archive(cnn): move CNN v1 to cnn_v1/ subdirectory

Consolidate CNN v1 (CNNEffect) into dedicated directory:
- C++ effect: src/effects → cnn_v1/src/
- Shaders: workspaces/main/shaders/cnn → cnn_v1/shaders/
- Training: training/train_cnn.py → cnn_v1/training/
- Docs: doc/CNN*.md → cnn_v1/docs/

Updated all references:
- CMake source list
- C++ includes (relative paths: ../../cnn_v1/src/)
- Asset paths (../../cnn_v1/shaders/)
- Documentation cross-references

CNN v1 remains active in timeline. For new work, use CNN v2 with
enhanced features (7D static, storage buffer, sigmoid activation).

Tests: 34/34 passing (100%)

1 files changed, 100 insertions, 0 deletions

diff --git a/cnn_v1/shaders/cnn_conv1x1.wgsl b/cnn_v1/shaders/cnn_conv1x1.wgsl
new file mode 100644
index 0000000..f77cfa8
--- /dev/null
+++ b/cnn_v1/shaders/cnn_conv1x1.wgsl
@@ -0,0 +1,100 @@
+// 1x1 convolution (vec4-optimized)
+
+// Inner layers: 7→4 channels (vec4-optimized)
+// Assumes 'tex' is already normalized to [-1,1]
+fn cnn_conv1x1_7to4(
+  tex: texture_2d<f32>,
+  samp: sampler,
+  uv: vec2<f32>,
+  resolution: vec2<f32>,
+  gray: f32,
+  weights: array<vec4<f32>, 8>
+) -> vec4<f32> {
+  let step = 1.0 / resolution;
+  let uv_norm = (uv - 0.5) * 2.0;
+
+  var sum = vec4<f32>(0.0);
+  var pos = 0;
+
+  for (var dy = -0; dy <= 0; dy++) {
+    for (var dx = -0; dx <= 0; dx++) {
+      let offset = vec2<f32>(f32(dx), f32(dy)) * step;
+      let rgbd = textureSample(tex, samp, uv + offset);
+      let in1 = vec4<f32>(uv_norm, gray, 1.0);
+
+      sum.r += dot(weights[pos+0], rgbd) + dot(weights[pos+1], in1);
+      sum.g += dot(weights[pos+2], rgbd) + dot(weights[pos+3], in1);
+      sum.b += dot(weights[pos+4], rgbd) + dot(weights[pos+5], in1);
+      sum.a += dot(weights[pos+6], rgbd) + dot(weights[pos+7], in1);
+      pos += 8;
+    }
+  }
+
+  return sum;
+}
+
+// Source layer: 7→4 channels (vec4-optimized)
+// Normalizes [0,1] input to [-1,1] internally
+fn cnn_conv1x1_7to4_src(
+  tex: texture_2d<f32>,
+  samp: sampler,
+  uv: vec2<f32>,
+  resolution: vec2<f32>,
+  weights: array<vec4<f32>, 8>
+) -> vec4<f32> {
+  let step = 1.0 / resolution;
+
+  var original = (textureSample(tex, samp, uv) - 0.5) * 2.0;
+  let gray = dot(original.rgb, vec3<f32>(0.2126, 0.7152, 0.0722));
+  let uv_norm = (uv - 0.5) * 2.0;
+  let in1 = vec4<f32>(uv_norm, gray, 1.0);
+
+  var sum = vec4<f32>(0.0);
+  var pos = 0;
+
+  for (var dy = -0; dy <= 0; dy++) {
+    for (var dx = -0; dx <= 0; dx++) {
+      let offset = vec2<f32>(f32(dx), f32(dy)) * step;
+      var rgbd = (textureSample(tex, samp, uv + offset) - 0.5) * 2.0;
+
+      sum.r += dot(weights[pos+0], rgbd) + dot(weights[pos+1], in1);
+      sum.g += dot(weights[pos+2], rgbd) + dot(weights[pos+3], in1);
+      sum.b += dot(weights[pos+4], rgbd) + dot(weights[pos+5], in1);
+      sum.a += dot(weights[pos+6], rgbd) + dot(weights[pos+7], in1);
+      pos += 8;
+    }
+  }
+
+  return sum;
+}
+
+// Final layer: 7→1 channel (vec4-optimized)
+// Assumes 'tex' is already normalized to [-1,1]
+// Returns raw sum (activation applied at call site)
+fn cnn_conv1x1_7to1(
+  tex: texture_2d<f32>,
+  samp: sampler,
+  uv: vec2<f32>,
+  resolution: vec2<f32>,
+  gray: f32,
+  weights: array<vec4<f32>, 2>
+) -> f32 {
+  let step = 1.0 / resolution;
+  let uv_norm = (uv - 0.5) * 2.0;
+  let in1 = vec4<f32>(uv_norm, gray, 1.0);
+
+  var sum = 0.0;
+  var pos = 0;
+
+  for (var dy = -0; dy <= 0; dy++) {
+    for (var dx = -0; dx <= 0; dx++) {
+      let offset = vec2<f32>(f32(dx), f32(dy)) * step;
+      let rgbd = textureSample(tex, samp, uv + offset);
+
+      sum += dot(weights[pos], rgbd) + dot(weights[pos+1], in1);
+      pos += 2;
+    }
+  }
+
+  return sum;
+}