refactor: Optimize CNN grayscale computation

Compute gray once per fragment using dot() instead of per-layer.
Pass gray as f32 parameter to conv functions instead of vec4 original.

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>

1 files changed, 3 insertions, 2 deletions

diff --git a/workspaces/main/shaders/cnn/cnn_layer.wgsl b/workspaces/main/shaders/cnn/cnn_layer.wgsl
index 3f970df..e67ad31 100644
--- a/workspaces/main/shaders/cnn/cnn_layer.wgsl
+++ b/workspaces/main/shaders/cnn/cnn_layer.wgsl
@@ -32,6 +32,7 @@ struct CNNLayerParams {
     let uv = p.xy / uniforms.resolution;
     let original_raw = textureSample(original_input, smplr, uv);
     let original = (original_raw - 0.5) * 2.0;  // Normalize to [-1,1]
+    let gray = dot(original.rgb, vec3<f32>(0.2126, 0.7152, 0.0722));
     var result = vec4<f32>(0.0);
 
     // Layer 0: 7→4 (RGBD output, normalizes [0,1] input)
@@ -42,12 +43,12 @@ struct CNNLayerParams {
     }
     else if (params.layer_index == 1) {
         result = cnn_conv5x5_7to4(txt, smplr, uv, uniforms.resolution,
-                                   original, weights_layer1);
+                                   gray, weights_layer1);
         result = cnn_tanh(result);  // Keep in [-1,1]
     }
     else if (params.layer_index == 2) {
         let gray_out = cnn_conv3x3_7to1(txt, smplr, uv, uniforms.resolution,
-                                         original, weights_layer2);
+                                         gray, weights_layer2);
         // gray_out already in [0,1] from clipped training
         result = vec4<f32>(gray_out, gray_out, gray_out, 1.0);
         return mix(original_raw, result, params.blend_amount); // [0,1]