dtourolle
acab0d7abb
dr-export gains write_linear_dng — LinearRaw, DNG 1.4, u16 samples at the sensor's scale, the body's matrices with their illuminants, the as-shot neutral, the EXIF block an export writes — streamed strip by strip through a closure so the composite is never held (FR-MRG-11). The tiff crate's directory is a map, so PhotometricInterpretation is written over what new_image set, which is the trick the S15.1 spike thought it had to hand-roll around. The test reads the file back through rawler. dr-decode's RawImage carries samples_per_pixel (a linear DNG is 3), the body's profile with its calibrations mapped back to EXIF illuminant codes, and the cleaned make and model. The GPU uploads a three-sample image as it is, normalised by black and white like a photosite, through a full f16 conversion — subnormals kept, because a 14-bit LSB sits at f16's smallest normal and rounding it to zero would crush exactly the shadows the file was written to keep.
45 lines
1.8 KiB
TOML
45 lines
1.8 KiB
TOML
[package]
|
|
name = "dr-export"
|
|
version.workspace = true
|
|
edition.workspace = true
|
|
rust-version.workspace = true
|
|
license.workspace = true
|
|
|
|
# No platform dependency and no filesystem, deliberately. This crate turns a
|
|
# rendered frame into *bytes* and a *name*; where those bytes go is the
|
|
# caller's problem, because the answer differs by more than a path. On Linux
|
|
# it is a file, on Android a SAF document descriptor with no path at all
|
|
# (ARCH §6.9), and on either it may be a `PUT` to the server. A crate that
|
|
# took a `Path` would work on exactly one of the three.
|
|
[dependencies]
|
|
dr-types.workspace = true
|
|
log.workspace = true
|
|
thiserror.workspace = true
|
|
|
|
# Encoders. All three are pure Rust and already in the tree, which is the same
|
|
# criterion that chose rustls, bundled SQLite and the Lensfun port: a C
|
|
# dependency here would be one more thing to satisfy under the Android NDK.
|
|
#
|
|
# AVIF and JPEG XL (FR-EXP-1) are deliberately absent. The mature encoders for
|
|
# both are C or C++ — libaom and libjxl — and ravif, the pure-Rust AVIF path,
|
|
# is slow enough to change what a batch export feels like. Neither belongs in
|
|
# the first version; see `format` in lib.rs for what happens when one is asked
|
|
# for.
|
|
jpeg-encoder.workspace = true
|
|
png = "0.18"
|
|
tiff = "0.11"
|
|
|
|
# The example runs the whole path — decode, GPU render, read back, encode,
|
|
# write — so it needs what the library deliberately does not: a GPU, a
|
|
# pipeline and a decoder. Dev-only, so none of it reaches a dependent.
|
|
[dev-dependencies]
|
|
dr-decode.workspace = true
|
|
dr-gpu.workspace = true
|
|
dr-pipeline.workspace = true
|
|
env_logger.workspace = true
|
|
pollster.workspace = true
|
|
# The DNG writer's test reads its output back through the decoder the
|
|
# library uses, which is the whole claim the writer makes (S15.1).
|
|
rawler.workspace = true
|
|
zune-jpeg.workspace = true
|