+ for (unsigned x = 0; x < 8; ++x) {
+ tf_switch_second_stage_inv(&c[1 * 8 + x], &c[3 * 8 + x], &c[5 * 8 + x], &c[7 * 8 + x]);
+ }
+ for (unsigned y = 0; y < 8; ++y) {
+ tf_switch_second_stage_inv(&c[y * 8 + 1], &c[y * 8 + 3], &c[y * 8 + 5], &c[y * 8 + 7]);
+ }
+ for (unsigned y = 0; y < 4; ++y) {
+ for (unsigned x = 0; x < 4; ++x) {
+ tf_switch(&c[(y*2) * 8 + x*2], &c[(y*2) * 8 + (x*2+1)], &c[(y*2+1)*8 + x*2], &c[(y*2+1)*8 + (x*2+1)]);
+ }
+ }
+ short d[64] = {
+ c[0*8 + 0], c[0*8 + 2], c[0*8 + 4], c[0*8 + 6], c[0*8 + 1], c[0*8 + 3], c[0*8 + 5], c[0*8 + 7],
+ c[2*8 + 0], c[2*8 + 2], c[2*8 + 4], c[2*8 + 6], c[2*8 + 1], c[2*8 + 3], c[2*8 + 5], c[2*8 + 7],
+ c[4*8 + 0], c[4*8 + 2], c[4*8 + 4], c[4*8 + 6], c[4*8 + 1], c[4*8 + 3], c[4*8 + 5], c[4*8 + 7],
+ c[6*8 + 0], c[6*8 + 2], c[6*8 + 4], c[6*8 + 6], c[6*8 + 1], c[6*8 + 3], c[6*8 + 5], c[6*8 + 7],
+ c[1*8 + 0], c[1*8 + 2], c[1*8 + 4], c[1*8 + 6], c[1*8 + 1], c[1*8 + 3], c[1*8 + 5], c[1*8 + 7],
+ c[3*8 + 0], c[3*8 + 2], c[3*8 + 4], c[3*8 + 6], c[3*8 + 1], c[3*8 + 3], c[3*8 + 5], c[3*8 + 7],
+ c[5*8 + 0], c[5*8 + 2], c[5*8 + 4], c[5*8 + 6], c[5*8 + 1], c[5*8 + 3], c[5*8 + 5], c[5*8 + 7],
+ c[7*8 + 0], c[7*8 + 2], c[7*8 + 4], c[7*8 + 6], c[7*8 + 1], c[7*8 + 3], c[7*8 + 5], c[7*8 + 7]
+ };
+ memcpy(c, d, sizeof(d));
+}
+
+static void convert_4x4to8x8(short *c)
+{
+ short d[64] = {
+ c[0*8 + 0], c[0*8 + 4], c[0*8 + 1], c[0*8 + 5], c[0*8 + 2], c[0*8 + 6], c[0*8 + 3], c[0*8 + 7],
+ c[4*8 + 0], c[4*8 + 4], c[4*8 + 1], c[4*8 + 5], c[4*8 + 2], c[4*8 + 6], c[4*8 + 3], c[4*8 + 7],
+ c[1*8 + 0], c[1*8 + 4], c[1*8 + 1], c[1*8 + 5], c[1*8 + 2], c[1*8 + 6], c[1*8 + 3], c[1*8 + 7],
+ c[5*8 + 0], c[5*8 + 4], c[5*8 + 1], c[5*8 + 5], c[5*8 + 2], c[5*8 + 6], c[5*8 + 3], c[5*8 + 7],
+ c[2*8 + 0], c[2*8 + 4], c[2*8 + 1], c[2*8 + 5], c[2*8 + 2], c[2*8 + 6], c[2*8 + 3], c[2*8 + 7],
+ c[6*8 + 0], c[6*8 + 4], c[6*8 + 1], c[6*8 + 5], c[6*8 + 2], c[6*8 + 6], c[6*8 + 3], c[6*8 + 7],
+ c[3*8 + 0], c[3*8 + 4], c[3*8 + 1], c[3*8 + 5], c[3*8 + 2], c[3*8 + 6], c[3*8 + 3], c[3*8 + 7],
+ c[7*8 + 0], c[7*8 + 4], c[7*8 + 1], c[7*8 + 5], c[7*8 + 2], c[7*8 + 6], c[7*8 + 3], c[7*8 + 7]
+ };
+
+ for (unsigned y = 0; y < 4; ++y) {
+ for (unsigned x = 0; x < 4; ++x) {
+ tf_switch(&d[(y*2) * 8 + x*2], &d[(y*2) * 8 + (x*2+1)], &d[(y*2+1)*8 + x*2], &d[(y*2+1)*8 + (x*2+1)]);
+ }
+ }
+ for (unsigned y = 0; y < 8; ++y) {
+ tf_switch_second_stage(&d[y * 8 + 1], &d[y * 8 + 3], &d[y * 8 + 5], &d[y * 8 + 7]);
+ }
+ for (unsigned x = 0; x < 8; ++x) {
+ tf_switch_second_stage(&d[1 * 8 + x], &d[3 * 8 + x], &d[5 * 8 + x], &d[7 * 8 + x]);
+ }
+
+ memcpy(c, d, sizeof(d));
+}
+
+int main(int argc, char **argv)
+{
+ if (argc >= 2) quant_scalefac = atof(argv[1]);
+ if (argc >= 3) lambda = atof(argv[2]);
+