summaryrefslogtreecommitdiff
AgeCommit message (Expand)Author
2 daysopt: Move invariant in1 calculation outside CNN convolution loopsskal
2 daysopt: Vec4-optimize CNN convolution shaders for SIMDskal
2 dayschore: Update CNN architecture to 3×3×3 with new trained weightsskal
2 daysdocs: Update CNN training documentation with patch extractionskal
2 daysfeat: Add salient-point patch extraction for CNN trainingskal
3 daysfix: Use ClampToEdge sampler for CNN to avoid edge wrappingskal
3 daysfix: Correct UV coordinate computation to match PyTorch linspaceskal
3 daysfix: Add clamp to CNN final layer to match PyTorch trainingskal
3 daysfix: PostProcessHelperTest signature and fixture sharingskal
3 daysrefactor: Optimize CNN grayscale computationskal
3 daysupdate train_cnn.py and shaderskal
3 daysdocs: Add inference mode to training documentationskal
3 daysfeat: Add inference mode to train_cnn.py for ground truth generationskal
3 daysfix: CNN training normalization pipeline consistencyskal
3 daysudpate CNN shader code.skal
3 daysfeat: Add --shader-only option to convert_shadertoy.pyskal
3 daysrefactor: Optimize CNN normalization to eliminate redundant conversionsskal
3 daysupdate timeline.seqskal
3 daysfix: Flip Y-axis to match ShaderToy coordinate conventionskal
3 daysrefactor: Improve convert_shadertoy.py to generate compile-ready codeskal
3 daysfeat: Add Scene1 effect from ShaderToy (raymarching cube & sphere)skal
3 dayschore: Remove incomplete CubeSphere effectskal
3 daysdocs: Fix EFFECT keyword syntax and add automation-friendly workflowskal
3 daysfeat: Add ShaderToy conversion toolsskal
3 daysfix: Support variable kernel sizes in CNN layer generationskal
3 daysdocs: Document WGPU builder refactoring in COMPLETED.mdskal
3 daysrefactor: Factor WGPU boilerplate into builder pattern helpersskal
3 daysfeat: CNN RGBD→grayscale with 7-channel augmented inputskal
3 daysudpateskal
3 daysupdate timelineskal
3 daysfix: Resolve CNN effect black screen bug (framebuffer capture + uniforms)skal
3 daysfix: Capture scene framebuffer before post-processing for CNN effectskal
3 daysfeat: Add multi-layer CNN support with framebuffer capture and blend controlskal
3 daysdocs: Update and streamline CNN training documentationskal
3 daysfeat: Add checkpointing support to CNN training scriptskal
3 daysfix: Auto-expand single kernel size to all layers in training scriptskal
3 daysupdate target imagesskal
3 daysfeat: Add coordinate-aware CNN layer 0 for position-dependent stylizationskal
3 daysdocs: Add CNN effect documentation and update project statusskal
3 daysfeat: Add CNN post-processing effect with modular WGSL architectureskal
3 daysinitial doc for the CNN projectskal
3 daysdocs: Reorganize shader editor documentationskal
3 daysfix: Use placeholder technique to prevent syntax highlighting overlapskal
3 daysrefactor: Improve syntax highlighting to avoid overlapping matchesskal
3 daysfeat: Add WGSL syntax highlightingskal
3 daysrefactor: Increase editor pane size by 20% (36% → 43%)skal
3 daysfix: Remove resolutionSelect references after removing UI elementskal
3 daysfeat: Remove resolution selector, enable autoplay on loadskal
3 daysfix: Use 16:9 aspect ratio and object-fit contain for canvasskal
3 daysfix: Canvas fills available height instead of maintaining square aspectskal