// // convolution - 5x5 convolution kernel matrix calculation // // written by Olaf Matthes // // 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 2 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, write to the Free Software // Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA // // usage: convolution // // // // // // // KNOWN BUG: produces ugly colors (due to rounding error I guess) // #include #include #include "plugin.h" INFO("5x5 convolution kernel matrix calculation") void perform_effect(struct frame f, struct args a) { short x, y; long matrix[25], scale, shift; pixel16 *p16[5], *c16; pixel24 *p24[5], *c24; pixel32 *p32[5], *c32; long r, g, b; char *t; // get params if(!a.s) return; matrix[0] = atoi(a.s); if(!(t = strstr(a.s, " "))) return; matrix[1] = atoi(t+1); if(!(t = strstr(t+1, " "))) return; matrix[2] = atoi(t+1); if(!(t = strstr(t+1, " "))) return; matrix[3] = atoi(t+1); if(!(t = strstr(t+1, " "))) return; matrix[4] = atoi(t+1); if(!(t = strstr(t+1, " "))) return; matrix[5] = atoi(t+1); if(!(t = strstr(t+1, " "))) return; matrix[6] = atoi(t+1); if(!(t = strstr(t+1, " "))) return; matrix[7] = atoi(t+1); if(!(t = strstr(t+1, " "))) return; matrix[8] = atoi(t+1); if(!(t = strstr(t+1, " "))) return; matrix[9] = atoi(t+1); if(!(t = strstr(t+1, " "))) return; matrix[10] = atoi(t+1); if(!(t = strstr(t+1, " "))) return; matrix[11] = atoi(t+1); if(!(t = strstr(t+1, " "))) return; matrix[12] = atoi(t+1); if(!(t = strstr(t+1, " "))) return; matrix[13] = atoi(t+1); if(!(t = strstr(t+1, " "))) return; matrix[14] = atoi(t+1); if(!(t = strstr(t+1, " "))) return; matrix[15] = atoi(t+1); if(!(t = strstr(t+1, " "))) return; matrix[16] = atoi(t+1); if(!(t = strstr(t+1, " "))) return; matrix[17] = atoi(t+1); if(!(t = strstr(t+1, " "))) return; matrix[18] = atoi(t+1); if(!(t = strstr(t+1, " "))) return; matrix[19] = atoi(t+1); if(!(t = strstr(t+1, " "))) return; matrix[20] = atoi(t+1); if(!(t = strstr(t+1, " "))) return; matrix[21] = atoi(t+1); if(!(t = strstr(t+1, " "))) return; matrix[22] = atoi(t+1); if(!(t = strstr(t+1, " "))) return; matrix[23] = atoi(t+1); if(!(t = strstr(t+1, " "))) return; matrix[24] = atoi(t+1); if(!(t = strstr(t+1, " "))) return; scale = atoi(t+1); if(!(t = strstr(t+1, " "))) return; shift = atoi(t+1); if(scale == 0) scale = 1; printf("convolution: matrix %d %d %d %d %d\n", matrix[0], matrix[1], matrix[2], matrix[3], matrix[4]); printf(" %d %d %d %d %d\n", matrix[5], matrix[6], matrix[7], matrix[8], matrix[9]); printf(" %d %d %d %d %d\n", matrix[10], matrix[11], matrix[12], matrix[13], matrix[14]); printf(" %d %d %d %d %d\n", matrix[15], matrix[16], matrix[17], matrix[18], matrix[19]); printf(" %d %d %d %d %d\n", matrix[20], matrix[21], matrix[22], matrix[23], matrix[24]); printf(" scale %d, shift %d\n", scale, shift); switch(f.pixelformat) { case 16: for(y = 0; y < f.height; y++) { if(y > 5) // we have read all 5 lines needed for calculation { c16 = scanline16(f, y - 3); // this will become our newly calculated line for(x = 2; x < f.width - 2; x++) { r = (r16(p16[(y-5)%5][x-2])*matrix[0] + r16(p16[(y-5)%5][x-1])*matrix[1] + r16(p16[(y-5)%5][x])*matrix[2] + r16(p16[(y-5)%5][x+1])*matrix[3] + r16(p16[(y-5)%5][x+2])*matrix[4] + r16(p16[(y-4)%5][x-2])*matrix[5] + r16(p16[(y-4)%5][x-1])*matrix[6] + r16(p16[(y-3)%5][x])*matrix[7] + r16(p16[(y-4)%5][x+1])*matrix[8] + r16(p16[(y-4)%5][x+2])*matrix[9] + r16(p16[(y-3)%5][x-2])*matrix[10] + r16(p16[(y-3)%5][x-1])*matrix[11] + r16(p16[(y-3)%5][x])*matrix[12] + r16(p16[(y-3)%5][x+1])*matrix[13] + r16(p16[(y-3)%5][x+2])*matrix[14] + r16(p16[(y-2)%5][x-2])*matrix[15] + r16(p16[(y-2)%5][x-1])*matrix[16] + r16(p16[(y-2)%5][x])*matrix[17] + r16(p16[(y-2)%5][x+1])*matrix[18] + r16(p16[(y-2)%5][x+2])*matrix[19] + r16(p16[(y-1)%5][x-2])*matrix[20] + r16(p16[(y-1)%5][x-1])*matrix[21] + r16(p16[(y-1)%5][x])*matrix[22] + r16(p16[(y-1)%5][x+1])*matrix[23] + r16(p16[(y-1)%5][x+2])*matrix[24])/scale - shift; g = (g16(p16[(y-5)%5][x-2])*matrix[0] + g16(p16[(y-5)%5][x-1])*matrix[1] + g16(p16[(y-5)%5][x])*matrix[2] + g16(p16[(y-5)%5][x+1])*matrix[3] + g16(p16[(y-5)%5][x+2])*matrix[4] + g16(p16[(y-4)%5][x-2])*matrix[5] + g16(p16[(y-4)%5][x-1])*matrix[6] + g16(p16[(y-3)%5][x])*matrix[7] + g16(p16[(y-4)%5][x+1])*matrix[8] + g16(p16[(y-4)%5][x+2])*matrix[9] + g16(p16[(y-3)%5][x-2])*matrix[10] + g16(p16[(y-3)%5][x-1])*matrix[11] + g16(p16[(y-3)%5][x])*matrix[12] + g16(p16[(y-3)%5][x+1])*matrix[13] + g16(p16[(y-3)%5][x+2])*matrix[14] + g16(p16[(y-2)%5][x-2])*matrix[15] + g16(p16[(y-2)%5][x-1])*matrix[16] + g16(p16[(y-2)%5][x])*matrix[17] + g16(p16[(y-2)%5][x+1])*matrix[18] + g16(p16[(y-2)%5][x+2])*matrix[19] + g16(p16[(y-1)%5][x-2])*matrix[20] + g16(p16[(y-1)%5][x-1])*matrix[21] + g16(p16[(y-1)%5][x])*matrix[22] + g16(p16[(y-1)%5][x+1])*matrix[23] + g16(p16[(y-1)%5][x+2])*matrix[24])/scale - shift; b = (b16(p16[(y-5)%5][x-2])*matrix[0] + b16(p16[(y-5)%5][x-1])*matrix[1] + b16(p16[(y-5)%5][x])*matrix[2] + b16(p16[(y-5)%5][x+1])*matrix[3] + b16(p16[(y-5)%5][x+2])*matrix[4] + b16(p16[(y-4)%5][x-2])*matrix[5] + b16(p16[(y-4)%5][x-1])*matrix[6] + b16(p16[(y-3)%5][x])*matrix[7] + b16(p16[(y-4)%5][x+1])*matrix[8] + b16(p16[(y-4)%5][x+2])*matrix[9] + b16(p16[(y-3)%5][x-2])*matrix[10] + b16(p16[(y-3)%5][x-1])*matrix[11] + b16(p16[(y-3)%5][x])*matrix[12] + b16(p16[(y-3)%5][x+1])*matrix[13] + b16(p16[(y-3)%5][x+2])*matrix[14] + b16(p16[(y-2)%5][x-2])*matrix[15] + b16(p16[(y-2)%5][x-1])*matrix[16] + b16(p16[(y-2)%5][x])*matrix[17] + b16(p16[(y-2)%5][x+1])*matrix[18] + b16(p16[(y-2)%5][x+2])*matrix[19] + b16(p16[(y-1)%5][x-2])*matrix[20] + b16(p16[(y-1)%5][x-1])*matrix[21] + b16(p16[(y-1)%5][x])*matrix[22] + b16(p16[(y-1)%5][x+1])*matrix[23] + b16(p16[(y-1)%5][x+2])*matrix[24])/scale - shift; c16[x] = rgbtocolor16(klamp255(r), klamp255(g), klamp255(b)); } } // read in lines needed for calculation p16[y % 5] = scanline16(f, y); } break; case 24: for(y=0; y 5) // we have read all 5 lines needed for calculation { c24 = scanline24(f, y - 3); // this will become our newly calculated line for(x = 2; x < f.width - 2; x++) { r = (r24(p24[(y-5)%5][x-2])*matrix[0] + r24(p24[(y-5)%5][x-1])*matrix[1] + r24(p24[(y-5)%5][x])*matrix[2] + r24(p24[(y-5)%5][x+1])*matrix[3] + r24(p24[(y-5)%5][x+2])*matrix[4] + r24(p24[(y-4)%5][x-2])*matrix[5] + r24(p24[(y-4)%5][x-1])*matrix[6] + r24(p24[(y-3)%5][x])*matrix[7] + r24(p24[(y-4)%5][x+1])*matrix[8] + r24(p24[(y-4)%5][x+2])*matrix[9] + r24(p24[(y-3)%5][x-2])*matrix[10] + r24(p24[(y-3)%5][x-1])*matrix[11] + r24(p24[(y-3)%5][x])*matrix[12] + r24(p24[(y-3)%5][x+1])*matrix[13] + r24(p24[(y-3)%5][x+2])*matrix[14] + r24(p24[(y-2)%5][x-2])*matrix[15] + r24(p24[(y-2)%5][x-1])*matrix[16] + r24(p24[(y-2)%5][x])*matrix[17] + r24(p24[(y-2)%5][x+1])*matrix[18] + r24(p24[(y-2)%5][x+2])*matrix[19] + r24(p24[(y-1)%5][x-2])*matrix[20] + r24(p24[(y-1)%5][x-1])*matrix[21] + r24(p24[(y-1)%5][x])*matrix[22] + r24(p24[(y-1)%5][x+1])*matrix[23] + r24(p24[(y-1)%5][x+2])*matrix[24])/scale - shift; g = (g24(p24[(y-5)%5][x-2])*matrix[0] + g24(p24[(y-5)%5][x-1])*matrix[1] + g24(p24[(y-5)%5][x])*matrix[2] + g24(p24[(y-5)%5][x+1])*matrix[3] + g24(p24[(y-5)%5][x+2])*matrix[4] + g24(p24[(y-4)%5][x-2])*matrix[5] + g24(p24[(y-4)%5][x-1])*matrix[6] + g24(p24[(y-3)%5][x])*matrix[7] + g24(p24[(y-4)%5][x+1])*matrix[8] + g24(p24[(y-4)%5][x+2])*matrix[9] + g24(p24[(y-3)%5][x-2])*matrix[10] + g24(p24[(y-3)%5][x-1])*matrix[11] + g24(p24[(y-3)%5][x])*matrix[12] + g24(p24[(y-3)%5][x+1])*matrix[13] + g24(p24[(y-3)%5][x+2])*matrix[14] + g24(p24[(y-2)%5][x-2])*matrix[15] + g24(p24[(y-2)%5][x-1])*matrix[16] + g24(p24[(y-2)%5][x])*matrix[17] + g24(p24[(y-2)%5][x+1])*matrix[18] + g24(p24[(y-2)%5][x+2])*matrix[19] + g24(p24[(y-1)%5][x-2])*matrix[20] + g24(p24[(y-1)%5][x-1])*matrix[21] + g24(p24[(y-1)%5][x])*matrix[22] + g24(p24[(y-1)%5][x+1])*matrix[23] + g24(p24[(y-1)%5][x+2])*matrix[24])/scale - shift; b = (b24(p24[(y-5)%5][x-2])*matrix[0] + b24(p24[(y-5)%5][x-1])*matrix[1] + b24(p24[(y-5)%5][x])*matrix[2] + b24(p24[(y-5)%5][x+1])*matrix[3] + b24(p24[(y-5)%5][x+2])*matrix[4] + b24(p24[(y-4)%5][x-2])*matrix[5] + b24(p24[(y-4)%5][x-1])*matrix[6] + b24(p24[(y-3)%5][x])*matrix[7] + b24(p24[(y-4)%5][x+1])*matrix[8] + b24(p24[(y-4)%5][x+2])*matrix[9] + b24(p24[(y-3)%5][x-2])*matrix[10] + b24(p24[(y-3)%5][x-1])*matrix[11] + b24(p24[(y-3)%5][x])*matrix[12] + b24(p24[(y-3)%5][x+1])*matrix[13] + b24(p24[(y-3)%5][x+2])*matrix[14] + b24(p24[(y-2)%5][x-2])*matrix[15] + b24(p24[(y-2)%5][x-1])*matrix[16] + b24(p24[(y-2)%5][x])*matrix[17] + b24(p24[(y-2)%5][x+1])*matrix[18] + b24(p24[(y-2)%5][x+2])*matrix[19] + b24(p24[(y-1)%5][x-2])*matrix[20] + b24(p24[(y-1)%5][x-1])*matrix[21] + b24(p24[(y-1)%5][x])*matrix[22] + b24(p24[(y-1)%5][x+1])*matrix[23] + b24(p24[(y-1)%5][x+2])*matrix[24])/scale - shift; c24[x] = rgbtocolor24(klamp255(r), klamp255(g), klamp255(b)); } } // read in lines needed for calculation p24[y % 5] = scanline24(f, y); } break; case 32: for(y=0; y 5) // we have read all 5 lines needed for calculation { c32 = scanline32(f, y - 3); // this will become our newly calculated line for(x = 2; x < f.width - 2; x++) { r = (r32(p32[(y-5)%5][x-2])*matrix[0] + r32(p32[(y-5)%5][x-1])*matrix[1] + r32(p32[(y-5)%5][x])*matrix[2] + r32(p32[(y-5)%5][x+1])*matrix[3] + r32(p32[(y-5)%5][x+2])*matrix[4] + r32(p32[(y-4)%5][x-2])*matrix[5] + r32(p32[(y-4)%5][x-1])*matrix[6] + r32(p32[(y-3)%5][x])*matrix[7] + r32(p32[(y-4)%5][x+1])*matrix[8] + r32(p32[(y-4)%5][x+2])*matrix[9] + r32(p32[(y-3)%5][x-2])*matrix[10] + r32(p32[(y-3)%5][x-1])*matrix[11] + r32(p32[(y-3)%5][x])*matrix[12] + r32(p32[(y-3)%5][x+1])*matrix[13] + r32(p32[(y-3)%5][x+2])*matrix[14] + r32(p32[(y-2)%5][x-2])*matrix[15] + r32(p32[(y-2)%5][x-1])*matrix[16] + r32(p32[(y-2)%5][x])*matrix[17] + r32(p32[(y-2)%5][x+1])*matrix[18] + r32(p32[(y-2)%5][x+2])*matrix[19] + r32(p32[(y-1)%5][x-2])*matrix[20] + r32(p32[(y-1)%5][x-1])*matrix[21] + r32(p32[(y-1)%5][x])*matrix[22] + r32(p32[(y-1)%5][x+1])*matrix[23] + r32(p32[(y-1)%5][x+2])*matrix[24])/scale - shift; g = (g32(p32[(y-5)%5][x-2])*matrix[0] + g32(p32[(y-5)%5][x-1])*matrix[1] + g32(p32[(y-5)%5][x])*matrix[2] + g32(p32[(y-5)%5][x+1])*matrix[3] + g32(p32[(y-5)%5][x+2])*matrix[4] + g32(p32[(y-4)%5][x-2])*matrix[5] + g32(p32[(y-4)%5][x-1])*matrix[6] + g32(p32[(y-3)%5][x])*matrix[7] + g32(p32[(y-4)%5][x+1])*matrix[8] + g32(p32[(y-4)%5][x+2])*matrix[9] + g32(p32[(y-3)%5][x-2])*matrix[10] + g32(p32[(y-3)%5][x-1])*matrix[11] + g32(p32[(y-3)%5][x])*matrix[12] + g32(p32[(y-3)%5][x+1])*matrix[13] + g32(p32[(y-3)%5][x+2])*matrix[14] + g32(p32[(y-2)%5][x-2])*matrix[15] + g32(p32[(y-2)%5][x-1])*matrix[16] + g32(p32[(y-2)%5][x])*matrix[17] + g32(p32[(y-2)%5][x+1])*matrix[18] + g32(p32[(y-2)%5][x+2])*matrix[19] + g32(p32[(y-1)%5][x-2])*matrix[20] + g32(p32[(y-1)%5][x-1])*matrix[21] + g32(p32[(y-1)%5][x])*matrix[22] + g32(p32[(y-1)%5][x+1])*matrix[23] + g32(p32[(y-1)%5][x+2])*matrix[24])/scale - shift; b = (b32(p32[(y-5)%5][x-2])*matrix[0] + b32(p32[(y-5)%5][x-1])*matrix[1] + b32(p32[(y-5)%5][x])*matrix[2] + b32(p32[(y-5)%5][x+1])*matrix[3] + b32(p32[(y-5)%5][x+2])*matrix[4] + b32(p32[(y-4)%5][x-2])*matrix[5] + b32(p32[(y-4)%5][x-1])*matrix[6] + b32(p32[(y-3)%5][x])*matrix[7] + b32(p32[(y-4)%5][x+1])*matrix[8] + b32(p32[(y-4)%5][x+2])*matrix[9] + b32(p32[(y-3)%5][x-2])*matrix[10] + b32(p32[(y-3)%5][x-1])*matrix[11] + b32(p32[(y-3)%5][x])*matrix[12] + b32(p32[(y-3)%5][x+1])*matrix[13] + b32(p32[(y-3)%5][x+2])*matrix[14] + b32(p32[(y-2)%5][x-2])*matrix[15] + b32(p32[(y-2)%5][x-1])*matrix[16] + b32(p32[(y-2)%5][x])*matrix[17] + b32(p32[(y-2)%5][x+1])*matrix[18] + b32(p32[(y-2)%5][x+2])*matrix[19] + b32(p32[(y-1)%5][x-2])*matrix[20] + b32(p32[(y-1)%5][x-1])*matrix[21] + b32(p32[(y-1)%5][x])*matrix[22] + b32(p32[(y-1)%5][x+1])*matrix[23] + b32(p32[(y-1)%5][x+2])*matrix[24])/scale - shift; c32[x] = rgbtocolor32(klamp255(r), klamp255(g), klamp255(b)); } } // read in lines needed for calculation p32[y % 5] = scanline32(f, y); } break; } } void perform_copy(struct frame f1, struct frame f2, struct args a) { printf("Using convolution as copy operation does nothing!\n"); }