+ __m128i color, mod, pix;
+ if (x + 4 <= endx && *(const unsigned int *)&pixelmask[x] == 0x01010101)
+ {
+ __m128i pix2, mod2;
+ __m128 z = _mm_loadu_ps(&buffer_z[x]);
+ color = _mm_loadu_si128((const __m128i *)&buffer_texture_colorbgra8[x*4]);
+ mod = _mm_cvtps_epi32(_mm_mul_ps(data, _mm_shuffle_ps(z, z, _MM_SHUFFLE(0, 0, 0, 0))));
+ data = _mm_add_ps(data, slope);
+ mod = _mm_packs_epi32(mod, _mm_cvtps_epi32(_mm_mul_ps(data, _mm_shuffle_ps(z, z, _MM_SHUFFLE(1, 1, 1, 1)))));
+ data = _mm_add_ps(data, slope);
+ mod2 = _mm_cvtps_epi32(_mm_mul_ps(data, _mm_shuffle_ps(z, z, _MM_SHUFFLE(2, 2, 2, 2))));
+ data = _mm_add_ps(data, slope);
+ mod2 = _mm_packs_epi32(mod2, _mm_cvtps_epi32(_mm_mul_ps(data, _mm_shuffle_ps(z, z, _MM_SHUFFLE(3, 3, 3, 3)))));
+ pix = _mm_mulhi_epu16(_mm_add_epi16(_mm_mulhi_epu16(Color_Diffusem, mod), Color_Ambientm),
+ _mm_unpacklo_epi8(_mm_setzero_si128(), color));
+ pix2 = _mm_mulhi_epu16(_mm_add_epi16(_mm_mulhi_epu16(Color_Diffusem, mod2), Color_Ambientm),
+ _mm_unpackhi_epi8(_mm_setzero_si128(), color));
+ _mm_storeu_si128((__m128i *)&pixel[x*4], _mm_packus_epi16(pix, pix2));
+ x += 3;
+ continue;
+ }
+ if(!pixelmask[x])
+ continue;
+ color = _mm_unpacklo_epi8(_mm_setzero_si128(), _mm_cvtsi32_si128(*(const int *)&buffer_texture_colorbgra8[x*4]));
+ mod = _mm_cvtps_epi32(_mm_mul_ps(data, _mm_load1_ps(&buffer_z[x])));
+ mod = _mm_packs_epi32(mod, mod);
+ pix = _mm_mulhi_epu16(_mm_add_epi16(_mm_mulhi_epu16(mod, Color_Diffusem), Color_Ambientm), color);
+ *(int *)&pixel[x*4] = _mm_cvtsi128_si32(_mm_packus_epi16(pix, pix));