Replace the position capture cutoff with a camera-relative dark threshold
shared by every backend, and carry explicit outcome/reason provenance
through the ray, RayPool, adaptive mesh, lens-map and replay paths.
- eval/eval_slab return SpacetimePointStatus; remove SPACETIME_RAY_CAPTURED
and the Schwarzschild capture radius; decouple observer construction from
ray position.
- RayEndpoint stores RayOutcome/RayReason plus the last trusted state;
budget exhaustion is retryable UNRESOLVED, data/integration failures are
INCOMPLETE.
- Normal dark terminal is L - L0 >= --dark-threshold (default 8), with L0
taken at the camera event and kept distinct from the worldtube entry
energy; photon energy and frequency ratio are never reset.
- Implement E/D/U triangle decisions with merged budget retries, persistent
probe witnesses promoted in place by vertex identity, conformity settling,
and approximate-black boundary provenance with achieved-scale statistics.
- Add RayPool continuation state and per-ray step budgets.
- Bump lens-map to v2 with explicit end/outcome/reason, approx_black,
threshold/retry/geometry provenance and per-frame retry counts; reject v1.
- Gate production output on incomplete/error results, overridable with
--allow-incomplete.
- Update AGENTS.md, the design document and usage docs; add the termination
oracle and regression coverage.
make -B -j4 BUILD_TYPE=Debug test passes with bit-identical reference HDRs.
Add an optional CPU preview path that deposits each point-source image as a
supersampled delta and resolves the whole frame with one global Moffat
convolution plus an N x N box average, instead of splatting a per-event PSF.
- optics: FastPsfAccumulator builds the pixel-area-integral kernel of the
target Moffat at the supersampled scale (width N*alpha, same beta). The
1/N^2 box average then reproduces the final pixel-area integral, so the
requested FWHM and beta are preserved without renormalisation. Deposits are
per-cell atomic adds; resolve accumulates into the caller's HDR buffer.
- frame: fast branch in frame_splat_catalog with one shared supersampled
buffer and a single resolve per frame; the accumulator is reused across
movie frames and built from the map dimensions on lens-map import.
- main: --fast-mode, --fast-supersample N (1..8, default 2) and
--fast-deposit nearest|bilinear (default nearest). CPU-only and rejected in
the HIP/dummy backends; --psf-min-y still applies per event while
--max-cache-psf-flux does not.
- The deposition scheme was chosen by scripts/fast_mode_deposit_error.py:
nearest keeps the PSF shape exactly with <= 0.5/N px position quantization;
bilinear keeps the exact centroid but broadens FWHM and beta. Recorded in
benchmarks/fast_mode_deposit_2026-09-18.md.
- tests/test_frame.c covers fast nearest vs the direct evaluator at the snapped
centre, flux conservation, bilinear centroid, min-Y discard, frame plumbing,
and HDR accumulation onto a non-zero background.
- benchmarks/fast_mode_cpu_2026-09-18.md records a ~10x speedup on the 2MASS
galactic-centre field with small tone-mapped differences.
Replace the in-tree Wyman analytic CIE fit with a repository CIE 1931
2-degree LUT derived from the 360-830 nm, 1 nm-linear CSV reference. The
GRBBLUT3 table stores 1024 uniform log(T) XYZ nodes over
[670.146556, 101408.88] K with four-point cubic Lagrange interpolation, and
the same CSV reference supplies a three-term Rayleigh-Jeans form above the
table and an inverse-temperature/log-XYZ crossover with channel-specific
endpoint forms below it. The loader validates the header and payload
checksum and rejects legacy formats and runtime generation.
Initialize the backend in the renderer, frame test, and PSF capture tool so
the new default is active wherever colors are produced.
Capture prepared events through a test-only producer consumer and measure serial and indexed parallel generation independently. Compare the production atomic kernel with bounded pixel-owned tile reductions while preserving double HDR, cache support, and direct fallback ordering.
Validate captured inputs, edge and multichunk fixtures, CPU/HIP renderer regressions, and 45 isolated final replays. Keep the production HIP algorithm unchanged because gains depend on event distribution.
wyj