, size);
count++;
}
key1 = GetKey(i - 1, keys, size);
if (frame <= key2->fFrame && frame >= key1->fFrame)
{
k2 = i;
k1 = i - 1;
(*kF2) = key2;
(*kF1) = key1;
*p = (time - (*kF1)->fFrame / MAX_FRAMES_PER_SEC) / (((*kF2)->fFrame - (*kF1)->fFrame) / MAX_FRAMES_PER_SEC);
goto ret;
}
key2=key1;
}
}
ret:
;
#if 0
char str[128];
sprintf(str, "k1=%d, k2=%d, p=%f\n", k1, k2, *p);
OutputDebugString(str);
#endif
*lastKeyIdx = k1;
}