demo.git - Vide-coded 64k demo system

Age	Commit message (Expand)	Author
8 hours	Replace hard clamp with sigmoid activation in CNN v2	skal
9 hours	Fix --mix option: blend prev layer with static p4-p7, not p0-p3	skal
11 hours	CNN v2: Alpha channel depth handling and layer visualization	skal
17 hours	CNN v2: Fix WebGPU validation error in uniform buffer alignment	skal
17 hours	CNN v2: Change feature #6 from sin(10x) to sin(20y)	skal
17 hours	CNN v2: Add TODO for flexible feature layout in binary format v3	skal
17 hours	CNN v2: Add mip-level support to runtime effect	skal
20 hours	Add shader snippet test assets to fix test_shader_composer	skal
20 hours	Add test asset support with STRIP_ALL guards	skal
21 hours	Add pop-punk drum track from documented sequence	skal
21 hours	CNNv2Effect: Document per-layer kernel sizes support	skal
21 hours	CNN v2: Refactor to uniform 12D→4D architecture	skal
25 hours	Remediation: Implement shared common/shaders/ directory	skal
25 hours	Update workspace.cfg for new directory structure	skal
25 hours	Add weights/ subdirectory to workspaces for CNN training outputs	skal
25 hours	Refactor: Reorganize workspaces and remove assets/ directory	skal
26 hours	Refactor: Move application entry points to src/app/	skal
43 hours	test_demo: Add beat-synchronized CNN post-processing with version selection	skal
45 hours	Timeline editor: CPU load visualization + UX improvements	skal
46 hours	Refine training script output and validation	skal
46 hours	TODO: 8-bit weight quantization for 2× size reduction	skal
46 hours	CNN v2: storage buffer architecture foundation	skal
46 hours	CNN v2: parametric static features - Phases 1-4	skal
48 hours	timeline editor: quantize grid, drag fixes, hotkeys	skal
2 days	cleanup: remove test-only files from main workspace	skal
2 days	fix: update shader files to use beat_phase instead of beat	skal
2 days	feat: implement beat-based timing system	skal
2 days	add trained layers	skal
2 days	docs: Update CNN comments and add bias fix summary	skal
2 days	fix: CNN bias accumulation and output format improvements	skal
3 days	update cnn code	skal
3 days	fix: Register cnn_conv1x1 snippet and add verification	skal
3 days	fix: Move sigmoid activation to call site in CNN layer shader	skal
3 days	fix: Replace clamp with sigmoid in CNN final layer	skal
3 days	format .wgsl layer code (cosmetics)	skal
3 days	opt: Move invariant in1 calculation outside CNN convolution loops	skal
3 days	opt: Vec4-optimize CNN convolution shaders for SIMD	skal
3 days	chore: Update CNN architecture to 3×3×3 with new trained weights	skal
4 days	fix: Correct UV coordinate computation to match PyTorch linspace	skal
4 days	fix: Add clamp to CNN final layer to match PyTorch training	skal
4 days	refactor: Optimize CNN grayscale computation	skal
4 days	update train_cnn.py and shader	skal
4 days	fix: CNN training normalization pipeline consistency	skal
4 days	udpate CNN shader code.	skal
4 days	refactor: Optimize CNN normalization to eliminate redundant conversions	skal
4 days	update timeline.seq	skal
4 days	fix: Flip Y-axis to match ShaderToy coordinate convention	skal
4 days	feat: Add Scene1 effect from ShaderToy (raymarching cube & sphere)	skal
4 days	chore: Remove incomplete CubeSphere effect	skal
4 days	docs: Fix EFFECT keyword syntax and add automation-friendly workflow	skal