Add a simple luma wipe transition.

[movit] / luma_mix_effect.frag

diff --git a/luma_mix_effect.frag b/luma_mix_effect.frag
new file mode 100644 (file)
index 0000000..21365a6
--- /dev/null
+++ b/luma_mix_effect.frag
@@ -0,0 +1,44 @@
+uniform float PREFIX(progress_mul_w_plus_one);
+
+vec4 FUNCNAME(vec2 tc) {
+       vec4 first = INPUT1(tc);
+       vec4 second = INPUT2(tc);
+
+       // We treat the luma as going from 0 to w, where w is the transition width
+       // (wider means the boundary between transitioned and non-transitioned
+       // will be harder, while w=0 is essentially just a straight fade).
+       // We need to map this 0..w range in the luma image to a (clamped) 0..1
+       // range for how far this pixel has come in a fade. At the very
+       // beginning, we can visualize it like this, where every pixel is in
+       // the state 0.0 (100% first image, 0% second image):
+       //
+       //         0                     w
+       //   luma: |---------------------|
+        //   mix:                        |----|
+        //                               0    1
+       //
+       // Then as we progress, eventually the luma range should move to the right
+       // so that more pixels start moving towards higher mix value:
+       //
+       //           0                     w
+       //   luma:   |---------------------|
+        //   mix:                        |----|
+        //                               0    1
+       //
+       // and at the very end, all pixels should be in the state 1.0 (0% first image,
+       // 100% second image):
+       //
+       //                                    0                     w
+       //   luma:                            |---------------------|
+        //   mix:                        |----|
+        //                               0    1
+       //
+       // So clearly, it should move (w+1) units to the right, and apart from that
+       // just stay a simple mapping.
+       float w = PREFIX(transition_width);
+       float luma = INPUT3(tc).x * w;
+       float m = clamp((luma - w) + PREFIX(progress_mul_w_plus_one), 0.0, 1.0);
+
+       return mix(first, second, m);
+//     return vec4(luma, luma, luma, 1.0);
+}