Research · Tooling · CLI
HCAISSR
A six-engine offline photo pipeline: auto-sorts, rebuilds structural defects, upscales with hybrid classical and AI methods, simulates depth of field, corrects artifacts, and compresses for delivery.
- 6
- Engines
- 114/114
- Tests passing
- ~820ms
- 1024² on M2 Max
- 20
- Pipeline configs
HCAISSR is a modular image processing system built as six functional engines. E1 (AutoSort) groups photos by viewpoint and selects best references. E2 (Rebuilder) repairs structural defects using coverage from other frames. E3 (HSSR) performs hybrid super-resolution combining classical and AI techniques. E4 (Bokeh) adds depth-of-field simulation. E5 (Polisher) corrects artifacts and applies colour grading. E6 (Compressor) handles output format selection. An optional augmentation layer provides hardware profiling, model routing and perceptual analysis. Everything runs locally — no network, no account.
Results
- 114/114 tests passing across all engines and pipeline combinations
- Full pipeline processes 1024×1024 images in ~820ms on Apple M2 Max
- Supports 20 pipeline configurations selectable at runtime
- Hardware acceleration via NEON SIMD, Metal GPU, CUDA and Vulkan