+clipping

11 files changed, 694 insertions, 212 deletions

diff --git a/src/core/device.c b/src/core/device.c
index ad640c6..f6b1a4e 100644
--- a/src/core/device.c
+++ b/src/core/device.c
@@ -1,5 +1,7 @@
 #include "device.h"
 #include "rasterizer.h"
+#include "clip.h"
+#include "shader.h"
 #include "../util/assert.h"
 
 ssr_Config config;
@@ -34,7 +36,7 @@ struct {
 	uint* zbuffer;
 
 	/* 顶点数据 */
-	Vert** verts; uint nverts;
+	Vert* verts; uint nverts;
 	uint* indices; uint nprims;
 
 	/* shader program */
@@ -246,7 +248,7 @@ void ssrU_viewport(Vec2* p, Vec2* out) {
 	//out->y = (int)((1 - p->y) * (halfh - 0.5f));
 }
 
-void ssr_bindvertices(Vert** verts, int nverts, uint* indices, int nprims) {
+void ssr_bindvertices(Vert* verts, int nverts, uint* indices, int nprims) {
 	ssr_assert(verts && indices);
 	state.verts = verts;
 	state.nverts = nverts;
@@ -263,69 +265,119 @@ void ssr_unuseprogram() {
 	state.program = NULL;
 }
 
-static Vec4* homos = NULL;
-static uint tri[3];
-static Vec3 ab, ac;
-static Mat4 mvp, mv;
+VertexShaderIn ssr_vert_in;
+FragmentShaderIn ssr_frag_in;
+
+static struct {
+	Vec4* coords;
+	uint length;
+} clipping_buffer;
+
+static ClippedBuffer clipped_buffer; /*clipping result*/
+
 void ssr_draw(ssr_PrimitiveType primitive) {
 	ssr_assert(state.verts && state.indices);
 
+	static Mat4 mvp, mv;
+
 	ssr_getmvp(&mvp);
 	ssr_getmv(&mv);
-	state.uniforms.model = &GETMATRIX(MATRIX_MODEL);
-	state.uniforms.view = &GETMATRIX(MATRIX_VIEW);
+	state.uniforms.model      = &GETMATRIX(MATRIX_MODEL);
+	state.uniforms.view       = &GETMATRIX(MATRIX_VIEW);
 	state.uniforms.projection = &GETMATRIX(MATRIX_PROJECTION);
-	state.uniforms.mvp = &mvp;
-	state.uniforms.mv = &mv;
+	state.uniforms.mvp        = &mvp;
+	state.uniforms.mv         = &mv;
+
+	uint varying_flag = state.program->varying_flag;
+	bool use_extra_varyings = varying_flag & VARYING_EXTRA;
 
-	VertexShaderIn vertIn = { NULL };
-	if (!homos)
-		homos = ssrM_newvector(Vec4, state.nverts);
+	/*prepare registers if necessary*/
+	if (use_extra_varyings) {
+		ssrS_setregisters(varying_flag, state.nverts);
+	}
+
+	/*resize clipping coords buffer*/
+	if (clipping_buffer.length < state.nverts) {
+		ssrM_resizevector(
+			Vec4, 
+			clipping_buffer.coords, 
+			clipping_buffer.length, 
+			state.nverts, 
+			FALSE
+		);
+	}
+
+	/*vertex operation*/
 	for (int i = 0; i < state.nverts; ++i) {
-		Vert* vert = state.verts[i];
-		vertIn.vertex = vert;
-		state.program->vertexshader(&state.uniforms, &vertIn, &homos[i]);
+		Vert* vert = &state.verts[i];
+		ssr_vert_in.vertex = vert;
+		/*set register pointers*/
+		if (use_extra_varyings) {
+			ssrS_setupregisterpoints(varying_flag, vert->index);
+		}
+		state.program->vertexshader(&state.uniforms, &ssr_vert_in, &clipping_buffer.coords[i]);
 	}
 
+	/*set register pointer to frag-in*/
+	if (use_extra_varyings) {
+		ssrS_setregtofragin(varying_flag, &ssr_frag_in);
+	}
+
+	/*primitive assembly*/
 	if (primitive == PRIMITIVE_TRIANGLE) {
+		uint i0, i1, i2;
+		Vec4 *h0, *h1, *h2;
+		Vert *v0, *v1, *v2;
 		for (int i = 0; i < state.nprims; ++i) {
-			tri[0] = state.indices[i * 3];
-			tri[1] = state.indices[i * 3 + 1];
-			tri[2] = state.indices[i * 3 + 2];
+			i0 = state.indices[i * 3];
+			i1 = state.indices[i * 3 + 1];
+			i2 = state.indices[i * 3 + 2];
+
+			h0 = &clipping_buffer.coords[i0],
+			h1 = &clipping_buffer.coords[i1],
+			h2 = &clipping_buffer.coords[i2];
 
-			/* back face culling */
+			v0 = &state.verts[i0];
+			v1 = &state.verts[i1];
+			v2 = &state.verts[i2];
+
+			/*back face culling*/
 			if (ssr_isenable(ENABLEMASK_BACKFACECULL)) { 
-				float w0 = 1 / homos[tri[0]].w, w1 = 1 / homos[tri[1]].w, w2 = 1 / homos[tri[2]].w;
-				Vec2* h0 = &homos[tri[0]], *h1 = &homos[tri[1]], *h2 = &homos[tri[2]];
-				ab.x = h1->x * w1 - h0->x * w0; /*在屏幕空间做背面剔除*/
+				float w0 = 1 / h0->w, w1 = 1 / h1->w, w2 = 1 / h2->w;
+				Vec3 ab, ac;
+				ab.x = h1->x * w1 - h0->x * w0; 
 				ab.y = h1->y * w1 - h0->y * w0;
 				ac.x = h2->x * w2 - h0->x * w0;
 				ac.y = h2->y * w2 - h0->y * w0;
-				if (ab.x * ac.y - ab.y * ac.x <= 0) {
-					continue;
+				if (ab.x * ac.y - ab.y * ac.x <= 0) 
+					continue; /*cull*/
+			}
+
+			/*clipping*/
+			bool clipped = clip_triangle(h0, h1, h2, v0, v1, v2, varying_flag, &clipped_buffer);
+
+			/*rasterization*/
+			if (!clipped) {
+				ssrR_triangle(h0, h1, h2, v0, v1, v2, TRUE, NULL, NULL, NULL, state.program, &state.uniforms);
+			} else {
+				if (clipped_buffer.count >= 3) {
+					ClippedVert* vt0 = &clipped_buffer.vertices[0], *vt1, *vt2;
+					for (int i = 1; i <= clipped_buffer.count - 2; ++i) { /*0->i->i+1*/
+						vt1 = &clipped_buffer.vertices[i];
+						vt2 = &clipped_buffer.vertices[i + 1];
+						h0 = &vt0->clip_coord; h1 = &vt1->clip_coord; h2 = &vt2->clip_coord;
+						v0 = &vt0->vertex; v1 = &vt1->vertex; v2 = &vt2->vertex;
+						ssrR_triangle(h0, h1, h2, v0, v1, v2, FALSE, vt0, vt1, vt2, state.program, &state.uniforms);
+					}
 				}
 			}
 
-			/* render triangle */
-			ssrR_triangle(
-				&homos[tri[0]],
-				&homos[tri[1]],
-				&homos[tri[2]],
-				state.verts[tri[0]],
-				state.verts[tri[1]],
-				state.verts[tri[2]],
-				state.program,
-				&state.uniforms
-			);
 		}
-	}
-	else if (primitive == PRIMITIVE_POINT) {
-
-	}
-	else if (primitive == PRIMITIVE_LINE) {
+	} else if (primitive == PRIMITIVE_POINT) {
 
+	} else if (primitive == PRIMITIVE_LINE) {
+		
 	}
-	//ssrM_free(homos);
 }
 
 void ssr_setuniformmat4(uint idx, Mat4* src) {
diff --git a/src/core/device.h b/src/core/device.h
index 1350105..10db509 100644
--- a/src/core/device.h
+++ b/src/core/device.h
@@ -64,7 +64,7 @@ void ssr_disable(uint mask);
 bool ssr_isenable(uint mask);
 
 /* 绑定顶点数据 */
-void ssr_bindvertices(Vert** verts, int nverts, uint* indices, int nprims);
+void ssr_bindvertices(Vert* verts, int nverts, uint* indices, int nprims);
 void ssr_unbindvertices();
 
 void ssr_useprogram(Program* program);
diff --git a/src/core/mem.h b/src/core/mem.h
index 1a763f9..e6cc714 100644
--- a/src/core/mem.h
+++ b/src/core/mem.h
@@ -3,10 +3,6 @@
 
 #include <memory.h>
 
-/*
-** 内存分配相关 
-*/
-
 void ssrM_freemem();
 void ssrM_realloc();
 
@@ -15,6 +11,32 @@ void ssrM_realloc();
 #define ssrM_newvector(T, c) ((T*)calloc(c, sizeof(T)))
 #define ssrM_free(obj) free(obj)
 
+/*vector扩容两倍*/
+#define ssrM_scalevector(T, v, curlen, copy) \
+do{\
+ssr_assert(v);\
+T* t = (T*)calloc(curlen << 1, sizeof(T));\
+if (copy) memcpy(t, v, curlen * sizeof(T));\
+free(v);\
+v = t;\
+}while(0)
+
+/*重设vector容量*/
+#define ssrM_resizevector(T, v, curlen, tarlen, copy) \
+do {\
+T* t = (T*)calloc(tarlen, sizeof(T));\
+if (v && copy) memcpy(t, v, min(curlen, tarlen)* sizeof(T));\
+if (v) free(v);\
+v = t;\
+curlen = tarlen;\
+}while(0)
+
+#define ssrM_rescalevector(T, v, curlen, tarlen, copy)\
+do{\
+if (curlen >= tarlen) {}\
+else { ssrM_resizevector(T, v, curlen, tarlen, copy); }\
+} while (0)
+
 #define cast(v, T) (*(T*)(&v))
 
 #endif
\ No newline at end of file
diff --git a/src/core/rasterizer.c b/src/core/rasterizer.c
index 0e5b785..5503e69 100644
--- a/src/core/rasterizer.c
+++ b/src/core/rasterizer.c
@@ -82,18 +82,21 @@ static void puttriangle(Vec2* A, Vec2* B, Vec2* C, Color c) {
 
 #define discardif(condition) if(condition) continue
 
-static Vec3 SA, SB, SC; // screen coords
-static Vec3 bcp; // 重心坐标
-void ssrR_triangle(Vec4* CA, Vec4* CB, Vec4* CC, Vert* A, Vert* B, Vert* C, Program* program, UniformCollection* uniforms) {
+extern FragmentShaderIn ssr_frag_in;
+
+void ssrR_triangle(Vec4* CA, Vec4* CB, Vec4* CC, Vert* A, Vert* B, Vert* C, bool from_reg, ClippedVert* CVA, ClippedVert* CVB, ClippedVert* CVC, Program* program, UniformCollection* uniforms) {
 	ssr_assert(CA && CB && CC && program);
 
+	static Vec3 SA, SB, SC;
+	static Vec3 bcp;
+
 	vec4_dividew(CA, &SA); ssrU_viewport(&SA, &SA);
 	vec4_dividew(CB, &SB); ssrU_viewport(&SB, &SB);
 	vec4_dividew(CC, &SC); ssrU_viewport(&SC, &SC);
-
-	//puttriangle(&SA, &SB, &SC, 0xffff0000);
-	//return;
-
+/*
+	puttriangle(&SA, &SB, &SC, 0xffff0000);
+	return;
+*/
 	Vec3 *sa = &SA, *sb = &SB, *sc = &SC, *tmp; Vec4* v4tmp; Vert* vtemp;
 #define swap(t, a, b) {t = a; a = b; b = t;} /*sort in y axis*/
 	if (sb->y < sa->y) { swap(tmp, sa, sb); swap(v4tmp, CA, CB); swap(vtemp, A, B); }
@@ -102,83 +105,174 @@ void ssrR_triangle(Vec4* CA, Vec4* CB, Vec4* CC, Vert* A, Vert* B, Vert* C, Prog
 #undef swap
 
 	Vec2 AB = {sb->x - sa->x, sb->y - sa->y}, AC = { sc->x - sa->x, sc->y - sa->y };
-	int order = (AB.x * AC.y - AC.x * AB.y) > 0 ? 1 : -1; /*1代表B在AC上方*/
+	int order = (AB.x * AC.y - AC.x * AB.y) > 0 ? 1 : -1;
 
 	float invkAC = (sc->x - sa->x) / (sc->y - sa->y + EPSILON);
 	float invkAB = (sb->x - sa->x) / (sb->y - sa->y + EPSILON);
 	float invkBC = (sc->x - sb->x) / (sc->y - sb->y + EPSILON);
-
-	FragmentShaderIn in = {A, B, C, 0};
+	
 	Color color;
-	float from, to; /*scan line*/
+	float from, to;
 	Vec2 p;
-	float depth; Vec3 cdepth = {sa->z, sb->z, sc->z};
+	float depth;
 	float CAw = 1.f / CA->w, CBw = 1.f / CB->w, CCw = 1.f / CC->w;
-	for (p.y = sa->y; p.y < sb->y; ++p.y) {
-		from = (int)(invkAC * (p.y - sa->y) + sa->x);
-		to = (int)(invkAB * (p.y - sa->y) + sa->x);
-		for (p.x = from; order * (p.x - to) <= 0; p.x += order) {
-
-			if (!ssrR_barycentric(sa, sb, sc, &p, &bcp)) {
-				continue; /*discard fragment*/
-			}
+	
+	uint varying_flag = program->varying_flag;
+
+#define DO_INTERPOLATION  \
+	if (varying_flag & VARYING_BASIC) {																																										 																		 \
+		if (varying_flag & VARYING_POSITION) ssrS_bcpvec3(&bcp, &A->position, &B->position, &C->position, &ssr_frag_in.position); 															 \
+		if (varying_flag & VARYING_NORMAL) ssrS_bcpvec3(&bcp, &A->normal, &B->normal, &C->normal, &ssr_frag_in.normal);					 																 \
+		if (varying_flag & VARYING_TANGENT) ssrS_bcpvec3(&bcp, &A->tangent, &B->tangent, &C->tangent, &ssr_frag_in.tangent);			 															 \
+		if (varying_flag & VARYING_TEXCOORD) ssrS_bcpvec2(&bcp, &A->texcoord, &B->texcoord, &C->texcoord, &ssr_frag_in.texcoord); 															 \
+		if (varying_flag & VARYING_JOINT) ssrS_bcpvec4(&bcp, &A->joint, &B->joint, &C->joint, &ssr_frag_in.joint);								 															 \
+		if (varying_flag & VARYING_WEIGHT) ssrS_bcpvec4(&bcp, &A->weight, &B->weight, &C->weight, &ssr_frag_in.weight);					 																 \
+		if (varying_flag & VARYING_COLOR) ssrS_bcpcolor(&bcp, A->color, B->color, C->color, &ssr_frag_in.color);									 															 \
+	}																																																																													 \
+	if (from_reg) { /*extra varying*/																																																													 \
+		uint i_a = A->index, i_b = B->index, i_c = C->index;																																																		 \
+		if (varying_flag && VARYING_NUM) {																																																											 \
+			if (varying_flag & VARYING_NUM_00) reg_num_00 = ssrS_bcpnum(&bcp, reg_num[0].num[i_a], reg_num[0].num[i_b], reg_num[0].num[i_c], &ssr_frag_in.num[0]); \
+			if (varying_flag & VARYING_NUM_01) reg_num_01 = ssrS_bcpnum(&bcp, reg_num[1].num[i_a], reg_num[1].num[i_b], reg_num[1].num[i_c], &ssr_frag_in.num[1]); \
+			if (varying_flag & VARYING_NUM_02) reg_num_02 = ssrS_bcpnum(&bcp, reg_num[2].num[i_a], reg_num[2].num[i_b], reg_num[2].num[i_c], &ssr_frag_in.num[2]); \
+			if (varying_flag & VARYING_NUM_03) reg_num_03 = ssrS_bcpnum(&bcp, reg_num[3].num[i_a], reg_num[3].num[i_b], reg_num[3].num[i_c], &ssr_frag_in.num[3]); \
+		}																																																																												 \
+		if (varying_flag && VARYING_V2) {																																																												 \
+			if (varying_flag & VARYING_V2_00) reg_v2_00 = ssrS_bcpvec2(&bcp, &reg_v2[0].v2[i_a], &reg_v2[0].v2[i_b], &reg_v2[0].v2[i_c], &ssr_frag_in.v2[0]);			 \
+			if (varying_flag & VARYING_V2_01) reg_v2_01 = ssrS_bcpvec2(&bcp, &reg_v2[1].v2[i_a], &reg_v2[1].v2[i_b], &reg_v2[1].v2[i_c], &ssr_frag_in.v2[1]);			 \
+			if (varying_flag & VARYING_V2_02) reg_v2_02 = ssrS_bcpvec2(&bcp, &reg_v2[2].v2[i_a], &reg_v2[2].v2[i_b], &reg_v2[2].v2[i_c], &ssr_frag_in.v2[2]);			 \
+			if (varying_flag & VARYING_V2_03) reg_v2_03 = ssrS_bcpvec2(&bcp, &reg_v2[3].v2[i_a], &reg_v2[3].v2[i_b], &reg_v2[3].v2[i_c], &ssr_frag_in.v2[3]);			 \
+			if (varying_flag & VARYING_V2_04) reg_v2_04 = ssrS_bcpvec2(&bcp, &reg_v2[4].v2[i_a], &reg_v2[4].v2[i_b], &reg_v2[4].v2[i_c], &ssr_frag_in.v2[4]);			 \
+			if (varying_flag & VARYING_V2_05) reg_v2_05 = ssrS_bcpvec2(&bcp, &reg_v2[5].v2[i_a], &reg_v2[5].v2[i_b], &reg_v2[5].v2[i_c], &ssr_frag_in.v2[5]);			 \
+		}																																																																												 \
+		if (varying_flag && VARYING_V3) {																																																												 \
+			if (varying_flag & VARYING_V3_00) reg_v3_00 = ssrS_bcpvec3(&bcp, &reg_v3[0].v3[i_a], &reg_v3[0].v3[i_b], &reg_v3[0].v3[i_c], &ssr_frag_in.v3[0]);			 \
+			if (varying_flag & VARYING_V3_01) reg_v3_01 = ssrS_bcpvec3(&bcp, &reg_v3[1].v3[i_a], &reg_v3[1].v3[i_b], &reg_v3[1].v3[i_c], &ssr_frag_in.v3[1]);			 \
+			if (varying_flag & VARYING_V3_02) reg_v3_02 = ssrS_bcpvec3(&bcp, &reg_v3[2].v3[i_a], &reg_v3[2].v3[i_b], &reg_v3[2].v3[i_c], &ssr_frag_in.v3[2]);			 \
+			if (varying_flag & VARYING_V3_03) reg_v3_03 = ssrS_bcpvec3(&bcp, &reg_v3[3].v3[i_a], &reg_v3[3].v3[i_b], &reg_v3[3].v3[i_c], &ssr_frag_in.v3[3]);			 \
+			if (varying_flag & VARYING_V3_04) reg_v3_04 = ssrS_bcpvec3(&bcp, &reg_v3[4].v3[i_a], &reg_v3[4].v3[i_b], &reg_v3[4].v3[i_c], &ssr_frag_in.v3[4]);			 \
+			if (varying_flag & VARYING_V3_05) reg_v3_05 = ssrS_bcpvec3(&bcp, &reg_v3[5].v3[i_a], &reg_v3[5].v3[i_b], &reg_v3[5].v3[i_c], &ssr_frag_in.v3[5]);			 \
+			if (varying_flag & VARYING_V3_06) reg_v3_06 = ssrS_bcpvec3(&bcp, &reg_v3[6].v3[i_a], &reg_v3[6].v3[i_b], &reg_v3[6].v3[i_c], &ssr_frag_in.v3[6]);			 \
+			if (varying_flag & VARYING_V3_07) reg_v3_07 = ssrS_bcpvec3(&bcp, &reg_v3[7].v3[i_a], &reg_v3[7].v3[i_b], &reg_v3[7].v3[i_c], &ssr_frag_in.v3[7]);			 \
+		}																																																																												 \
+		if (varying_flag && VARYING_V4) {																																																												 \
+			if (varying_flag & VARYING_V4_00) reg_v4_00 = ssrS_bcpvec4(&bcp, &reg_v4[0].v4[i_a], &reg_v4[0].v4[i_b], &reg_v4[0].v4[i_c], &ssr_frag_in.v4[0]);			 \
+			if (varying_flag & VARYING_V4_01) reg_v4_01 = ssrS_bcpvec4(&bcp, &reg_v4[1].v4[i_a], &reg_v4[1].v4[i_b], &reg_v4[1].v4[i_c], &ssr_frag_in.v4[1]);			 \
+			if (varying_flag & VARYING_V4_02) reg_v4_02 = ssrS_bcpvec4(&bcp, &reg_v4[2].v4[i_a], &reg_v4[2].v4[i_b], &reg_v4[2].v4[i_c], &ssr_frag_in.v4[2]);			 \
+			if (varying_flag & VARYING_V4_03) reg_v4_03 = ssrS_bcpvec4(&bcp, &reg_v4[3].v4[i_a], &reg_v4[3].v4[i_b], &reg_v4[3].v4[i_c], &ssr_frag_in.v4[3]);			 \
+			if (varying_flag & VARYING_V4_04) reg_v4_04 = ssrS_bcpvec4(&bcp, &reg_v4[4].v4[i_a], &reg_v4[4].v4[i_b], &reg_v4[4].v4[i_c], &ssr_frag_in.v4[4]);			 \
+			if (varying_flag & VARYING_V4_05) reg_v4_05 = ssrS_bcpvec4(&bcp, &reg_v4[5].v4[i_a], &reg_v4[5].v4[i_b], &reg_v4[5].v4[i_c], &ssr_frag_in.v4[5]);			 \
+		}																																																																												 \
+	} else {																																																																									 \
+		if (varying_flag && VARYING_NUM) {																																																											 \
+			if (varying_flag & VARYING_NUM_00) reg_num_00 = ssrS_bcpnum(&bcp, CVA->num[0], CVB->num[0], CVC->num[0], &ssr_frag_in.num[0]);												 \
+			if (varying_flag & VARYING_NUM_01) reg_num_01 = ssrS_bcpnum(&bcp, CVA->num[1], CVB->num[1], CVC->num[1], &ssr_frag_in.num[1]);												 \
+			if (varying_flag & VARYING_NUM_02) reg_num_02 = ssrS_bcpnum(&bcp, CVA->num[2], CVB->num[2], CVC->num[2], &ssr_frag_in.num[2]);												 \
+			if (varying_flag & VARYING_NUM_03) reg_num_03 = ssrS_bcpnum(&bcp, CVA->num[3], CVB->num[3], CVC->num[3], &ssr_frag_in.num[3]);												 \
+		}																																																																												 \
+		if (varying_flag && VARYING_V2) {																																																												 \
+			if (varying_flag & VARYING_V2_00) reg_v2_00 = ssrS_bcpvec2(&bcp, &CVA->v2[0], &CVB->v2[0], &CVC->v2[0], &ssr_frag_in.v2[0]);													 \
+			if (varying_flag & VARYING_V2_01) reg_v2_01 = ssrS_bcpvec2(&bcp, &CVA->v2[1], &CVB->v2[1], &CVC->v2[1], &ssr_frag_in.v2[1]);													 \
+			if (varying_flag & VARYING_V2_02) reg_v2_02 = ssrS_bcpvec2(&bcp, &CVA->v2[2], &CVB->v2[2], &CVC->v2[2], &ssr_frag_in.v2[2]);													 \
+			if (varying_flag & VARYING_V2_03) reg_v2_03 = ssrS_bcpvec2(&bcp, &CVA->v2[3], &CVB->v2[3], &CVC->v2[3], &ssr_frag_in.v2[3]);													 \
+			if (varying_flag & VARYING_V2_04) reg_v2_04 = ssrS_bcpvec2(&bcp, &CVA->v2[4], &CVB->v2[4], &CVC->v2[4], &ssr_frag_in.v2[4]);													 \
+			if (varying_flag & VARYING_V2_05) reg_v2_05 = ssrS_bcpvec2(&bcp, &CVA->v2[5], &CVB->v2[5], &CVC->v2[5], &ssr_frag_in.v2[5]);													 \
+		}																																																																												 \
+		if (varying_flag && VARYING_V3) {																																																												 \
+			if (varying_flag & VARYING_V3_00) reg_v3_00 = ssrS_bcpvec3(&bcp, &CVA->v3[0], &CVB->v3[0], &CVC->v3[0], &ssr_frag_in.v3[0]);													 \
+			if (varying_flag & VARYING_V3_01) reg_v3_01 = ssrS_bcpvec3(&bcp, &CVA->v3[1], &CVB->v3[1], &CVC->v3[1], &ssr_frag_in.v3[1]);													 \
+			if (varying_flag & VARYING_V3_02) reg_v3_02 = ssrS_bcpvec3(&bcp, &CVA->v3[2], &CVB->v3[2], &CVC->v3[2], &ssr_frag_in.v3[2]);													 \
+			if (varying_flag & VARYING_V3_03) reg_v3_03 = ssrS_bcpvec3(&bcp, &CVA->v3[3], &CVB->v3[3], &CVC->v3[3], &ssr_frag_in.v3[3]);													 \
+			if (varying_flag & VARYING_V3_04) reg_v3_04 = ssrS_bcpvec3(&bcp, &CVA->v3[4], &CVB->v3[4], &CVC->v3[4], &ssr_frag_in.v3[4]);													 \
+			if (varying_flag & VARYING_V3_05) reg_v3_05 = ssrS_bcpvec3(&bcp, &CVA->v3[5], &CVB->v3[5], &CVC->v3[5], &ssr_frag_in.v3[5]);													 \
+			if (varying_flag & VARYING_V3_06) reg_v3_06 = ssrS_bcpvec3(&bcp, &CVA->v3[6], &CVB->v3[6], &CVC->v3[6], &ssr_frag_in.v3[6]);													 \
+			if (varying_flag & VARYING_V3_07) reg_v3_07 = ssrS_bcpvec3(&bcp, &CVA->v3[7], &CVB->v3[7], &CVC->v3[7], &ssr_frag_in.v3[7]);													 \
+		}																																																																												 \
+		if (varying_flag && VARYING_V4) {																																																												 \
+			if (varying_flag & VARYING_V4_00) reg_v4_00 = ssrS_bcpvec4(&bcp, &CVA->v4[0], &CVB->v4[0], &CVC->v4[0], &ssr_frag_in.v4[0]);													 \
+			if (varying_flag & VARYING_V4_01) reg_v4_01 = ssrS_bcpvec4(&bcp, &CVA->v4[1], &CVB->v4[1], &CVC->v4[1], &ssr_frag_in.v4[1]);													 \
+			if (varying_flag & VARYING_V4_02) reg_v4_02 = ssrS_bcpvec4(&bcp, &CVA->v4[2], &CVB->v4[2], &CVC->v4[2], &ssr_frag_in.v4[2]);													 \
+			if (varying_flag & VARYING_V4_03) reg_v4_03 = ssrS_bcpvec4(&bcp, &CVA->v4[3], &CVB->v4[3], &CVC->v4[3], &ssr_frag_in.v4[3]);													 \
+			if (varying_flag & VARYING_V4_04) reg_v4_04 = ssrS_bcpvec4(&bcp, &CVA->v4[4], &CVB->v4[4], &CVC->v4[4], &ssr_frag_in.v4[4]);													 \
+			if (varying_flag & VARYING_V4_05) reg_v4_05 = ssrS_bcpvec4(&bcp, &CVA->v4[5], &CVB->v4[5], &CVC->v4[5], &ssr_frag_in.v4[5]);													 \
+		}																																																																												 \
+	}
 
-			discardif(bcp.x < 0 || bcp.y < 0 || bcp.z < 0);
+	Vec3 s[2], u;
+	s[0].x = sc->x - sa->x; s[0].y = sb->x - sa->x;
+	s[1].x = sc->y - sa->y; s[1].y = sb->y - sa->y;
+
+#define RENDER_TRIANGLE \
+	for (p.y = FROM->y; p.y < TO->y; ++p.y) {									       \
+		SET_FROM_AND_TO	                                               \
+		s[1].z = sa->y - p.y;																			     \
+		for (p.x = from; order * (p.x - to) <= 0; p.x += order) {	     \
+			s[0].z = sa->x - p.x;																		     \
+			vec3_cross(&s[0], &s[1], &u);													       \
+			discardif(compare(u.z, 0));														       \
+			u.z = 1.f / u.z;																			       \
+			bcp.x = 1.f - (u.x + u.y) * u.z;											       \
+			bcp.y = u.y * u.z;																		       \
+			bcp.z = u.x * u.z;																		       \
+			discardif(bcp.x < 0 || bcp.y < 0 || bcp.z < 0);				       \
+			/*perspective correction*/																	 \
+			bcp.x *= CAw; bcp.y *= CBw; bcp.z *= CCw;							       \
+			vec3_scale(&bcp, 1.f / (bcp.x + bcp.y + bcp.z), &bcp);       \
+			/*depth test*/																					     \
+			depth = (bcp.x*sa->z+bcp.y*sb->z+bcp.z*sc->z) * 0.5f + 0.5f; \
+			discardif(!ssr_testdepthf(p.x, p.y, depth));						     \
+			/*set varying variables*/																		 \
+			DO_INTERPOLATION																						 \
+			if(program->fragmentshader(uniforms, &ssr_frag_in, &color))	 \
+				ssr_putpoint(p.x, p.y, color);													   \
+		}																													     \
+	}
 
-			bcp.x *= CAw; bcp.y *= CBw; bcp.z *= CCw;
-			vec3_scale(&bcp, 1.f / (bcp.x + bcp.y + bcp.z), &bcp);
-			/*depth test*/
-			depth = vec3_dot(&bcp, &cdepth) * 0.5f + 0.5f;
-			if (!ssr_testdepthf(p.x, p.y, depth)) {
-				continue; /*discard fragment*/
-			}
-			/*enter fragment shader*/
-			in.bc = &bcp;
-			program->fragmentshader(uniforms, &in, &color);
-			ssr_putpoint(p.x, p.y, color);
-		}
+#define FROM    sa 
+#define TO      sb 
+#define FROMK   invkAC 
+#define TOK     invkAB
+#define SET_FROM_AND_TO \
+	from = (int)(invkAC * (p.y - sa->y) + sa->x); \
+	to = (int)(invkAB * (p.y - sa->y) + sa->x);   \
+
+	RENDER_TRIANGLE
+
+#define FROM    sb 
+#define TO      sc
+#define FROMK   invkAC 
+#define TOK     invkBC
+#define SET_FROM_AND_TO \
+	if (p.y == sb->y && sb->y == sc->y) {					  \
+		from = sc->x;																  \
+		to = sb->x;																	  \
+	}																							  \
+	else {																				  \
+		from = (int)(invkAC * (p.y - sa->y) + sa->x); \
+		to = (int)(invkBC * (p.y - sb->y) + sb->x);	  \
 	}
-	for (p.y = sb->y; p.y <= sc->y; ++p.y) {
-		if (p.y == sb->y && sb->y == sc->y) {
-			from = sc->x; 
-			to = sb->x;
-		}
-		else {
-			from = (int)(invkAC * (p.y - sa->y) + sa->x);
-			to = (int)(invkBC * (p.y - sb->y) + sb->x);
-		}
-		for (p.x = from; order * (p.x - to) <= 0; p.x += order) {
-			/*计算线性的（齐次裁剪空间）的重心坐标，分三步*/
-			if (!ssrR_barycentric(sa, sb, sc, &p, &bcp)) {
-				continue; /*discard fragment*/
-			}
 
-			discardif(bcp.x < 0 || bcp.y < 0 || bcp.z < 0);
+	RENDER_TRIANGLE
 
-			bcp.x *= CAw; bcp.y *= CBw; bcp.z *= CCw;
-			vec3_scale(&bcp, 1.f / (bcp.x + bcp.y + bcp.z), &bcp);
-			/*depth test*/
-			depth = vec3_dot(&bcp, &cdepth) * 0.5f + 0.5f;
-			if (!ssr_testdepthf(p.x, p.y, depth)) {
-				continue; /*discard fragment*/
-			}
-			/*enter fragment shader*/
-			in.bc = &bcp;
-			program->fragmentshader(uniforms, &in, &color);
-			ssr_putpoint(p.x, p.y, color);
-		}
-	}
+#undef FROM 
+#undef TO
+#undef FROMK
+#undef TOK
+#undef SET_FROM_AND_TO
+#undef RENDER_TRIANGLE
 }
 
-static Vec2 bboxmin, bboxmax;
-void _ssrR_triangle(Vec4* CA, Vec4* CB, Vec4* CC, Vert* A, Vert* B, Vert* C, Program* program, UniformCollection* uniforms) {
+/*deprecated*/
+/*
+static Vec2 bboxmin, bboxmax; 
+void ssrR_triangle(Vec4* CA, Vec4* CB, Vec4* CC, Vert* A, Vert* B, Vert* C, Program* program, UniformCollection* uniforms) {
 	ssr_assert(CA && CB && CC && program && A && B && C && uniforms);
 
 	vec4_dividew(CA, &SA); ssrU_viewport(&SA, &SA);
 	vec4_dividew(CB, &SB); ssrU_viewport(&SB, &SB);
 	vec4_dividew(CC, &SC); ssrU_viewport(&SC, &SC);
 
-	bboxmin.x = (int)min(SA.x, min(SB.x, SC.x)); /*这里一定要int，确保没有缝隙*/
+	bboxmin.x = (int)min(SA.x, min(SB.x, SC.x));
 	bboxmax.x = (int)max(SA.x, max(SB.x, SC.x));
 	bboxmin.y = (int)min(SA.y, min(SB.y, SC.y));
 	bboxmax.y = (int)max(SA.y, max(SB.y, SC.y));
@@ -195,7 +289,6 @@ void _ssrR_triangle(Vec4* CA, Vec4* CB, Vec4* CC, Vert* A, Vert* B, Vert* C, Pro
 	for (p.y = bboxmin.y; p.y <= bboxmax.y; ++p.y) {
 		s[1].z = SA.y - p.y;
 		for (p.x = bboxmin.x; p.x <= bboxmax.x; ++p.x) {
-			/*计算线性的（齐次裁剪空间）的重心坐标*/
 			s[0].z = SA.x - p.x;
 			vec3_cross(&s[0], &s[1], &u);
 
@@ -211,13 +304,11 @@ void _ssrR_triangle(Vec4* CA, Vec4* CB, Vec4* CC, Vert* A, Vert* B, Vert* C, Pro
 			bcp.x *= CAw; bcp.y *= CBw; bcp.z *= CCw;
 			vec3_scale(&bcp, 1.f / (bcp.x + bcp.y + bcp.z), &bcp);
 
-			/*depth test*/
 			if (ssr_isenable(ENABLEMASK_DEPTHTEST)) {
 				depth = vec3_dot(&bcp, &cdepth) * 0.5f + 0.5f;
 				discardif(!ssr_testdepthf(p.x, p.y, depth));
 			}
 
-			/*enter fragment shader*/
 			in.bc = &bcp;
 			dontdiscad = program->fragmentshader(uniforms, &in, &color);
 			if (dontdiscad)
@@ -225,6 +316,7 @@ void _ssrR_triangle(Vec4* CA, Vec4* CB, Vec4* CC, Vert* A, Vert* B, Vert* C, Pro
 		}
 	}
 }
+*/
 
 void ssrR_line(Vec4* CA, Vec4* CB, Vert* A, Vert* B, Program* program, UniformCollection* uniforms) {
 	ssr_assert(CA && CB && program && A && B && uniforms);
diff --git a/src/core/rasterizer.h b/src/core/rasterizer.h
index e35b24a..eceb8c7 100644
--- a/src/core/rasterizer.h
+++ b/src/core/rasterizer.h
@@ -3,6 +3,7 @@
 
 #include "../math/math.h"
 #include "shader.h"
+#include "clip.h"
 
 void ssrR_putline(int x0, int y0, int x1, int y1, Color color);
 
@@ -12,6 +13,11 @@ bool ssrR_barycentric(Vec2* A, Vec2* B, Vec2* C, Vec2* p, Vec3* out); /*计算重心
 void ssrR_center(Vec2* A, Vec2* B, Vec2* C, Vec2* out); /*获得重心*/
 bool ssrR_ispointintriangle(Vec2* A, Vec2* B, Vec2* C, Vec2* p);
 
-void ssrR_triangle(Vec4* CA, Vec4* CB, Vec4* CC, Vert* A, Vert* B, Vert* C, Program* program, UniformCollection* uniforms); /*绘制三角形图元*/
+void ssrR_triangle(Vec4* CA, Vec4* CB, Vec4* CC, 
+	Vert* A, Vert* B, Vert* C, 
+	bool from_reg, 
+	ClippedVert* CVA, ClippedVert* CVB, ClippedVert* CVC, 
+	Program* program, UniformCollection* uniforms
+);
 
 #endif
\ No newline at end of file
diff --git a/src/core/shader.c b/src/core/shader.c
index fe25408..b5d8407 100644
--- a/src/core/shader.c
+++ b/src/core/shader.c
@@ -1,8 +1,226 @@
+#include "../math/math.h"
 #include "shader.h"
 #include "vert.h"
 #include "device.h"
 
-void ssrS_bcpcolor(Vec3* bc, Color A, Color B, Color C, Color* out) {
+float* reg_num_00;
+float* reg_num_01;
+float* reg_num_02;
+float* reg_num_03;
+
+Vec2* reg_v2_00;
+Vec2* reg_v2_01;
+Vec2* reg_v2_02;
+Vec2* reg_v2_03;
+Vec2* reg_v2_04;
+Vec2* reg_v2_05;
+
+Vec3* reg_v3_00;
+Vec3* reg_v3_01;
+Vec3* reg_v3_02;
+Vec3* reg_v3_03;
+Vec3* reg_v3_04;
+Vec3* reg_v3_05;
+Vec3* reg_v3_06;
+Vec3* reg_v3_07;
+
+Vec4* reg_v4_00;
+Vec4* reg_v4_01;
+Vec4* reg_v4_02;
+Vec4* reg_v4_03;
+Vec4* reg_v4_04;
+Vec4* reg_v4_05;
+
+Register reg_num[4] = {
+	{ 0, REGTYPE_NUM, NULL },
+	{ 0, REGTYPE_NUM, NULL },
+	{ 0, REGTYPE_NUM, NULL },
+	{ 0, REGTYPE_NUM, NULL },
+};
+
+Register reg_v2[6] = {
+	{ 0, REGTYPE_VEC2, NULL },
+	{ 0, REGTYPE_VEC2, NULL },
+	{ 0, REGTYPE_VEC2, NULL },
+	{ 0, REGTYPE_VEC2, NULL },
+	{ 0, REGTYPE_VEC2, NULL },
+	{ 0, REGTYPE_VEC2, NULL },
+};
+
+Register reg_v3[8] = {
+	{ 0, REGTYPE_VEC3, NULL },
+	{ 0, REGTYPE_VEC3, NULL },
+	{ 0, REGTYPE_VEC3, NULL },
+	{ 0, REGTYPE_VEC3, NULL },
+	{ 0, REGTYPE_VEC3, NULL },
+	{ 0, REGTYPE_VEC3, NULL },
+	{ 0, REGTYPE_VEC3, NULL },
+	{ 0, REGTYPE_VEC3, NULL },
+};
+
+Register reg_v4[6] = {
+	{ 0, REGTYPE_VEC4, NULL },
+	{ 0, REGTYPE_VEC4, NULL },
+	{ 0, REGTYPE_VEC4, NULL },
+	{ 0, REGTYPE_VEC4, NULL },
+	{ 0, REGTYPE_VEC4, NULL },
+	{ 0, REGTYPE_VEC4, NULL },
+};
+
+void ssrS_setregisters(uint flag, int capacity) {
+#define reg_scale_num(i) \
+if (flag & VARYING_NUM_##i) ssrM_rescalevector(float, reg_num[##i].num, reg_num[##i].length, capacity, FALSE)
+
+#define reg_scale_v2(i) \
+if (flag & VARYING_V2_##i) ssrM_rescalevector(float, reg_v2[##i].v2, reg_v2[##i].length, capacity, FALSE)
+
+#define reg_scale_v3(i) \
+if (flag & VARYING_V3_##i) ssrM_rescalevector(float, reg_v3[##i].v3, reg_v3[##i].length, capacity, FALSE)
+
+#define reg_scale_v4(i) \
+if (flag & VARYING_V4_##i) ssrM_rescalevector(float, reg_v4[##i].v4, reg_v4[##i].length, capacity, FALSE)
+
+	if (!(flag & VARYING_EXTRA))
+		return;
+	if (flag & VARYING_NUM) {
+		reg_scale_num(00);
+		reg_scale_num(01);
+		reg_scale_num(02);
+		reg_scale_num(03);
+	}
+	if (flag & VARYING_V2) {
+		reg_scale_v2(00);
+		reg_scale_v2(01);
+		reg_scale_v2(02);
+		reg_scale_v2(03);
+		reg_scale_v2(04);
+		reg_scale_v2(05);
+	}
+	if (flag & VARYING_V3) {
+		reg_scale_v3(00);
+		reg_scale_v3(01);
+		reg_scale_v3(02);
+		reg_scale_v3(03);
+		reg_scale_v3(04);
+		reg_scale_v3(05);
+		reg_scale_v3(06);
+		reg_scale_v3(07);
+	}
+	if (flag & VARYING_V4) {
+		reg_scale_v4(00);
+		reg_scale_v4(01);
+		reg_scale_v4(02);
+		reg_scale_v4(03);
+		reg_scale_v4(04);
+		reg_scale_v4(05);
+	}
+}
+
+void ssrS_setupregisterpoints(uint extra_varying_flag, int idx) {
+#define set_reg_num_pointer(i)\
+		if (extra_varying_flag & VARYING_NUM_##i) reg_num_##i = &reg_num[##i].num[idx];
+
+#define set_reg_v2_pointer(i)\
+		if (extra_varying_flag & VARYING_V2_##i) reg_v2_##i = &reg_v2[##i].v2[idx];
+
+#define set_reg_v3_pointer(i)\
+		if (extra_varying_flag & VARYING_V3_##i) reg_v3_##i = &reg_v3[##i].v3[idx];
+
+#define set_reg_v4_pointer(i)\
+		if (extra_varying_flag & VARYING_V4_##i) reg_v4_##i = &reg_v4[##i].v4[idx];
+
+	if (!(extra_varying_flag & VARYING_EXTRA))
+		return;
+	if (extra_varying_flag & VARYING_NUM) {
+		set_reg_num_pointer(00);
+		set_reg_num_pointer(01);
+		set_reg_num_pointer(02);
+		set_reg_num_pointer(03);
+	}
+	if (extra_varying_flag & VARYING_V2) {
+		set_reg_v2_pointer(00);
+		set_reg_v2_pointer(01);
+		set_reg_v2_pointer(02);
+		set_reg_v2_pointer(03);
+		set_reg_v2_pointer(04);
+		set_reg_v2_pointer(05);
+	}
+	if (extra_varying_flag & VARYING_V3) {
+		set_reg_v3_pointer(00);
+		set_reg_v3_pointer(01);
+		set_reg_v3_pointer(02);
+		set_reg_v3_pointer(03);
+		set_reg_v3_pointer(04);
+		set_reg_v3_pointer(05);
+		set_reg_v3_pointer(06);
+		set_reg_v3_pointer(07);
+	}
+	if (extra_varying_flag & VARYING_V4) {
+		set_reg_v4_pointer(00);
+		set_reg_v4_pointer(01);
+		set_reg_v4_pointer(02);
+		set_reg_v4_pointer(03);
+		set_reg_v4_pointer(04);
+		set_reg_v4_pointer(05);
+	}
+}
+
+void ssrS_setregtofragin(uint extra_varying_flag, FragmentShaderIn* frag_in){
+#define set_reg_num_pointer_to_fragin(i)\
+		if (extra_varying_flag & VARYING_NUM_##i) reg_num_##i = &frag_in->num[##i];
+
+#define set_reg_v2_pointer_to_fragin(i)\
+		if (extra_varying_flag & VARYING_V2_##i) reg_v2_##i = &frag_in->v2[##i];
+
+#define set_reg_v3_pointer_to_fragin(i)\
+		if (extra_varying_flag & VARYING_V3_##i) reg_v3_##i = &frag_in->v3[##i];
+
+#define set_reg_v4_pointer_to_fragin(i)\
+		if (extra_varying_flag & VARYING_V4_##i) reg_v4_##i = &frag_in->v4[##i];
+
+	if (!(extra_varying_flag & VARYING_EXTRA))
+		return;
+	if (extra_varying_flag & VARYING_NUM) {
+		set_reg_num_pointer_to_fragin(00);
+		set_reg_num_pointer_to_fragin(01);
+		set_reg_num_pointer_to_fragin(02);
+		set_reg_num_pointer_to_fragin(03);
+	}
+	if (extra_varying_flag & VARYING_V2) {
+		set_reg_v2_pointer_to_fragin(00);
+		set_reg_v2_pointer_to_fragin(01);
+		set_reg_v2_pointer_to_fragin(02);
+		set_reg_v2_pointer_to_fragin(03);
+		set_reg_v2_pointer_to_fragin(04);
+		set_reg_v2_pointer_to_fragin(05);
+	}
+	if (extra_varying_flag & VARYING_V3) {
+		set_reg_v3_pointer_to_fragin(00);
+		set_reg_v3_pointer_to_fragin(01);
+		set_reg_v3_pointer_to_fragin(02);
+		set_reg_v3_pointer_to_fragin(03);
+		set_reg_v3_pointer_to_fragin(04);
+		set_reg_v3_pointer_to_fragin(05);
+		set_reg_v3_pointer_to_fragin(06);
+		set_reg_v3_pointer_to_fragin(07);
+	}
+	if (extra_varying_flag & VARYING_V4) {
+		set_reg_v4_pointer_to_fragin(00);
+		set_reg_v4_pointer_to_fragin(01);
+		set_reg_v4_pointer_to_fragin(02);
+		set_reg_v4_pointer_to_fragin(03);
+		set_reg_v4_pointer_to_fragin(04);
+		set_reg_v4_pointer_to_fragin(05);
+	}
+}
+
+float* ssrS_bcpnum(Vec3* bc, float A, float B, float C, float* out) {
+	ssr_assert(bc && out);
+	*out = bc->A * A + bc->B * B + bc->C * C;
+	return out;
+}
+
+Color* ssrS_bcpcolor(Vec3* bc, Color A, Color B, Color C, Color* out) {
 	ssr_assert(bc && out);
 	*out = ssr_color(
 		bc->A * COLOR_R(A) + bc->B * COLOR_R(B) + bc->C * COLOR_R(C),
@@ -10,25 +228,60 @@ void ssrS_bcpcolor(Vec3* bc, Color A, Color B, Color C, Color* out) {
 		bc->A * COLOR_B(A) + bc->B * COLOR_B(B) + bc->C * COLOR_B(C),
 		bc->A * COLOR_A(A) + bc->B * COLOR_A(B) + bc->C * COLOR_A(C)
 	);
+	return out;
 }
 
-void ssrS_bcpvec2(Vec3* bc, Vec2* A, Vec2* B, Vec2* C, Vec2* out) {
+Vec2* ssrS_bcpvec2(Vec3* bc, Vec2* A, Vec2* B, Vec2* C, Vec2* out) {
 	ssr_assert(bc && A && B && C && out);
 	out->x = bc->A * A->x + bc->B * B->x + bc->C * C->x;
 	out->y = bc->A * A->y + bc->B * B->y + bc->C * C->y;
+	return out;
 }
 
-void ssrS_bcpvec3(Vec3* bc, Vec3* A, Vec3* B, Vec3* C, Vec3* out) {
+Vec3* ssrS_bcpvec3(Vec3* bc, Vec3* A, Vec3* B, Vec3* C, Vec3* out) {
 	ssr_assert(bc && A && B && C && out);
 	out->x = bc->A * A->x + bc->B * B->x + bc->C * C->x;
 	out->y = bc->A * A->y + bc->B * B->y + bc->C * C->y;
 	out->z = bc->A * A->z + bc->B * B->z + bc->C * C->z;
+	return out;
 }
 
-void ssrS_bcpvec4(Vec3* bc, Vec4* A, Vec4* B, Vec4* C, Vec4* out) {
+Vec4* ssrS_bcpvec4(Vec3* bc, Vec4* A, Vec4* B, Vec4* C, Vec4* out) {
 	ssr_assert(bc && A && B && C && out);
 	out->x = bc->A * A->x + bc->B * B->x + bc->C * C->x;
 	out->y = bc->A * A->y + bc->B * B->y + bc->C * C->y;
 	out->z = bc->A * A->z + bc->B * B->z + bc->C * C->z;
 	out->w = bc->A * A->w + bc->B * B->w + bc->C * C->w;
+	return out;
+}
+
+void ssrS_lerpnum(float t, float A, float B, float* out) {
+	*out = lerp(A, B, t);
+}
+
+void ssrS_lerpcolor(float t, Color A, Color B, Color* out) {
+	*out = ssr_color(
+		lerp(COLOR_R(A), COLOR_R(B), t),
+		lerp(COLOR_G(A), COLOR_G(B), t),
+		lerp(COLOR_B(A), COLOR_B(B), t),
+		lerp(COLOR_A(A), COLOR_A(B), t)
+	);
+}
+
+void ssrS_lerpvec2(float t, Vec2* A, Vec2* B, Vec2* out) {
+	out->x = lerp(A->x, B->x, t);
+	out->y = lerp(A->y, B->y, t);
+}
+
+void ssrS_lerpvec3(float t, Vec3* A, Vec3* B, Vec3* out) {
+	out->x = lerp(A->x, B->x, t);
+	out->y = lerp(A->y, B->y, t);
+	out->z = lerp(A->z, B->z, t);
+}
+
+void ssrS_lerpvec4(float t, Vec4* A, Vec4* B, Vec4* out) {
+	out->x = lerp(A->x, B->x, t);
+	out->y = lerp(A->y, B->y, t);
+	out->z = lerp(A->z, B->z, t);
+	out->w = lerp(A->w, B->w, t);
 }
diff --git a/src/core/shader.h b/src/core/shader.h
index 15dfc26..3024a0d 100644
--- a/src/core/shader.h
+++ b/src/core/shader.h
@@ -11,17 +11,21 @@ typedef struct UniformCollection {
 	Mat4* projection;
 	Mat4* mvp;
 	Mat4* mv;
-	/*userside variables*/
+	/*extra variables*/
 	Mat4 var_mat4[8];
 	Vec4 var_vec4[8];
 	Vec3 var_vec3[8];
 	Vec2 var_vec2[8];
+	float var_num[8];
+	/*userdata*/
+	void* userdata;
 } UniformCollection;
 
 #define UM4(i) (&uniforms->var_mat4[i])
 #define UV2(i) (&uniforms->var_vec2[i])
 #define UV3(i) (&uniforms->var_vec3[i])
 #define UV4(i) (&uniforms->var_vec4[i])
+#define UN(i) (&uniforms->var_num[i])
 
 typedef struct VertexShaderIn {
 	Vert* vertex;
@@ -30,58 +34,163 @@ typedef struct VertexShaderIn {
 typedef void(*VertexShader)(UniformCollection* uniforms, VertexShaderIn* in, Vec4* homocoord);
 
 typedef struct FragmentShaderIn {
-	Vert *A, *B, *C;
-	Vec3* bc; 
+	/*value from vertices interpolation*/
+	Vec3 position;
+	Vec3 normal; 
+	Vec3 tangent; 
+	Vec2 texcoord; 
+	Color color; 
+	Vec4 joint;
+	Vec4 weight;
+	/*value from registers interpolation*/
+	float num[4];
+	Vec2 v2[6];
+	Vec3 v3[8];
+	Vec4 v4[6];
 } FragmentShaderIn;
 
 typedef bool(*FragmentShader)(UniformCollection* uniforms, FragmentShaderIn* in, Color* color);
 
 typedef struct Program {
-	VertexShader vertexshader; 
+	VertexShader vertexshader;
 	FragmentShader fragmentshader; 
+	uint varying_flag;
 } Program;
 
-void ssrS_bcpcolor(Vec3* bc, Color A, Color B, Color C, Color* out);
-void ssrS_bcpvec2(Vec3* bc, Vec2* A, Vec2* B, Vec2* C, Vec2* out);
-void ssrS_bcpvec3(Vec3* bc, Vec3* A, Vec3* B, Vec3* C, Vec3* out);
-void ssrS_bcpvec4(Vec3* bc, Vec4* A, Vec4* B, Vec4* C, Vec4* out);
+#define VARYING_NONE     (0)
 
+#define VARYING_POSITION (1 << 31)
+#define VARYING_NORMAL   (1 << 30)
+#define VARYING_TANGENT  (1 << 29)
+#define VARYING_TEXCOORD (1 << 28)
+#define VARYING_COLOR		 (1 << 27)
+#define VARYING_JOINT		 (1 << 26)
+#define VARYING_WEIGHT   (1 << 25)
+
+#define VARYING_UNUSED   (1 << 24) 
+
+#define VARYING_BASIC  ( VARYING_UNUSED | VARYING_POSITION | VARYING_NORMAL | VARYING_TANGENT | VARYING_TEXCOORD | VARYING_COLOR | VARYING_JOINT | VARYING_WEIGHT )
+
+#define VARYING_NUM_00		 (1 << 20)
+#define VARYING_NUM_01		 (1 << 21)
+#define VARYING_NUM_02		 (1 << 22)
+#define VARYING_NUM_03		 (1 << 23)
+
+#define VARYING_NUM      (VARYING_NUM_00 | VARYING_NUM_01 | VARYING_NUM_02 | VARYING_NUM_03 )
+												 
+#define VARYING_V2_00		 (1)
+#define VARYING_V2_01		 (1 << 1)
+#define VARYING_V2_02		 (1 << 2)
+#define VARYING_V2_03		 (1 << 3)
+#define VARYING_V2_04		 (1 << 4)
+#define VARYING_V2_05		 (1 << 5)
+
+#define VARYING_V2       (VARYING_V2_00 | VARYING_V2_01 | VARYING_V2_02 | VARYING_V2_03 | VARYING_V2_04 | VARYING_V2_05)
+												 
+#define VARYING_V3_00		 (1 << 6)
+#define VARYING_V3_01		 (1 << 7)
+#define VARYING_V3_02		 (1 << 8)
+#define VARYING_V3_03		 (1 << 9)
+#define VARYING_V3_04		 (1 << 10)
+#define VARYING_V3_05		 (1 << 11)
+#define VARYING_V3_06		 (1 << 12)
+#define VARYING_V3_07		 (1 << 13)
+
+#define VARYING_V3       (VARYING_V3_00 | VARYING_V3_01 | VARYING_V3_02 | VARYING_V3_03 | VARYING_V3_04 | VARYING_V3_05| VARYING_V3_06| VARYING_V3_07)
+												 
+#define VARYING_V4_00		 (1 << 14)
+#define VARYING_V4_01		 (1 << 15)
+#define VARYING_V4_02		 (1 << 16)
+#define VARYING_V4_03		 (1 << 17)
+#define VARYING_V4_04		 (1 << 18)
+#define VARYING_V4_05		 (1 << 19)
+
+#define VARYING_V4       (VARYING_V4_00 | VARYING_V4_01 | VARYING_V4_02 | VARYING_V4_03 | VARYING_V4_04 | VARYING_V4_05)
+
+#define VARYING_EXTRA    (~VARYING_BASIC)
+
+float* ssrS_bcpnum(Vec3* bc, float A, float B, float C, float* out);
+Color* ssrS_bcpcolor(Vec3* bc, Color A, Color B, Color C, Color* out);
+Vec2*  ssrS_bcpvec2(Vec3* bc, Vec2* A, Vec2* B, Vec2* C, Vec2* out);
+Vec3* ssrS_bcpvec3(Vec3* bc, Vec3* A, Vec3* B, Vec3* C, Vec3* out);
+Vec4* ssrS_bcpvec4(Vec3* bc, Vec4* A, Vec4* B, Vec4* C, Vec4* out);
+
+void ssrS_lerpnum(float t, float A, float B, float* out);
 void ssrS_lerpcolor(float t, Color A, Color B, Color* out);
 void ssrS_lerpvec2(float t, Vec2* A, Vec2* B, Vec2* out);
 void ssrS_lerpvec3(float t, Vec3* A, Vec3* B, Vec3* out);
 void ssrS_lerpvec4(float t, Vec4* A, Vec4* B, Vec4* out);
 
+void ssrS_setregisters(uint flag, int capacity);
+
 /*
-** 顶点数据外，还提供额外的寄存器
+** 顶点数据外，还提供额外的寄存器用来存储vertex shader输出的顶点额外数据
+** 共24个
 */
 enum RegisterType {
-	REGTYPE_VEC2 = 1,
-	REGTYPE_VEC4 = 2,
-	REGTYPE_VEC3 = 3,
+	REGTYPE_NUM = 1,
+	REGTYPE_VEC2,
+	REGTYPE_VEC4,
+	REGTYPE_VEC3,
 };
 
 typedef struct Register {
 	uint length;
 	uint type;
 	union {
-		Vec4* v4;
+		float* num;
 		Vec2* v2;
+		Vec4* v4;
 		Vec3* v3;
 	};
 } Register;
 
-void ssrS_setregvec4_01(Vert* vert, Vec4* value);
-void ssrS_setregvec4_02(Vert* vert, Vec4* value);
-void ssrS_setregvec2_01(Vert* vert, Vec2* value);
-void ssrS_setregvec2_02(Vert* vert, Vec2* value);
-void ssrS_setregvec3_01(Vert* vert, Vec3* value);
-void ssrS_setregvec3_02(Vert* vert, Vec3* value);
-
-Vec4* ssrS_getregvec4_01(Vert* vert);
-Vec4* ssrS_getregvec4_02(Vert* vert);
-Vec2* ssrS_getregvec2_01(Vert* vert);
-Vec2* ssrS_getregvec2_02(Vert* vert);
-Vec3* ssrS_getregvec3_01(Vert* vert);
-Vec3* ssrS_getregvec3_02(Vert* vert);
+#define REG_NUM_COUNT 4
+#define REG_V2_COUNT 6 
+#define REG_V3_COUNT 8
+#define REG_V4_COUNT 6
+
+Register reg_num[4];
+Register reg_v2[6];
+Register reg_v3[8];
+Register reg_v4[6];
+
+void reg_all_wipe(); /*标记清空所有寄存器*/
+
+/*寄存器指针，使用错误可能会出现野指针*/
+
+float* reg_num_00;
+float* reg_num_01;
+float* reg_num_02;
+float* reg_num_03;
+
+Vec2* reg_v2_00;
+Vec2* reg_v2_01;
+Vec2* reg_v2_02;
+Vec2* reg_v2_03;
+Vec2* reg_v2_04;
+Vec2* reg_v2_05;
+
+Vec3* reg_v3_00;
+Vec3* reg_v3_01;
+Vec3* reg_v3_02;
+Vec3* reg_v3_03;
+Vec3* reg_v3_04;
+Vec3* reg_v3_05;
+Vec3* reg_v3_06;
+Vec3* reg_v3_07;
+
+Vec4* reg_v4_00;
+Vec4* reg_v4_01;
+Vec4* reg_v4_02;
+Vec4* reg_v4_03;
+Vec4* reg_v4_04;
+Vec4* reg_v4_05;
+
+/*设置寄存器指针，指向寄存器（注意在fragment阶段，指针会指向fragmentIn结构）*/
+void ssrS_setupregisterpoints(uint extra_varying_flag, int idx);
+
+/*设置寄存器指针，指向fragIn结构*/
+void ssrS_setregtofragin(uint extra_varying_flag, FragmentShaderIn* frag_in);
 
 #endif
\ No newline at end of file
diff --git a/src/core/tris.c b/src/core/tris.c
deleted file mode 100644
index e69de29..0000000
--- a/src/core/tris.c
+++ /dev/null
diff --git a/src/core/tris.h b/src/core/tris.h
deleted file mode 100644
index 1f4208f..0000000
--- a/src/core/tris.h
+++ /dev/null
@@ -1,17 +0,0 @@
-#ifndef _SOFTSHADEROOM_TRIS_H_
-#define _SOFTSHADEROOM_TRIS_H_
-
-#include "../util/type.h"
-
-/* 用索引表示的三角形，在一系列顶点中索引 */
-typedef union Triangle
-{
-	uint v[3];
-	struct {
-		uint v1, v2, v3;
-	};
-} Triangle;
-
-
-
-#endif
\ No newline at end of file
diff --git a/src/core/vert.c b/src/core/vert.c
index 0e62087..95469d3 100644
--- a/src/core/vert.c
+++ b/src/core/vert.c
@@ -1,42 +1,16 @@
 #include "vert.h"
 
-Vert* vert_new(uint comp) {
+Vert* vert_new() {
 	Vert* vert = ssrM_new(Vert);
-	vert->comp = comp;
-	vert->position = ssrM_new(Vec3);
-	vert->normal = ssrM_new(Vec3);
-	vert->tangent = ssrM_new(Vec3);
-	vert->uv = ssrM_new(Vec2);
 	return vert;
 }
 
-void vert_init(Vert* v, uint comp) {
+void vert_init(Vert* v) {
 	ssr_assert(v);
-	v->comp = comp;
 	v->index = 0;
-	if ((comp & VERTMASK_POSITION) && !v->position)
-		v->position = ssrM_new(Vec3);
-	else if (!(comp & VERTMASK_POSITION) && v->position)
-		ssrM_free(v->position);
-	if ((comp & VERTMASK_NORMAL) && !v->normal)
-		v->normal = ssrM_new(Vec3);
-	else if (!(comp & VERTMASK_NORMAL) && v->normal)
-		ssrM_free(v->normal);
-	if ((comp & VERTMASK_TANGENT) && !v->tangent)
-		v->tangent = ssrM_new(Vec3);
-	else if (!(comp & VERTMASK_TANGENT) && v->tangent)
-		ssrM_free(v->tangent);
-	if ((comp & VERTMASK_UV) && !v->uv)
-		v->uv = ssrM_new(Vec2);
-	else if (!(comp & VERTMASK_UV) && v->uv)
-		ssrM_free(v->uv);
 }
 
 void vert_free(Vert* v) {
 	ssr_assert(v);
-	if (v->position) ssrM_free(v->position);
-	if (v->normal) ssrM_free(v->normal);
-	if (v->tangent) ssrM_free(v->tangent);
-	if (v->uv) ssrM_free(v->uv);
 	ssrM_free(v);
 }
\ No newline at end of file
diff --git a/src/core/vert.h b/src/core/vert.h
index a2fdb74..0542ded 100644
--- a/src/core/vert.h
+++ b/src/core/vert.h
@@ -7,42 +7,33 @@
 #include "../util/type.h"
 #include "../util/assert.h"
 
-//typedef uint Color; // ARGB by default
-
-typedef uint Color; // ARGB by default
+typedef uint Color; // ARGB
 
 #define COLOR_A(c) ((c >> 24) & 0xff) 
 #define COLOR_R(c) ((c >> 16) & 0xff) 
 #define COLOR_G(c) ((c >> 8) & 0xff) 
 #define COLOR_B(c) (c & 0xff) 
 
-typedef Vec4 Color32; // 4个float的颜色
+typedef Vec4 Color32;
 
 Color color32_tocolor(Color32* c);
 void color_tocolor32(Color c, Color32* out);
 void color32_saturate(Color32* c);
 
-enum VertMask {
-	VERTMASK_POSITION = 1 ,
-	VERTMASK_NORMAL = 1 << 1,
-	VERTMASK_COLOR = 1 << 2,
-	VERTMASK_TANGENT = 1 << 3,
-	VERTMASK_UV = 1 << 4,
-};
-
-/*标准的顶点结构，禁止修改*/
+/*readonly*/
 typedef struct Vert {
-	uint index;
-	uint comp;
-	Vec3* position;
-	Vec3* normal;
-	Vec3* tangent;
-	Vec2* uv;
+	int index;
+	Vec3 position;
+	Vec3 normal;
+	Vec3 tangent;
+	Vec2 texcoord;
 	Color color;
+	Vec4 joint;
+	Vec4 weight;
 } Vert;
 
-Vert* vert_new(uint comp); /*新建vert*/
-void vert_init(Vert* v, uint comp); /*如果某些component没创建，创建它*/
-void vert_free(Vert* v); /*释放vert*/
+Vert* vert_new(uint comp);
+void vert_init(Vert* v, uint comp); 
+void vert_free(Vert* v); 
 
 #endif
\ No newline at end of file