Commit: 6aaf10c79a900549065125f4044cd07bed9f8549
Parent: bd9b2b697ac73f17919eeefba56136b459a8254e
Author: Randy Palamar
Date: Mon, 29 Jul 2024 22:16:12 -0600
fix hsv conversion code when contrast is 0
Clean up a few other things while we are here.
Diffstat:
| M | util.c | | | 37 | ++++++++++++++++++++----------------- |
1 file changed, 20 insertions(+), 17 deletions(-)
diff --git a/util.c b/util.c
@@ -198,23 +198,20 @@ rgb_to_hsv(v4 rgb)
__m128 Min = _mm_min_ps(rgba, _mm_min_ps(gbra, brga));
__m128 C = _mm_sub_ps(Max, Min);
- __m128 t = _mm_div_ps(_mm_sub_ps(gbra, brga), C);
-
- _Alignas(16) f32 aval[4] = { 0, 2, 4, 0 };
- t = _mm_add_ps(t, _mm_load_ps(aval));
-
+ __m128 zero = _mm_set1_ps(0);
+ __m128 six = _mm_set1_ps(6);
+
+ _Alignas(16) f32 aval[4] = {0, 2, 4, 0};
+ _Alignas(16) f32 scale[4] = {1.0/6.0f, 0, 0, 0};
+ /* NOTE if C == 0 then take H as 0/1 (which are equivalent in HSV) */
+ __m128 t = _mm_div_ps(_mm_sub_ps(gbra, brga), C);
+ t = _mm_blendv_ps(t, zero, _mm_cmpeq_ps(zero, C));
+ t = _mm_add_ps(t, _mm_load_ps(aval));
/* TODO: does (G - B) / C ever exceed 6.0? */
- /* NOTE: 1e9 ensures that the remainder after floor is 0.
- * This limits the fmodf to apply only to element [0] */
- _Alignas(16) f32 div[4] = { 6, 1e9, 1e9, 1e9 };
- __m128 six = _mm_set1_ps(6);
- __m128 rem = _mm_floor_ps(_mm_div_ps(t, _mm_load_ps(div)));
- t = _mm_sub_ps(t, _mm_mul_ps(rem, six));
+ /* NOTE: Compute fmodf on element [0] */
+ t = _mm_sub_ps(t, _mm_mul_ps(_mm_floor_ps(_mm_mul_ps(t, _mm_load_ps(scale))), six));
- __m128 zero = _mm_set1_ps(0);
- __m128 maxmask = _mm_cmpeq_ps(rgba, Max);
-
- __m128 H = _mm_div_ps(_mm_blendv_ps(zero, t, maxmask), six);
+ __m128 H = _mm_div_ps(_mm_blendv_ps(zero, t, _mm_cmpeq_ps(rgba, Max)), six);
__m128 S = _mm_div_ps(C, Max);
/* NOTE: Make sure H & S are 0 instead of NaN when V == 0 */
@@ -228,9 +225,15 @@ rgb_to_hsv(v4 rgb)
H = _mm_or_ps(H0, _mm_or_ps(H1, H2));
/* NOTE: keep only element [0] from H vector; Max contains V & A */
- __m128 hva = _mm_blend_ps(Max, H, 0x01);
+ __m128 hva = _mm_blend_ps(Max, H, 0x01);
+ __m128 hsva = _mm_blend_ps(hva, S, 0x02);
+
+ /* NOTE: Clamp values to 1 */
+ __m128 one = _mm_set1_ps(1);
+ hsva = _mm_blendv_ps(hsva, one, _mm_cmplt_ps(one, hsva));
+
v4 res;
- _mm_storeu_ps(res.E, _mm_blend_ps(hva, S, 0x02));
+ _mm_storeu_ps(res.E, hsva);
return res;
}