Solve from spectra
Solve white balance and a camera RGB to ACES matrix from measured
spectra. rta::core::SpectralSolver and
rawtoaces.SpectralSolver do not decode a RAW file. There is
no command-line equivalent. Pass the spectral database directories to
the constructor, or set RAWTOACES_DATA_PATH before a solver that
uses the default search.
Before calculate_transform, the camera needs three channels, the
illuminant one, the observer three, and the training set must be
non-empty. find_camera and the name form of find_illuminant load
those two objects. d55 and 3200k are generated; any other name
must exist in the database. The name match is case-insensitive. The name
form does not calculate white balance. Call calculate_WB after it, or
use the three-multiplier form of find_illuminant when you already
have multipliers.
On success get_WB_multipliers returns the RGB multipliers and
transform_matrix holds the 3×3 transform. A missing camera,
illuminant, observer, or training set makes calculate_transform
return false and sets last_error_message. Leaving white balance
unset keeps the multipliers at [1.0, 1.0, 1.0] and is not reported
as an error. Read both results only after the calls that fill them
return true.
The dataset layout is documented with the rawtoaces-data repository. The paths below are relative to a database directory from that layout.
#include <rawtoaces/rawtoaces_core.h>
#include <iostream>
#include <string>
#include <vector>
int main()
{
const std::vector<std::string> search_paths = {
"/path/to/rawtoaces-data/data"
};
rta::core::SpectralSolver solver(search_paths);
if (!solver.find_camera("nikon", "d200"))
{
std::cerr << solver.last_error_message << '\n';
return 1;
}
if (!solver.find_illuminant("d55"))
{
std::cerr << solver.last_error_message << '\n';
return 1;
}
if (!solver.load_spectral_data(
"cmf/cmf_1931.json", solver.observer))
{
std::cerr << solver.last_error_message << '\n';
return 1;
}
if (!solver.load_spectral_data(
"training/training_spectral.json", solver.training_data))
{
std::cerr << solver.last_error_message << '\n';
return 1;
}
if (!solver.calculate_WB())
{
std::cerr << solver.last_error_message << '\n';
return 1;
}
if (!solver.calculate_transform())
{
std::cerr << solver.last_error_message << '\n';
return 1;
}
const auto &white_balance = solver.get_WB_multipliers();
const auto &transform = solver.transform_matrix;
return white_balance.empty() || transform.empty() ? 1 : 0;
}
C++ can also fill a spectrum with calculate_daylight_SPD or
calculate_blackbody_SPD and adapt D65 XYZ with
XYZD65Solver.
An empty illuminant name, or a multiplier list whose length is not
three, raises ValueError.
import rawtoaces
solver = rawtoaces.SpectralSolver([
"/path/to/rawtoaces-data/data"
])
if not solver.find_camera("nikon", "d200"):
raise RuntimeError(solver.last_error_message or "camera not found")
if not solver.find_illuminant("d55"):
raise RuntimeError(
solver.last_error_message or "illuminant not found"
)
if not solver.load_spectral_data("cmf/cmf_1931.json", solver.observer):
raise RuntimeError(
solver.last_error_message or "failed to load observer"
)
if not solver.load_spectral_data(
"training/training_spectral.json", solver.training_data
):
raise RuntimeError(
solver.last_error_message or "failed to load training data"
)
if not solver.calculate_WB():
raise RuntimeError(solver.last_error_message)
if not solver.calculate_transform():
raise RuntimeError(solver.last_error_message)
white_balance = solver.get_WB_multipliers()
transform = solver.transform_matrix
calculate_daylight_SPD, calculate_blackbody_SPD, and
XYZD65Solver are C++ only. Python reaches the built-in
illuminants through find_illuminant, as this example does.
The Python core guide explains the containers. Signatures are in the C++ Reference and the Python Reference.