nuclear@9: #include "config.h"
nuclear@14: #include <stdio.h>
nuclear@19: #include <stdlib.h>
nuclear@7: #include <string.h>
nuclear@12: #include <math.h>
nuclear@7: #include "x3d.h"
nuclear@7: #include "fixed.h"
nuclear@7: #include "sincos.h"
nuclear@8: #include "logger.h"
nuclear@8: #include "polyfill.h"
nuclear@8: #include "gbasys.h"
nuclear@17: #include "x3d_impl.h"
nuclear@7: 
nuclear@14: int dbg_fill_dump;
nuclear@14: 
nuclear@7: #define MAT_STACK_SIZE	4
nuclear@7: 
nuclear@7: struct matrix {
nuclear@8: 	int32_t m[12];
nuclear@7: };
nuclear@7: 
nuclear@15: static void proc_vertex(const int32_t *vin, const int32_t *cin, const int32_t *tin,
nuclear@15: 		pvec3 *vout, pvec3 *cout, pvec2 *tout);
nuclear@14: static int dump_frame(struct pixel_buffer *frame);
nuclear@8: 
nuclear@8: 
nuclear@7: static int32_t proj_fov = M_PI_X16;
nuclear@7: static int32_t proj_aspect = 65536;
nuclear@13: static int32_t inv_proj_aspect = 65536;
nuclear@7: static int32_t proj_near = ftox16(0.5);
nuclear@7: static int32_t proj_far = 500 << 16;
nuclear@13: static int32_t inv_tan_half_xfov, inv_tan_half_yfov;
nuclear@7: 
nuclear@8: #define ID_INIT {65536, 0, 0, 0, 0, 65536, 0, 0, 0, 0, 65536, 0}
nuclear@8: 
nuclear@8: static struct matrix identity = { ID_INIT };
nuclear@7: 
nuclear@7: static short mtop;
nuclear@8: static struct matrix mstack[MAT_STACK_SIZE] = { {ID_INIT}, {ID_INIT} };
nuclear@8: 
nuclear@8: static const int32_t *vertex_array;
nuclear@8: static unsigned short vertex_count;
nuclear@8: static const int32_t *color_array;
nuclear@8: static unsigned short color_count;
nuclear@15: static const int32_t *texcoord_array;
nuclear@15: static unsigned short texcoord_count;
nuclear@8: 
nuclear@8: static int32_t im_color[3];
nuclear@15: static int32_t im_texcoord[2];
nuclear@9: static uint8_t im_color_index;
nuclear@7: 
nuclear@17: #define MAX_TEXTURES	64
nuclear@17: static struct texture textures[MAX_TEXTURES];
nuclear@17: static short cur_tex = -1;
nuclear@17: 
nuclear@17: 
nuclear@12: void x3d_projection(int fov, int32_t aspect, int32_t nearz, int32_t farz)
nuclear@7: {
nuclear@12: 	proj_fov = (M_PI_X16 * fov) / 180;
nuclear@7: 	proj_aspect = aspect;
nuclear@13: 	inv_proj_aspect = x16div(65536, proj_aspect);
nuclear@7: 	proj_near = nearz;
nuclear@7: 	proj_far = farz;
nuclear@12: 
nuclear@13: 	inv_tan_half_yfov = (int32_t)(65536.0 / tan(0.5 * proj_fov / 65536.0));
nuclear@13: 	inv_tan_half_xfov = x16mul(inv_tan_half_yfov, aspect);
nuclear@7: }
nuclear@7: 
nuclear@7: int x3d_push_matrix(void)
nuclear@7: {
nuclear@7: 	short newtop = mtop + 1;
nuclear@7: 	if(newtop >= MAT_STACK_SIZE) {
nuclear@7: 		return -1;
nuclear@7: 	}
nuclear@7: 	memcpy(mstack + newtop, mstack + mtop, sizeof *mstack);
nuclear@7: 	mtop = newtop;
nuclear@7: 	return 0;
nuclear@7: }
nuclear@7: 
nuclear@7: int x3d_pop_matrix(void)
nuclear@7: {
nuclear@7: 	if(mtop <= 0) {
nuclear@7: 		return -1;
nuclear@7: 	}
nuclear@7: 	--mtop;
nuclear@7: 	return 0;
nuclear@7: }
nuclear@7: 
nuclear@7: void x3d_load_matrix(int32_t *m)
nuclear@7: {
nuclear@8: 	memcpy(mstack[mtop].m, m, sizeof *mstack);
nuclear@7: }
nuclear@7: 
nuclear@7: 
nuclear@7: #define M(i,j)	(((i) << 2) + (j))
nuclear@7: void x3d_mult_matrix(int32_t *m)
nuclear@7: {
nuclear@7: 	int i, j;
nuclear@7: 	struct matrix tmp;
nuclear@7: 
nuclear@8: 	memcpy(tmp.m, mstack[mtop].m, sizeof tmp);
nuclear@7: 
nuclear@7: 	for(i=0; i<3; i++) {
nuclear@7: 		for(j=0; j<4; j++) {
nuclear@8: 			mstack[mtop].m[M(i, j)] =
nuclear@14: 				x16mul(m[M(0, j)], tmp.m[M(i, 0)]) +
nuclear@14: 				x16mul(m[M(1, j)], tmp.m[M(i, 1)]) +
nuclear@14: 				x16mul(m[M(2, j)], tmp.m[M(i, 2)]);
nuclear@7: 		}
nuclear@14: 		mstack[mtop].m[M(i, 3)] += tmp.m[M(i, 3)];
nuclear@7: 	}
nuclear@7: }
nuclear@7: 
nuclear@7: void x3d_load_identity(void)
nuclear@7: {
nuclear@8: 	memcpy(mstack[mtop].m, identity.m, sizeof identity);
nuclear@7: }
nuclear@7: 
nuclear@8: void x3d_translate(int32_t x, int32_t y, int32_t z)
nuclear@8: {
nuclear@8: 	int32_t m[] = ID_INIT;
nuclear@8: 	m[3] = x;
nuclear@8: 	m[7] = y;
nuclear@8: 	m[11] = z;
nuclear@8: 
nuclear@8: 	x3d_mult_matrix(m);
nuclear@8: }
nuclear@8: 
nuclear@8: void x3d_rotate(int32_t deg, int32_t x, int32_t y, int32_t z)
nuclear@8: {
nuclear@8: 	int32_t xform[] = ID_INIT;
nuclear@8: 
nuclear@8: 	int32_t angle = x16mul(M_PI_X16, deg) / 180;
nuclear@8: 	int32_t sina = sin_x16(angle);
nuclear@8: 	int32_t cosa = cos_x16(angle);
nuclear@8: 	int32_t one_minus_cosa = 65536 - cosa;
nuclear@8: 	int32_t nxsq = x16sq(x);
nuclear@8: 	int32_t nysq = x16sq(y);
nuclear@8: 	int32_t nzsq = x16sq(z);
nuclear@8: 
nuclear@8: 	xform[0] = nxsq + x16mul(65536 - nxsq, cosa);
nuclear@8: 	xform[4] = x16mul(x16mul(x, y), one_minus_cosa) - x16mul(z, sina);
nuclear@8: 	xform[8] = x16mul(x16mul(x, z), one_minus_cosa) + x16mul(y, sina);
nuclear@8: 	xform[1] = x16mul(x16mul(x, y), one_minus_cosa) + x16mul(z, sina);
nuclear@8: 	xform[5] = nysq + x16mul(65536 - nysq, cosa);
nuclear@8: 	xform[9] = x16mul(x16mul(y, z), one_minus_cosa) - x16mul(x, sina);
nuclear@8: 	xform[2] = x16mul(x16mul(x, z), one_minus_cosa) - x16mul(y, sina);
nuclear@8: 	xform[6] = x16mul(x16mul(y, z), one_minus_cosa) + x16mul(x, sina);
nuclear@8: 	xform[10] = nzsq + x16mul(65536 - nzsq, cosa);
nuclear@8: 
nuclear@8: 	x3d_mult_matrix(xform);
nuclear@8: }
nuclear@8: 
nuclear@8: void x3d_scale(int32_t x, int32_t y, int32_t z)
nuclear@8: {
nuclear@8: 	int32_t m[] = ID_INIT;
nuclear@8: 
nuclear@8: 	m[0] = x;
nuclear@8: 	m[5] = y;
nuclear@8: 	m[10] = z;
nuclear@8: 
nuclear@8: 	x3d_mult_matrix(m);
nuclear@8: }
nuclear@8: 
nuclear@8: void x3d_vertex_array(int count, const int32_t *ptr)
nuclear@8: {
nuclear@8: 	vertex_array = ptr;
nuclear@8: 	vertex_count = count;
nuclear@8: }
nuclear@8: 
nuclear@8: void x3d_color_array(int count, const int32_t *ptr)
nuclear@8: {
nuclear@8: 	color_array = ptr;
nuclear@8: 	color_count = count;
nuclear@8: }
nuclear@8: 
nuclear@15: void x3d_texcoord_array(int count, const int32_t *ptr)
nuclear@15: {
nuclear@15: 	texcoord_array = ptr;
nuclear@15: 	texcoord_count = count;
nuclear@15: }
nuclear@15: 
nuclear@12: int x3d_draw(int prim, int vnum)
nuclear@8: {
nuclear@8: 	int i, j, pverts = prim;
nuclear@8: 	const int32_t *vptr = vertex_array;
nuclear@8: 	const int32_t *cptr = color_array;
nuclear@15: 	const int32_t *tptr = texcoord_array;
nuclear@9: #ifndef PALMODE
nuclear@8: 	short cr, cg, cb;
nuclear@9: #endif
nuclear@9: 	uint16_t color;
nuclear@8: 
nuclear@8: 	if(!vertex_array) return -1;
nuclear@8: 
nuclear@8: 	if(vnum > vertex_count) {
nuclear@8: 		logmsg(LOG_DBG, "%s called with vnum=%d, but current vertex array has %d vertices\n",
nuclear@8: 				__FUNCTION__, vnum, vertex_count);
nuclear@8: 		vnum = vertex_count;
nuclear@8: 	}
nuclear@8: 	if(color_array && vnum > color_count) {
nuclear@8: 		logmsg(LOG_DBG, "%s called with vnum=%d, but current color array has %d elements\n",
nuclear@8: 				__FUNCTION__, vnum, color_count);
nuclear@8: 		vnum = color_count;
nuclear@8: 	}
nuclear@15: 	if(texcoord_array && vnum > texcoord_count) {
nuclear@15: 		logmsg(LOG_DBG, "%s called with vnum=%d, but current texcoord array has %d elements\n",
nuclear@15: 				__FUNCTION__, vnum, texcoord_count);
nuclear@15: 		vnum = texcoord_count;
nuclear@15: 	}
nuclear@8: 
nuclear@8: 	for(i=0; i<vnum; i+=pverts) {
nuclear@8: 		/* process vertices */
nuclear@8: 		pvec3 vpos[4];
nuclear@8: 		pvec3 col[4];
nuclear@15: 		pvec2 tex[4];
nuclear@8: 
nuclear@8: 		for(j=0; j<pverts; j++) {
nuclear@15: 			proc_vertex(vptr, cptr, tptr, vpos + j, col + j, tex + j);
nuclear@12: 
nuclear@12: 			if(vpos[j].z <= proj_near) {
nuclear@12: 				goto skip_prim;
nuclear@12: 			}
nuclear@12: 
nuclear@8: 			vptr += 3;
nuclear@8: 			if(cptr) cptr += 3;
nuclear@15: 			if(tptr) tptr += 2;
nuclear@8: 		}
nuclear@8: 
nuclear@9: #ifdef PALMODE
nuclear@9: 		color = im_color_index;
nuclear@9: #else
nuclear@8: 		cr = col[0].x >> 8;
nuclear@8: 		cg = col[0].y >> 8;
nuclear@8: 		cb = col[0].z >> 8;
nuclear@8: 
nuclear@8: 		if(cr > 255) cr = 255;
nuclear@8: 		if(cg > 255) cg = 255;
nuclear@8: 		if(cb > 255) cb = 255;
nuclear@8: 
nuclear@9: 		color = RGB(cr, cg, cb);
nuclear@9: #endif
nuclear@9: 
nuclear@12: 		/* project & viewport */
nuclear@12: 		for(j=0; j<pverts; j++) {
nuclear@12: 			int32_t x, y;
nuclear@12: 
nuclear@13: 			x = x16mul(vpos[j].x, inv_tan_half_xfov);
nuclear@12: 			x = x16div(x, vpos[j].z);
nuclear@13: 			vpos[j].x = (x16mul(x, inv_proj_aspect) + 65536) * (WIDTH / 2);
nuclear@12: 
nuclear@13: 			y = x16mul(vpos[j].y, inv_tan_half_yfov);
nuclear@12: 			y = x16div(y, vpos[j].z);
nuclear@12: 			vpos[j].y = (65536 - y) * (HEIGHT / 2);
nuclear@12: 		}
nuclear@12: 
nuclear@8: 		switch(pverts) {
nuclear@8: 		case X3D_POINTS:
nuclear@9: 			draw_point(vpos, color);
nuclear@8: 			break;
nuclear@8: 
nuclear@8: 		case X3D_LINES:
nuclear@8: 			break;
nuclear@8: 
nuclear@8: 		case X3D_TRIANGLES:
nuclear@8: 		case X3D_QUADS:
nuclear@17: 			draw_poly(pverts, vpos, tex, color, cur_tex >= 0 ? textures + cur_tex : 0);
nuclear@14: 			if(dbg_fill_dump) {
nuclear@14: 				dump_frame(back_buffer);
nuclear@14: 			}
nuclear@8: 			break;
nuclear@8: 		}
nuclear@12: skip_prim: ;
nuclear@8: 	}
nuclear@14: 
nuclear@14: 	dbg_fill_dump = 0;
nuclear@8: 	return 0;
nuclear@8: }
nuclear@8: 
nuclear@15: static void proc_vertex(const int32_t *vin, const int32_t *cin, const int32_t *tin,
nuclear@15: 		pvec3 *vout, pvec3 *cout, pvec2 *tout)
nuclear@8: {
nuclear@8: 	int i;
nuclear@8: 	int32_t tvert[3];
nuclear@8: 	int32_t *mvmat = mstack[mtop].m;
nuclear@8: 
nuclear@8: 	/* transform vertex with current matrix */
nuclear@8: 	for(i=0; i<3; i++) {
nuclear@8: 		tvert[i] = x16mul(mvmat[0], vin[0]) +
nuclear@8: 			x16mul(mvmat[1], vin[1]) +
nuclear@8: 			x16mul(mvmat[2], vin[2]) +
nuclear@8: 			mvmat[3];
nuclear@8: 		mvmat += 4;
nuclear@8: 	}
nuclear@8: 
nuclear@8: 	vout->x = tvert[0];
nuclear@8: 	vout->y = tvert[1];
nuclear@8: 	vout->z = tvert[2];
nuclear@8: 	/*logmsg(LOG_DBG, "%s: (%g %g %g) -> (%g %g %g)\n", __FUNCTION__,
nuclear@8: 			x16tof(vin[0]), x16tof(vin[1]), x16tof(vin[2]),
nuclear@8: 			x16tof(vout->x), x16tof(vout->y), x16tof(vout->z));*/
nuclear@8: 
nuclear@8: 	if(color_array) {
nuclear@8: 		cout->x = cin[0];
nuclear@8: 		cout->y = cin[1];
nuclear@8: 		cout->z = cin[2];
nuclear@8: 	} else {
nuclear@8: 		cout->x = im_color[0];
nuclear@8: 		cout->y = im_color[1];
nuclear@8: 		cout->z = im_color[2];
nuclear@8: 	}
nuclear@15: 
nuclear@15: 	if(texcoord_array) {
nuclear@15: 		tout->x = tin[0];
nuclear@15: 		tout->y = tin[1];
nuclear@15: 	} else {
nuclear@15: 		tout->x = im_texcoord[0];
nuclear@15: 		tout->y = im_texcoord[1];
nuclear@15: 	}
nuclear@8: }
nuclear@8: 
nuclear@9: void x3d_color_index(int cidx)
nuclear@9: {
nuclear@9: 	im_color_index = cidx;
nuclear@9: }
nuclear@9: 
nuclear@8: void x3d_color(int32_t r, int32_t g, int32_t b)
nuclear@8: {
nuclear@8: 	im_color[0] = r;
nuclear@8: 	im_color[1] = g;
nuclear@8: 	im_color[2] = b;
nuclear@8: }
nuclear@14: 
nuclear@17: static int count_bits(int x)
nuclear@17: {
nuclear@17: 	int i, count = 0;
nuclear@17: 	for(i=0; i<32; i++) {
nuclear@17: 		if(x & 1) count++;
nuclear@17: 		x >>= 1;
nuclear@17: 	}
nuclear@17: 	return count;
nuclear@17: }
nuclear@17: 
nuclear@17: int x3d_create_texture_rgb(int xsz, int ysz, const uint16_t *pixels)
nuclear@17: {
nuclear@17: 	int i, j;
nuclear@17: 
nuclear@17: 	if(xsz == 0 || count_bits(xsz) > 1 || ysz == 0 || count_bits(ysz) > 1) {
nuclear@17: 		logmsg(LOG_DBG, "%s: texture size (%dx%d) not power of two!\n", __func__, xsz, ysz);
nuclear@17: 		return -1;
nuclear@17: 	}
nuclear@17: 
nuclear@17: 	for(i=0; i<MAX_TEXTURES; i++) {
nuclear@17: 		if(!textures[i].pixels) {
nuclear@19: 			/*textures[i].pixels = pixels;*/
nuclear@19: 			textures[i].pixels = malloc(xsz * ysz * 2);
nuclear@19: 			memcpy(textures[i].pixels, pixels, xsz * ysz * 2);
nuclear@17: 			textures[i].xsz = xsz;
nuclear@17: 			textures[i].ysz = ysz;
nuclear@17: 			textures[i].umask = xsz - 1;
nuclear@17: 			textures[i].vmask = ysz - 1;
nuclear@17: 
nuclear@17: 			for(j=0; j<32; j++) {
nuclear@17: 				if((1 << j) == xsz) {
nuclear@17: 					textures[i].ushift = j;
nuclear@17: 				}
nuclear@17: 				if((1 << j) == ysz) {
nuclear@17: 					textures[i].vshift = j;
nuclear@17: 				}
nuclear@17: 			}
nuclear@17: 
nuclear@19: 			logmsg(LOG_DBG, "create texture %dx%d: %p\n", xsz, ysz, pixels);
nuclear@19: 
nuclear@17: 			return i;
nuclear@17: 		}
nuclear@17: 	}
nuclear@17: 	return -1;
nuclear@17: }
nuclear@17: 
nuclear@17: void x3d_enable_texture(int texid)
nuclear@17: {
nuclear@17: 	cur_tex = texid;
nuclear@17: }
nuclear@17: 
nuclear@17: void x3d_disable_texture(void)
nuclear@17: {
nuclear@17: 	cur_tex = 0;
nuclear@17: }
nuclear@17: 
nuclear@17: int x3d_get_active_texture(void)
nuclear@17: {
nuclear@17: 	return cur_tex;
nuclear@17: }
nuclear@17: 
nuclear@14: static int dump_frame(struct pixel_buffer *frame)
nuclear@14: {
nuclear@14: 	static int frameno;
nuclear@14: 	char buf[128];
nuclear@14: 	FILE *fp;
nuclear@14: 	int i, npix;
nuclear@14: 	uint16_t *ptr = frame->pixels;
nuclear@14: 
nuclear@14: 	sprintf(buf, "dump%03d.ppm", ++frameno);
nuclear@14: 
nuclear@14: 	if(!(fp = fopen(buf, "wb"))) {
nuclear@14: 		fprintf(stderr, "failed to dump file: %s\n", buf);
nuclear@14: 		return -1;
nuclear@14: 	}
nuclear@14: 
nuclear@14: 	fprintf(fp, "P6\n%d %d\n255\n", frame->x, frame->y);
nuclear@14: 
nuclear@14: 	npix = frame->x * frame->y;
nuclear@14: 	for(i=0; i<npix; i++) {
nuclear@14: 		uint16_t pixel = *ptr++;
nuclear@14: 		fputc(GET_R(pixel), fp);
nuclear@14: 		fputc(GET_G(pixel), fp);
nuclear@14: 		fputc(GET_B(pixel), fp);
nuclear@14: 	}
nuclear@14: 	fclose(fp);
nuclear@14: 	return 0;
nuclear@14: }