RadiationModel.py

Go to the documentation of this file.

"""
High-level RadiationModel interface for PyHelios.
 
This module provides a user-friendly interface to the radiation modeling
capabilities with graceful plugin handling and informative error messages.
"""
 
import logging
import math
import tempfile
from typing import Dict, List, Optional, Union
from contextlib import contextmanager
from pathlib import Path
import os
 
import numpy as np
 
from .plugins.registry import get_plugin_registry, require_plugin, graceful_plugin_fallback
from .wrappers import URadiationModelWrapper as radiation_wrapper
# Bulk primitive-data reads go straight to the Context wrapper: the band results
# live in Context primitive data, and getAbsorbedFlux has already validated the
# UUIDs and probed the label, so Context.getPrimitiveDataArray() would only repeat
# that work.
from .wrappers import UContextWrapper as context_wrapper
from .validation.plugins import (
    validate_wavelength_range, validate_flux_value, validate_ray_count,
    validate_direction_vector, validate_band_label, validate_source_id, validate_source_id_list,
    validate_position_like, validate_direction_like, validate_size_like
)
from .validation.plugin_decorators import (
    validate_radiation_band_params, validate_collimated_source_params, validate_sphere_source_params,
    validate_sun_sphere_params, validate_get_source_flux_params,
    validate_update_geometry_params, validate_run_band_params, validate_scattering_depth_params,
    validate_min_scatter_energy_params
)
from .Context import Context, check_context_alive
from .assets import get_asset_manager
 
logger = logging.getLogger(__name__)
 
 
@contextmanager
def _radiation_working_directory():
    """
    Context manager that temporarily changes working directory to where RadiationModel assets are located.
    
    RadiationModel C++ code uses hardcoded relative paths like "plugins/radiation/" for GPU
    backend files (SPIR-V shaders for Vulkan, PTX files for OptiX), expecting assets relative
    to working directory. This manager temporarily changes to the build directory where assets
    are actually located.
    
    Raises:
        RuntimeError: If build directory or RadiationModel assets are not found, indicating a build system error.
    """
    # Find the build directory containing RadiationModel assets
    # Try asset manager first (works for both development and wheel installations)
    asset_manager = get_asset_manager()
    working_dir = asset_manager._get_helios_build_path()
    
    if working_dir and working_dir.exists():
        radiation_assets = working_dir / 'plugins' / 'radiation'
    else:
        # For wheel installations, check packaged assets  
        current_dir = Path(__file__).parent
        packaged_build = current_dir / 'assets' / 'build'
        
        if packaged_build.exists():
            working_dir = packaged_build
            radiation_assets = working_dir / 'plugins' / 'radiation'
        else:
            # Fallback to development paths
            repo_root = current_dir.parent
            build_lib_dir = repo_root / 'pyhelios_build' / 'build' / 'lib'
            working_dir = build_lib_dir.parent
            radiation_assets = working_dir / 'plugins' / 'radiation'
            
            if not build_lib_dir.exists():
                raise RuntimeError(
                    f"PyHelios build directory not found at {build_lib_dir}. "
                    f"Run: python build_scripts/build_helios.py --plugins radiation"
                )
    
    if not radiation_assets.exists():
        raise RuntimeError(
            f"RadiationModel assets not found at {radiation_assets}. "
            f"This indicates a build system error. The build script should copy shader/backend files to this location."
        )
    
    # Change to the build directory temporarily
    original_dir = os.getcwd()
    try:
        os.chdir(working_dir)
        logger.debug(f"Changed working directory to {working_dir} for RadiationModel asset access")
        yield working_dir
    finally:
        os.chdir(original_dir)
        logger.debug(f"Restored working directory to {original_dir}")
 
 
class RadiationModelError(Exception):
    """Raised when RadiationModel operations fail."""
    pass
 
 
class CameraProperties:
    """
    Camera properties for radiation model cameras.
 
    This class encapsulates the properties needed to configure a radiation camera,
    providing sensible defaults and validation for camera parameters. Carries the
    v1.3.60 ``camera_zoom`` and v1.3.85 ``exposure_target`` fields.
    """
 
    def __init__(self, camera_resolution=None, focal_plane_distance=1.0, lens_diameter=0.05,
                 HFOV=20.0, FOV_aspect_ratio=0.0, lens_focal_length=0.05,
                 sensor_width_mm=35.0, manufacturer="", model="generic", lens_make="",
                 lens_model="", lens_specification="", exposure="auto", shutter_speed=1.0/125.0,
                 white_balance="auto", camera_zoom=1.0, exposure_target=0.18):
        """
        Initialize camera properties with defaults matching C++ CameraProperties.
 
        Args:
            camera_resolution: Camera resolution as (width, height) tuple or list. Default: (512, 512)
            focal_plane_distance: Distance from viewing plane to focal plane (working distance). Default: 1.0
            lens_diameter: Diameter of camera lens (0 = pinhole camera). Default: 0.05
            HFOV: Horizontal field of view in degrees. Default: 20.0
            FOV_aspect_ratio: Ratio of horizontal to vertical FOV. Default: 0.0 (auto-calculate from resolution)
            lens_focal_length: Camera lens optical focal length in meters (physical, not 35mm equiv). Default: 0.05 (50mm)
            sensor_width_mm: Physical sensor width in mm. Default: 35.0 (full-frame)
            manufacturer: Camera manufacturer (e.g., "Canon", "Nikon", "Apple"). In helios-core
                v1.3.73 this maps to the EXIF Make tag (empty ⇒ "Helios"). Default: ""
            model: Camera model name (e.g., "Nikon D700", "Canon EOS 5D"). Default: "generic"
            lens_make: Lens manufacturer (e.g., "Canon", "Nikon"). Default: ""
            lens_model: Lens model name (e.g., "AF-S NIKKOR 50mm f/1.8G"). Default: ""
            lens_specification: Lens specification (e.g., "50mm f/1.8"). Default: ""
            exposure: Exposure mode - "auto", "ISOXXX" (e.g., "ISO100"), or "manual". Default: "auto"
            shutter_speed: Camera shutter speed in seconds (e.g., 0.008 for 1/125s). Default: 0.008 (1/125s)
            white_balance: White balance mode - "auto" or "off". Default: "auto".
                "auto" scales the camera's first three bands (red, green, blue) so that a
                spectrally flat surface renders neutral under the light actually reaching the
                surfaces in view, keeping the brightest band at unit gain. A scene lit by a
                spectrally flat source is therefore left unchanged. If the camera sees no
                light in one of those bands -- only sky, for instance -- the image is left
                unbalanced and the reason is reported as a warning.
            camera_zoom: Camera optical zoom multiplier. 1.0 = no zoom, 2.0 = 2x zoom.
                        Scales effective HFOV: effective_HFOV = HFOV / camera_zoom. Default: 1.0
            exposure_target: Target median scene luminance for ``"auto"`` exposure, as a
                        fraction of full scale (helios-core v1.3.85+). Auto exposure scales every
                        band so the median pixel luminance lands here. The default 0.18 is the
                        middle-grey convention, which assumes a scene of grey-card mean
                        reflectance; a canopy imaged from above is darker, so matching a real
                        camera may need a lower value. Ignored for ``"manual"`` and ``"ISOXXX"``
                        exposure. Must be > 0. Default: 0.18
        """
        # Set camera resolution with validation
        
        if camera_resolution is None:
            self.camera_resolution = (512, 512)
        else:
            if isinstance(camera_resolution, (list, tuple)) and len(camera_resolution) == 2:
                self.camera_resolution = (int(camera_resolution[0]), int(camera_resolution[1]))
            else:
                raise ValueError("camera_resolution must be a tuple or list of 2 integers")
        
 
        # Validate and set numeric properties
        if focal_plane_distance <= 0:
            raise ValueError("focal_plane_distance must be greater than 0")
        if lens_diameter < 0:
            raise ValueError("lens_diameter must be non-negative")
        if HFOV <= 0 or HFOV > 180:
            raise ValueError("HFOV must be between 0 and 180 degrees")
        if FOV_aspect_ratio < 0:
            raise ValueError("FOV_aspect_ratio must be non-negative (0 = auto-calculate)")
        if lens_focal_length <= 0:
            raise ValueError("lens_focal_length must be greater than 0")
        if sensor_width_mm <= 0:
            raise ValueError("sensor_width_mm must be greater than 0")
        if shutter_speed <= 0:
            raise ValueError("shutter_speed must be greater than 0")
        if camera_zoom <= 0:
            raise ValueError("camera_zoom must be greater than 0")
        if exposure_target <= 0:
            raise ValueError("exposure_target must be greater than 0")
 
        self.focal_plane_distance = float(focal_plane_distance)
        self.lens_diameter = float(lens_diameter)
        self.HFOV = float(HFOV)
        self.FOV_aspect_ratio = float(FOV_aspect_ratio)
        self.lens_focal_length = float(lens_focal_length)
        self.sensor_width_mm = float(sensor_width_mm)
        self.shutter_speed = float(shutter_speed)
        self.camera_zoom = float(camera_zoom)
        self.exposure_target = float(exposure_target)
 
        # Validate and set string properties
        if not isinstance(manufacturer, str):
            raise ValueError("manufacturer must be a string")
        if not isinstance(model, str):
            raise ValueError("model must be a string")
        if not isinstance(lens_make, str):
            raise ValueError("lens_make must be a string")
        if not isinstance(lens_model, str):
            raise ValueError("lens_model must be a string")
        if not isinstance(lens_specification, str):
            raise ValueError("lens_specification must be a string")
        if not isinstance(exposure, str):
            raise ValueError("exposure must be a string")
        if not isinstance(white_balance, str):
            raise ValueError("white_balance must be a string")
 
        # Validate exposure mode
        if exposure not in ["auto", "manual"] and not exposure.startswith("ISO"):
            raise ValueError("exposure must be 'auto', 'manual', or 'ISOXXX' (e.g., 'ISO100')")
 
        # Validate white balance mode
        if white_balance not in ["auto", "off"]:
            raise ValueError("white_balance must be 'auto' or 'off'")
 
        self.manufacturer = str(manufacturer)
        self.model = str(model)
        self.lens_make = str(lens_make)
        self.lens_model = str(lens_model)
        self.lens_specification = str(lens_specification)
        self.exposure = str(exposure)
        self.white_balance = str(white_balance)
 
    def to_array(self):
        """
        Convert to array format expected by C++ interface.
 
        Note: Returns numeric fields only. String fields (model, lens_make, etc.) are
        currently initialized with defaults in the C++ wrapper and cannot be set via
        this interface. Use the upcoming camera library methods for full metadata control.
 
        Returns:
            List of 11 float values: [resolution_x, resolution_y, focal_distance, lens_diameter,
                                      HFOV, FOV_aspect_ratio, lens_focal_length, sensor_width_mm,
                                      shutter_speed, camera_zoom, exposure_target]
        """
        return [
            float(self.camera_resolution[0]),  # resolution_x
            float(self.camera_resolution[1]),  # resolution_y
            self.focal_plane_distance,
            self.lens_diameter,
            self.HFOV,
            self.FOV_aspect_ratio,
            self.lens_focal_length,
            self.sensor_width_mm,
            self.shutter_speed,
            self.camera_zoom,
            self.exposure_target
        ]
 
    def __repr__(self):
        return (f"CameraProperties("
                f"camera_resolution={self.camera_resolution}, "
                f"focal_plane_distance={self.focal_plane_distance}, "
                f"lens_diameter={self.lens_diameter}, "
                f"HFOV={self.HFOV}, "
                f"FOV_aspect_ratio={self.FOV_aspect_ratio}, "
                f"lens_focal_length={self.lens_focal_length}, "
                f"sensor_width_mm={self.sensor_width_mm}, "
                f"manufacturer='{self.manufacturer}', "
                f"model='{self.model}', "
                f"lens_make='{self.lens_make}', "
                f"lens_model='{self.lens_model}', "
                f"lens_specification='{self.lens_specification}', "
                f"exposure='{self.exposure}', "
                f"shutter_speed={self.shutter_speed}, "
                f"white_balance='{self.white_balance}', "
                f"camera_zoom={self.camera_zoom}, "
                f"exposure_target={self.exposure_target})")
 
 
class SIFCameraProperties(CameraProperties):
    """
    Camera properties for a solar-induced chlorophyll fluorescence (SIF) camera.
 
    Extends :class:`CameraProperties` with two SIF-specific fields used by the
    Fluspect-B emission pipeline introduced in helios-core v1.3.72. Image geometry,
    resolution, exposure, and spectral-response handling are inherited from
    :class:`CameraProperties` unchanged.
 
    Attributes:
        excitation_bin_width_nm: Excitation wavelength bin width in nm. Helios
            auto-creates internal radiation bands spanning 400–750 nm at this
            resolution to compute per-leaf APAR. Must be > 0. Default 10.0.
        excitation_scattering_depth: Scattering depth for the auto-generated
            excitation bands. ``0`` (default) treats every leaf hit as fully
            absorbed. Set to ``>=1`` to include inter-leaf scattering at the
            cost of additional excitation-band ray traces.
 
    Note:
        String fields inherited from :class:`CameraProperties` (``model``,
        ``lens_make``, ``lens_model``, ``lens_specification``, ``exposure``,
        ``white_balance``) are passed through to the C++ camera, so they affect
        rendering and EXIF metadata just as they do for ``addRadiationCamera``.
    """
 
    def __init__(self, excitation_bin_width_nm: float = 10.0,
                 excitation_scattering_depth: int = 0,
                 **kwargs):
        super().__init__(**kwargs)
        if excitation_bin_width_nm <= 0:
            raise ValueError("excitation_bin_width_nm must be greater than 0")
        if excitation_scattering_depth < 0:
            raise ValueError("excitation_scattering_depth must be >= 0")
        self.excitation_bin_width_nm = float(excitation_bin_width_nm)
        self.excitation_scattering_depth = int(excitation_scattering_depth)
 
    def __repr__(self):
        base = super().__repr__()
        # Drop the trailing ')' from the parent repr and append SIF-specific fields.
        return (
            base[:-1]
            + f", excitation_bin_width_nm={self.excitation_bin_width_nm}"
            + f", excitation_scattering_depth={self.excitation_scattering_depth})"
        )
 
 
class CameraMetadata:
    """
    Metadata for radiation camera image export (Helios v1.3.58+).
 
    This class encapsulates comprehensive metadata for camera images including
    camera properties, location, acquisition settings, image processing, and
    agronomic properties derived from plant architecture data.
    """
 
    class CameraPropertiesMetadata:
        """Camera intrinsic properties for metadata export."""
        def __init__(self, height=512, width=512, channels=3, type="rgb",
                     focal_length=50.0, aperture="f/2.8", sensor_width=35.0,
                     sensor_height=24.0, model="generic", lens_make="",
                     lens_model="", lens_specification="", exposure="auto",
                     shutter_speed=0.008, white_balance="auto"):
            self.height = int(height)
            self.width = int(width)
            self.channels = int(channels)
            self.type = str(type)
            self.focal_length = float(focal_length)
            self.aperture = str(aperture)
            self.sensor_width = float(sensor_width)
            self.sensor_height = float(sensor_height)
            self.model = str(model)
            self.lens_make = str(lens_make)
            self.lens_model = str(lens_model)
            self.lens_specification = str(lens_specification)
            self.exposure = str(exposure)
            self.shutter_speed = float(shutter_speed)
            self.white_balance = str(white_balance)
 
    class LocationProperties:
        """Geographic location properties."""
        def __init__(self, latitude=0.0, longitude=0.0):
            self.latitude = float(latitude)
            self.longitude = float(longitude)
 
    class AcquisitionProperties:
        """Image acquisition properties."""
        def __init__(self, date="", time="", UTC_offset=0.0, camera_height_m=0.0,
                     camera_angle_deg=0.0, light_source="sunlight"):
            self.date = str(date)
            self.time = str(time)
            self.UTC_offset = float(UTC_offset)
            self.camera_height_m = float(camera_height_m)
            self.camera_angle_deg = float(camera_angle_deg)
            self.light_source = str(light_source)
 
    class ImageProcessingProperties:
        """Image processing corrections applied to the image."""
        def __init__(self, saturation_adjustment=1.0, brightness_adjustment=1.0,
                     contrast_adjustment=1.0, color_space="linear"):
            self.saturation_adjustment = float(saturation_adjustment)
            self.brightness_adjustment = float(brightness_adjustment)
            self.contrast_adjustment = float(contrast_adjustment)
            self.color_space = str(color_space)
 
    class AgronomicProperties:
        """Agronomic properties derived from plant architecture data."""
        def __init__(self, plant_species=None, plant_count=None, plant_height_m=None,
                     plant_age_days=None, plant_stage=None, leaf_area_m2=None,
                     weed_pressure=""):
            self.plant_species = plant_species if plant_species is not None else []
            self.plant_count = plant_count if plant_count is not None else []
            self.plant_height_m = plant_height_m if plant_height_m is not None else []
            self.plant_age_days = plant_age_days if plant_age_days is not None else []
            self.plant_stage = plant_stage if plant_stage is not None else []
            self.leaf_area_m2 = leaf_area_m2 if leaf_area_m2 is not None else []
            self.weed_pressure = str(weed_pressure)
 
    def __init__(self, path=""):
        """
        Initialize CameraMetadata with default values.
 
        Args:
            path: Full path to the associated image file. Default: ""
        """
        self.path = str(path)
        self.camera_properties = self.CameraPropertiesMetadata()
        self.location_properties = self.LocationProperties()
        self.acquisition_properties = self.AcquisitionProperties()
        self.image_processing = self.ImageProcessingProperties()
        self.agronomic_properties = self.AgronomicProperties()
 
    def __repr__(self):
        return (f"CameraMetadata(path='{self.path}', "
                f"camera={self.camera_properties.model}, "
                f"resolution={self.camera_properties.width}x{self.camera_properties.height}, "
                f"location=({self.location_properties.latitude},{self.location_properties.longitude}))")
 
 
class RadiationModel:
    """
    High-level interface for radiation modeling and ray tracing.
    
    This class provides a user-friendly wrapper around the native Helios
    radiation plugin with automatic plugin availability checking and
    graceful error handling.
    """
    
    def __init__(self, context: Context):
        """
        Initialize RadiationModel with graceful plugin handling.
        
        Args:
            context: Helios Context instance
            
        Raises:
            TypeError: If context is not a Context instance
            RadiationModelError: If radiation plugin is not available
        """
        # Validate context type
        if not isinstance(context, Context):
            raise TypeError(f"RadiationModel requires a Context instance, got {type(context).__name__}")
        
        self.context = context
        self.radiation_model = None
        # Tracks whether scene geometry has been pushed to the radiation model, either
        # by updateGeometry() or by runBand() building it automatically. Some queries
        # (e.g. calculateGtheta) silently return NaN from the native layer when no
        # geometry is loaded; this flag lets us auto-update or fail with an actionable
        # message instead.
        self._geometry_updated = False
        # Tracks whether the last build was a user-specified subset (updateGeometry(uuids)).
        # A subset must never be silently replaced by a full-Context build, so the
        # auto-update paths below refuse to do so.
        self._geometry_subset = False
 
        # Check plugin availability using registry
        registry = get_plugin_registry()
        
        if not registry.is_plugin_available('radiation'):
            # Get helpful information about the missing plugin
            plugin_info = registry.get_plugin_capabilities()
            available_plugins = registry.get_available_plugins()
            
            error_msg = (
                "RadiationModel requires the 'radiation' plugin which is not available.\n\n"
                "The radiation plugin provides GPU-accelerated ray tracing with runtime\n"
                "backend auto-detection (OptiX 8 -> OptiX 6 -> Vulkan).\n"
                "System requirements (at least one backend):\n"
                "- Vulkan: Vulkan loader library (macOS/Linux); no extra packages on Windows\n"
                "- OptiX 8.1: NVIDIA GPU with driver >= 560 and CUDA 12.0+\n"
                "- OptiX 6.5: NVIDIA GPU with driver < 560 and CUDA 9.0+\n\n"
                "To enable radiation modeling:\n"
                "1. Build PyHelios with radiation plugin:\n"
                "   build_scripts/build_helios --plugins radiation\n"
                "2. Or build with multiple plugins:\n"
                "   build_scripts/build_helios --plugins radiation,visualizer,weberpenntree\n"
                f"\nCurrently available plugins: {available_plugins}"
            )
            
            # Suggest alternatives if available
            alternatives = registry.suggest_alternatives('radiation')
            if alternatives:
                error_msg += f"\n\nAlternative plugins available: {alternatives}"
                error_msg += "\nConsider using energybalance or leafoptics for thermal modeling."
            
            raise RadiationModelError(error_msg)
        
        # Plugin is available - create radiation model using working directory context manager
        try:
            with _radiation_working_directory():
                self.radiation_model = radiation_wrapper.createRadiationModel(context.getNativePtr())
                if self.radiation_model is None:
                    raise RadiationModelError(
                        "Failed to create RadiationModel instance. "
                        "This may indicate a problem with the native library or GPU initialization."
                    )
            logger.info("RadiationModel created successfully")
 
            # Bands created with explicit wavelength bounds. Tracked so cameras
            # can warn about the bounded-band + spectral-response combination
            # (see _warn_if_bounded_bands_with_camera_response).
            self._bounded_bands = set()
 
        except Exception as e:
            raise RadiationModelError(f"Failed to initialize RadiationModel: {e}")
    
    def _check_context_alive(self):
        """Raise if the owning Context has been destroyed (see Context.check_context_alive)."""
        check_context_alive(getattr(self, "context", None), "RadiationModel")
 
    def _warn_if_bounded_bands_with_camera_response(self, camera_label: str,
                                                    band_labels: List[str],
                                                    operation: str):
        """Warn when bands have explicit wavelength bounds and the camera has a
        spectral response.
 
        The two combined silently produce wrong colors. Helios computes camera
        pixels by integrating surface reflectance against the camera's spectral
        response curve, but computes SCATTERED flux by integrating over the
        band's wavelength bounds instead. When bounds are absent Helios
        deliberately falls back to the camera-weighted average for scattering,
        so both paths agree; setting bounds breaks that agreement.
 
        With scattering enabled most camera-visible light is scattered, so the
        image carries band-limited color the camera never expected. Low-chroma
        natural surfaces are hit hardest: a measured soil spectrum whose true
        red/green ratio is 1.2 renders at 2.6 -- brown soil comes out pink --
        while saturated colorboard-style spectra survive and look fine, which
        makes the problem easy to miss.
 
        Nothing errors and no output looks obviously broken, so this is a
        warning at the point the camera is created rather than a failure.
        """
        bounded = [b for b in band_labels if b in getattr(self, "_bounded_bands", ())]
        if not bounded:
            return
 
        logger.warning(
            f"Camera '{camera_label}' ({operation}) uses a spectral response, but "
            f"band(s) {bounded} were created with explicit wavelength bounds. "
            f"Camera pixels integrate reflectance against the spectral response "
            f"while scattered flux integrates over the band bounds, so rendered "
            f"colors will be skewed (brown surfaces tend to render pink). Create "
            f"these bands without wavelength_min/wavelength_max -- the camera's "
            f"spectral response defines what each band measures."
        )
 
    def _check_camera_has_pixel_data(self, camera: str, bands: List[str], operation: str):
        """Raise an actionable error if a camera/band has no rendered pixel data.
 
        A camera's pixel data is populated only by ``runBand()``, and only for
        the bands passed to that call and for cameras that already existed when
        it ran. Requesting an image for an unrendered camera/band otherwise
        reaches the native layer as a bare ``invalid map<K, T> key`` from
        ``std::map::at`` -- see GitHub issue #4. This preflight turns that into a
        message naming the camera, the band, and the call the user is missing.
 
        This check is load-bearing rather than merely cosmetic, and must not be
        removed on the grounds that helios-core v1.3.79 addressed the same case
        upstream. The upstream change makes the native call return an empty
        filename instead of throwing, so without this preflight an unrendered
        camera would look like a successful write that produced no file.
        :meth:`_check_image_was_written` is the backstop for the failure modes
        this preflight cannot see, such as an unwritable output directory.
        """
        try:
            known_cameras = radiation_wrapper.getAllCameraLabels(self.radiation_model)
        except Exception:
            # Camera enumeration is only used to sharpen the message. If it is
            # unavailable, fall through and let the per-band probe report.
            known_cameras = None
 
        if known_cameras is not None and camera not in known_cameras:
            raise RadiationModelError(
                f"Cannot {operation}: camera '{camera}' does not exist. "
                f"Add it with addRadiationCamera() before calling runBand(). "
                f"Existing cameras: {sorted(known_cameras) if known_cameras else 'none'}"
            )
 
        for band in bands:
            try:
                radiation_wrapper.getCameraPixelData(self.radiation_model, camera, band)
            except Exception:
                raise RadiationModelError(
                    f"Cannot {operation}: camera '{camera}' has no rendered pixel data "
                    f"for band '{band}'. Call runBand() with this band after adding the "
                    f"camera -- e.g. runBand({list(bands)!r}). Note that runBand() only "
                    f"renders the bands passed to it, and only for cameras that already "
                    f"exist when it runs."
                ) from None
 
    def _check_image_was_written(self, filename: str, camera: str, bands: List[str],
                                 image_path: str, operation: str):
        """Raise if the native layer reported a failed write via an empty filename.
 
        helios-core v1.3.79 changed ``writeCameraImage``/``writeNormCameraImage`` to
        return an empty string on failure rather than throwing, so no exception and no
        error code reaches Python. Returning that empty string to the caller -- or
        logging it as "written to: " -- would present a write that produced no file as
        a success, which the fail-fast policy forbids.
 
        :meth:`_check_camera_has_pixel_data` already rejects the unrendered-camera and
        unrendered-band cases before the native call, so by the time an empty filename
        gets here the likeliest cause is an output directory that does not exist or
        cannot be written.
        """
        if filename:
            return
 
        raise RadiationModelError(
            f"Cannot {operation}: the native library reported a failed write for camera "
            f"'{camera}' (bands {list(bands)!r}) by returning an empty filename, so no "
            f"image file was created. The camera has rendered pixel data, so the most "
            f"likely cause is that the output directory '{image_path}' does not exist "
            f"or is not writable. Verify the path exists and is writable."
        )
 
    def __enter__(self):
        """Context manager entry."""
        return self
 
    def __exit__(self, exc_type, exc_value, traceback):
        """Context manager exit with proper cleanup."""
        if self.radiation_model is not None:
            try:
                radiation_wrapper.destroyRadiationModel(self.radiation_model)
                logger.debug("RadiationModel destroyed successfully")
            except Exception as e:
                logger.warning(f"Error destroying RadiationModel: {e}")
            finally:
                self.radiation_model = None  # Prevent double deletion
 
    def __del__(self):
        """Destructor to ensure GPU resources freed even without 'with' statement."""
        if hasattr(self, 'radiation_model') and self.radiation_model is not None:
            try:
                radiation_wrapper.destroyRadiationModel(self.radiation_model)
                self.radiation_model = None
            except Exception as e:
                import warnings
                warnings.warn(f"Error in RadiationModel.__del__: {e}")
 
    def get_native_ptr(self):
        """Get native pointer for advanced operations."""
        return self.radiation_model
    
    def getNativePtr(self):
        """Get native pointer for advanced operations. (Legacy naming for compatibility)"""
        return self.get_native_ptr()
    
    @require_plugin('radiation', 'disable status messages')
    def disableMessages(self):
        """Disable RadiationModel status messages."""
        self._check_context_alive()
        radiation_wrapper.disableMessages(self.radiation_model)
    
    @require_plugin('radiation', 'enable status messages')
    def enableMessages(self):
        """Enable RadiationModel status messages."""
        self._check_context_alive()
        radiation_wrapper.enableMessages(self.radiation_model)
    
    @require_plugin('radiation', 'add radiation band')
    def addRadiationBand(self, band_label: str, wavelength_min: float = None, wavelength_max: float = None):
        """
        Add radiation band with optional wavelength bounds.
        
        Args:
            band_label: Name/label for the radiation band
            wavelength_min: Optional minimum wavelength (nm)
            wavelength_max: Optional maximum wavelength (nm)
        """
        # Validate inputs
        validate_band_label(band_label, "band_label", "addRadiationBand")
        if wavelength_min is not None and wavelength_max is not None:
            validate_wavelength_range(wavelength_min, wavelength_max, "wavelength_min", "wavelength_max", "addRadiationBand")
            self._check_context_alive()
            radiation_wrapper.addRadiationBandWithWavelengths(self.radiation_model, band_label, wavelength_min, wavelength_max)
            self._bounded_bands.add(band_label)
            logger.debug(f"Added radiation band {band_label}: {wavelength_min}-{wavelength_max} nm")
        else:
            self._check_context_alive()
            radiation_wrapper.addRadiationBand(self.radiation_model, band_label)
            self._bounded_bands.discard(band_label)
            logger.debug(f"Added radiation band: {band_label}")
    
    @require_plugin('radiation', 'copy radiation band')
    @validate_radiation_band_params
    def copyRadiationBand(self, old_label: str, new_label: str, wavelength_min: float = None, wavelength_max: float = None):
        """
        Copy existing radiation band to new label, optionally with new wavelength range.
 
        Args:
            old_label: Existing band label to copy
            new_label: New label for the copied band
            wavelength_min: Optional minimum wavelength for new band (nm)
            wavelength_max: Optional maximum wavelength for new band (nm)
 
        Example:
            >>> # Copy band with same wavelength range
            >>> radiation.copyRadiationBand("SW", "SW_copy")
            >>>
            >>> # Copy band with different wavelength range
            >>> radiation.copyRadiationBand("full_spectrum", "PAR", 400, 700)
        """
        if wavelength_min is not None and wavelength_max is not None:
            validate_wavelength_range(wavelength_min, wavelength_max, "wavelength_min", "wavelength_max", "copyRadiationBand")
 
        self._check_context_alive()
        radiation_wrapper.copyRadiationBand(self.radiation_model, old_label, new_label, wavelength_min, wavelength_max)
        if wavelength_min is not None:
            self._bounded_bands.add(new_label)
            logger.debug(f"Copied radiation band {old_label} to {new_label} with wavelengths {wavelength_min}-{wavelength_max} nm")
        else:
            # Without explicit bounds the copy inherits the source band's bounds.
            if old_label in self._bounded_bands:
                self._bounded_bands.add(new_label)
            else:
                self._bounded_bands.discard(new_label)
            logger.debug(f"Copied radiation band {old_label} to {new_label}")
    
    @require_plugin('radiation', 'add radiation source')
    @validate_collimated_source_params
    def addCollimatedRadiationSource(self, direction=None) -> int:
        """
        Add collimated radiation source.
        
        Args:
            direction: Optional direction vector. Can be tuple (x, y, z), vec3, or None for default direction.
            
        Returns:
            Source ID
        """
        if direction is None:
            self._check_context_alive()
            source_id = radiation_wrapper.addCollimatedRadiationSourceDefault(self.radiation_model)
        else:
            # Handle vec3, SphericalCoord, and tuple types
            if hasattr(direction, 'x') and hasattr(direction, 'y') and hasattr(direction, 'z'):
                # vec3-like object
                x, y, z = direction.x, direction.y, direction.z
            elif hasattr(direction, 'radius') and hasattr(direction, 'elevation') and hasattr(direction, 'azimuth'):
                # SphericalCoord object - convert to Cartesian
                import math
                r = direction.radius
                elevation = direction.elevation
                azimuth = direction.azimuth
                x = r * math.cos(elevation) * math.cos(azimuth)
                y = r * math.cos(elevation) * math.sin(azimuth)
                z = r * math.sin(elevation)
            else:
                # Assume tuple-like object - validate it first
                self._check_context_alive()
                try:
                    if len(direction) != 3:
                        raise TypeError(f"Direction must be a 3-element tuple, vec3, or SphericalCoord, got {type(direction).__name__} with {len(direction)} elements")
                    x, y, z = direction
                except (TypeError, AttributeError):
                    # Not a valid sequence type
                    raise TypeError(f"Direction must be a tuple, vec3, or SphericalCoord, got {type(direction).__name__}")
            source_id = radiation_wrapper.addCollimatedRadiationSourceVec3(self.radiation_model, x, y, z)
        
        logger.debug(f"Added collimated radiation source: ID {source_id}")
        return source_id
    
    @require_plugin('radiation', 'add spherical radiation source')
    @validate_sphere_source_params
    def addSphereRadiationSource(self, position, radius: float) -> int:
        """
        Add spherical radiation source.
        
        Args:
            position: Position of the source. Can be tuple (x, y, z) or vec3.
            radius: Radius of the spherical source
            
        Returns:
            Source ID
        """
        validate_position_like(position, "position", "addSphereRadiationSource")
        # Handle both tuple and vec3 types
        if hasattr(position, 'x') and hasattr(position, 'y') and hasattr(position, 'z'):
            x, y, z = position.x, position.y, position.z
        else:
            x, y, z = position
        self._check_context_alive()
        source_id = radiation_wrapper.addSphereRadiationSource(self.radiation_model, x, y, z, radius)
        logger.debug(f"Added sphere radiation source: ID {source_id} at ({x}, {y}, {z}) with radius {radius}")
        return source_id
    
    @require_plugin('radiation', 'add sun radiation source')
    @validate_sun_sphere_params
    def addSunSphereRadiationSource(self, radius: float, zenith: float, azimuth: float,
                                    position_scaling: float = 1.0, angular_width: float = 0.53,
                                    flux_scaling: float = 1.0) -> int:
        """
        Add sun sphere radiation source.
        
        Args:
            radius: Radius of the sun sphere
            zenith: Zenith angle (degrees)
            azimuth: Azimuth angle (degrees)
            position_scaling: Position scaling factor
            angular_width: Angular width of the sun (degrees)
            flux_scaling: Flux scaling factor
            
        Returns:
            Source ID
        """
        self._check_context_alive()
        source_id = radiation_wrapper.addSunSphereRadiationSource(
            self.radiation_model, radius, zenith, azimuth, position_scaling, angular_width, flux_scaling
        )
        logger.debug(f"Added sun radiation source: ID {source_id}")
        return source_id
 
    @require_plugin('radiation', 'set source position')
    def setSourcePosition(self, source_id: int, position):
        """
        Set position of a radiation source.
 
        Allows dynamic repositioning of radiation sources during simulation,
        useful for time-series modeling or moving light sources.
 
        Args:
            source_id: ID of the radiation source
            position: New position as vec3, SphericalCoord, or list/tuple [x, y, z]
 
        Example:
            >>> source_id = radiation.addCollimatedRadiationSource()
            >>> radiation.setSourcePosition(source_id, [10, 20, 30])
            >>> from pyhelios.types import vec3
            >>> radiation.setSourcePosition(source_id, vec3(15, 25, 35))
        """
        if not isinstance(source_id, int) or source_id < 0:
            raise ValueError(f"Source ID must be a non-negative integer, got {source_id}")
        validate_direction_like(position, "position", "setSourcePosition")
        self._check_context_alive()
        radiation_wrapper.setSourcePosition(self.radiation_model, source_id, position)
        logger.debug(f"Updated position for radiation source {source_id}")
 
    @require_plugin('radiation', 'add rectangle radiation source')
    def addRectangleRadiationSource(self, position, size, rotation) -> int:
        """
        Add a rectangle (planar) radiation source.
 
        Rectangle sources are ideal for modeling artificial lighting such as
        LED panels, grow lights, or window light sources.
 
        Args:
            position: Center position as vec3 or list [x, y, z]
            size: Rectangle dimensions as vec2 or list [width, height]
            rotation: Rotation vector as vec3 or list [rx, ry, rz] (Euler angles in radians)
 
        Returns:
            Source ID
 
        Example:
            >>> from pyhelios.types import vec3, vec2
            >>> source_id = radiation.addRectangleRadiationSource(
            ...     position=vec3(0, 0, 5),
            ...     size=vec2(2, 1),
            ...     rotation=vec3(0, 0, 0)
            ... )
            >>> radiation.setSourceFlux(source_id, "PAR", 500.0)
        """
        validate_position_like(position, "position", "addRectangleRadiationSource")
        validate_size_like(size, "size", "addRectangleRadiationSource")
        validate_position_like(rotation, "rotation", "addRectangleRadiationSource")
        self._check_context_alive()
        return radiation_wrapper.addRectangleRadiationSource(self.radiation_model, position, size, rotation)
 
    @require_plugin('radiation', 'add disk radiation source')
    def addDiskRadiationSource(self, position, radius: float, rotation) -> int:
        """
        Add a disk (circular planar) radiation source.
 
        Disk sources are useful for modeling circular light sources such as
        spotlights, circular LED arrays, or solar simulators.
 
        Args:
            position: Center position as vec3 or list [x, y, z]
            radius: Disk radius
            rotation: Rotation vector as vec3 or list [rx, ry, rz] (Euler angles in radians)
 
        Returns:
            Source ID
 
        Example:
            >>> from pyhelios.types import vec3
            >>> source_id = radiation.addDiskRadiationSource(
            ...     position=vec3(0, 0, 5),
            ...     radius=1.5,
            ...     rotation=vec3(0, 0, 0)
            ... )
            >>> radiation.setSourceFlux(source_id, "PAR", 300.0)
        """
        validate_position_like(position, "position", "addDiskRadiationSource")
        validate_position_like(rotation, "rotation", "addDiskRadiationSource")
        if radius <= 0:
            raise ValueError(f"Radius must be positive, got {radius}")
        self._check_context_alive()
        return radiation_wrapper.addDiskRadiationSource(self.radiation_model, position, radius, rotation)
 
    # Source spectrum methods
    @require_plugin('radiation', 'manage source spectrum')
    def setSourceSpectrum(self, source_id, spectrum):
        """
        Set radiation spectrum for source(s).
 
        Spectral distributions define how radiation intensity varies with wavelength,
        essential for realistic modeling of different light sources (sunlight, LEDs, etc.).
 
        Args:
            source_id: Source ID (int) or list of source IDs
            spectrum: Either:
                - Spectrum data as list of (wavelength, value) tuples
                - Global data label string
 
        Example:
            >>> # Define custom LED spectrum
            >>> led_spectrum = [
            ...     (400, 0.0), (450, 0.3), (500, 0.8),
            ...     (550, 0.5), (600, 0.2), (700, 0.0)
            ... ]
            >>> radiation.setSourceSpectrum(source_id, led_spectrum)
            >>>
            >>> # Use predefined spectrum from global data
            >>> radiation.setSourceSpectrum(source_id, "D65_illuminant")
            >>>
            >>> # Apply same spectrum to multiple sources
            >>> radiation.setSourceSpectrum([src1, src2, src3], led_spectrum)
        """
        self._check_context_alive()
        radiation_wrapper.setSourceSpectrum(self.radiation_model, source_id, spectrum)
        logger.debug(f"Set spectrum for source(s) {source_id}")
 
    @require_plugin('radiation', 'configure source spectrum')
    def setSourceSpectrumIntegral(self, source_id: int, source_integral: float,
                                  wavelength_min: float = None, wavelength_max: float = None):
        """
        Set source spectrum integral value.
 
        Normalizes the spectrum so that its integral equals the specified value,
        useful for calibrating source intensity.
 
        Args:
            source_id: Source ID
            source_integral: Target integral value
            wavelength_min: Optional minimum wavelength for integration range
            wavelength_max: Optional maximum wavelength for integration range
 
        Example:
            >>> radiation.setSourceSpectrumIntegral(source_id, 1000.0)
            >>> radiation.setSourceSpectrumIntegral(source_id, 500.0, 400, 700)  # PAR range
        """
        if not isinstance(source_id, int) or source_id < 0:
            raise ValueError(f"Source ID must be a non-negative integer, got {source_id}")
        if source_integral < 0:
            raise ValueError(f"Source integral must be non-negative, got {source_integral}")
 
        self._check_context_alive()
        radiation_wrapper.setSourceSpectrumIntegral(self.radiation_model, source_id, source_integral,
                                                    wavelength_min, wavelength_max)
        logger.debug(f"Set spectrum integral for source {source_id}: {source_integral}")
 
    # Spectrum integration and analysis methods
    @require_plugin('radiation', 'integrate spectrum')
    def integrateSpectrum(self, object_spectrum, wavelength_min: float = None,
                         wavelength_max: float = None, source_id: int = None,
                         camera_spectrum=None) -> float:
        """
        Integrate spectrum with optional source/camera spectra and wavelength range.
 
        This unified method handles multiple integration scenarios:
        - Basic: Total spectrum integration
        - Range: Integration over wavelength range
        - Source: Integration weighted by source spectrum
        - Camera: Integration weighted by camera spectral response
        - Full: Integration with both source and camera spectra
 
        Args:
            object_spectrum: Object spectrum as list of (wavelength, value) tuples/vec2
            wavelength_min: Optional minimum wavelength for integration range
            wavelength_max: Optional maximum wavelength for integration range
            source_id: Optional source ID for source spectrum weighting
            camera_spectrum: Optional camera spectrum for camera response weighting
 
        Returns:
            Integrated value
 
        Example:
            >>> leaf_reflectance = [(400, 0.1), (500, 0.4), (600, 0.6), (700, 0.5)]
            >>>
            >>> # Total integration
            >>> total = radiation.integrateSpectrum(leaf_reflectance)
            >>>
            >>> # PAR range (400-700nm)
            >>> par = radiation.integrateSpectrum(leaf_reflectance, 400, 700)
            >>>
            >>> # With source spectrum
            >>> source_weighted = radiation.integrateSpectrum(
            ...     leaf_reflectance, 400, 700, source_id=sun_source
            ... )
            >>>
            >>> # With camera response
            >>> camera_response = [(400, 0.2), (550, 1.0), (700, 0.3)]
            >>> camera_weighted = radiation.integrateSpectrum(
            ...     leaf_reflectance, camera_spectrum=camera_response
            ... )
        """
        self._check_context_alive()
        return radiation_wrapper.integrateSpectrum(self.radiation_model, object_spectrum,
                                                  wavelength_min, wavelength_max,
                                                  source_id, camera_spectrum)
 
    @require_plugin('radiation', 'integrate source spectrum')
    def integrateSourceSpectrum(self, source_id: int, wavelength_min: float, wavelength_max: float) -> float:
        """
        Integrate source spectrum over wavelength range.
 
        Args:
            source_id: Source ID
            wavelength_min: Minimum wavelength
            wavelength_max: Maximum wavelength
 
        Returns:
            Integrated source spectrum value
 
        Example:
            >>> par_flux = radiation.integrateSourceSpectrum(source_id, 400, 700)
        """
        if not isinstance(source_id, int) or source_id < 0:
            raise ValueError(f"Source ID must be a non-negative integer, got {source_id}")
        self._check_context_alive()
        return radiation_wrapper.integrateSourceSpectrum(self.radiation_model, source_id,
                                                        wavelength_min, wavelength_max)
 
    # Spectral manipulation methods
    @require_plugin('radiation', 'scale spectrum')
    def scaleSpectrum(self, existing_label: str, new_label_or_scale, scale_factor: float = None):
        """
        Scale spectrum in-place or to new label.
 
        Useful for adjusting spectrum intensities or creating variations of
        existing spectra for sensitivity analysis.
 
        Supports two call patterns:
        - scaleSpectrum("label", scale) -> scales in-place
        - scaleSpectrum("existing", "new", scale) -> creates new scaled spectrum
 
        Args:
            existing_label: Existing global data label
            new_label_or_scale: Either new label string (if creating new) or scale factor (if in-place)
            scale_factor: Scale factor (required only if new_label_or_scale is a string)
 
        Example:
            >>> # In-place scaling
            >>> radiation.scaleSpectrum("leaf_reflectance", 1.2)
            >>>
            >>> # Create new scaled spectrum
            >>> radiation.scaleSpectrum("leaf_reflectance", "scaled_leaf", 1.5)
        """
        if not isinstance(existing_label, str) or not existing_label.strip():
            raise ValueError("Existing label must be a non-empty string")
 
        self._check_context_alive()
        radiation_wrapper.scaleSpectrum(self.radiation_model, existing_label,
                                       new_label_or_scale, scale_factor)
        logger.debug(f"Scaled spectrum '{existing_label}'")
 
    @require_plugin('radiation', 'scale spectrum randomly')
    def scaleSpectrumRandomly(self, existing_label: str, new_label: str,
                             min_scale: float, max_scale: float):
        """
        Scale spectrum with random factor and store as new label.
 
        Useful for creating stochastic variations in spectral properties for
        Monte Carlo simulations or uncertainty quantification.
 
        Args:
            existing_label: Existing global data label
            new_label: New global data label for scaled spectrum
            min_scale: Minimum scale factor
            max_scale: Maximum scale factor
 
        Example:
            >>> # Create random variation of leaf reflectance
            >>> radiation.scaleSpectrumRandomly("leaf_base", "leaf_variant", 0.8, 1.2)
        """
        if not isinstance(existing_label, str) or not existing_label.strip():
            raise ValueError("Existing label must be a non-empty string")
        if not isinstance(new_label, str) or not new_label.strip():
            raise ValueError("New label must be a non-empty string")
        if min_scale >= max_scale:
            raise ValueError(f"min_scale ({min_scale}) must be less than max_scale ({max_scale})")
 
        self._check_context_alive()
        radiation_wrapper.scaleSpectrumRandomly(self.radiation_model, existing_label, new_label,
                                               min_scale, max_scale)
        logger.debug(f"Scaled spectrum '{existing_label}' randomly to '{new_label}'")
 
    @require_plugin('radiation', 'blend spectra')
    def blendSpectra(self, new_label: str, spectrum_labels: List[str], weights: List[float]):
        """
        Blend multiple spectra with specified weights.
 
        Creates weighted combination of spectra, useful for mixing material properties
        or creating composite light sources.
 
        Args:
            new_label: New global data label for blended spectrum
            spectrum_labels: List of spectrum labels to blend
            weights: List of weights (must sum to reasonable values, same length as labels)
 
        Example:
            >>> # Mix two leaf types (70% type A, 30% type B)
            >>> radiation.blendSpectra("mixed_leaf",
            ...     ["leaf_type_a", "leaf_type_b"],
            ...     [0.7, 0.3]
            ... )
        """
        if not isinstance(new_label, str) or not new_label.strip():
            raise ValueError("New label must be a non-empty string")
        if len(spectrum_labels) != len(weights):
            raise ValueError(f"Number of labels ({len(spectrum_labels)}) must match number of weights ({len(weights)})")
        if not spectrum_labels:
            raise ValueError("At least one spectrum label required")
 
        self._check_context_alive()
        radiation_wrapper.blendSpectra(self.radiation_model, new_label, spectrum_labels, weights)
        logger.debug(f"Blended {len(spectrum_labels)} spectra into '{new_label}'")
 
    @require_plugin('radiation', 'blend spectra randomly')
    def blendSpectraRandomly(self, new_label: str, spectrum_labels: List[str]):
        """
        Blend multiple spectra with random weights.
 
        Creates random combinations of spectra, useful for generating diverse
        material properties in stochastic simulations.
 
        Args:
            new_label: New global data label for blended spectrum
            spectrum_labels: List of spectrum labels to blend
 
        Example:
            >>> # Create random mixture of leaf spectra
            >>> radiation.blendSpectraRandomly("random_leaf",
            ...     ["young_leaf", "mature_leaf", "senescent_leaf"]
            ... )
        """
        if not isinstance(new_label, str) or not new_label.strip():
            raise ValueError("New label must be a non-empty string")
        if not spectrum_labels:
            raise ValueError("At least one spectrum label required")
 
        self._check_context_alive()
        radiation_wrapper.blendSpectraRandomly(self.radiation_model, new_label, spectrum_labels)
        logger.debug(f"Blended {len(spectrum_labels)} spectra randomly into '{new_label}'")
 
    # Spectral interpolation methods
    @require_plugin('radiation', 'interpolate spectrum from data')
    def interpolateSpectrumFromPrimitiveData(self, primitive_uuids: List[int],
                                            spectra_labels: List[str], values: List[float],
                                            primitive_data_query_label: str,
                                            primitive_data_radprop_label: str):
        """
        Interpolate spectral properties based on primitive data values.
 
        Automatically assigns spectra to primitives by interpolating between
        reference spectra based on continuous data values (e.g., age, moisture, etc.).
 
        Args:
            primitive_uuids: List of primitive UUIDs to assign spectra
            spectra_labels: List of reference spectrum labels
            values: List of data values corresponding to each spectrum
            primitive_data_query_label: Primitive data label containing query values
            primitive_data_radprop_label: Primitive data label to store assigned spectra
 
        Example:
            >>> # Assign leaf reflectance based on age
            >>> leaf_patches = context.getAllUUIDs("patch")
            >>> radiation.interpolateSpectrumFromPrimitiveData(
            ...     primitive_uuids=leaf_patches,
            ...     spectra_labels=["young_leaf", "mature_leaf", "old_leaf"],
            ...     values=[0.0, 50.0, 100.0],  # Days since emergence
            ...     primitive_data_query_label="leaf_age",
            ...     primitive_data_radprop_label="reflectance"
            ... )
        """
        if not isinstance(primitive_uuids, (list, tuple)) or not primitive_uuids:
            raise ValueError("Primitive UUIDs must be a non-empty list")
        if not isinstance(spectra_labels, (list, tuple)) or not spectra_labels:
            raise ValueError("Spectra labels must be a non-empty list")
        if not isinstance(values, (list, tuple)) or not values:
            raise ValueError("Values must be a non-empty list")
        if len(spectra_labels) != len(values):
            raise ValueError(f"Number of spectra ({len(spectra_labels)}) must match number of values ({len(values)})")
 
        self._check_context_alive()
        radiation_wrapper.interpolateSpectrumFromPrimitiveData(
            self.radiation_model, primitive_uuids, spectra_labels, values,
            primitive_data_query_label, primitive_data_radprop_label
        )
        logger.debug(f"Interpolated spectra for {len(primitive_uuids)} primitives")
 
    @require_plugin('radiation', 'interpolate spectrum from object data')
    def interpolateSpectrumFromObjectData(self, object_ids: List[int],
                                         spectra_labels: List[str], values: List[float],
                                         object_data_query_label: str,
                                         primitive_data_radprop_label: str):
        """
        Interpolate spectral properties based on object data values.
 
        Automatically assigns spectra to object primitives by interpolating between
        reference spectra based on continuous object-level data values.
 
        Args:
            object_ids: List of object IDs
            spectra_labels: List of reference spectrum labels
            values: List of data values corresponding to each spectrum
            object_data_query_label: Object data label containing query values
            primitive_data_radprop_label: Primitive data label to store assigned spectra
 
        Example:
            >>> # Assign tree reflectance based on health index
            >>> tree_ids = [tree1_id, tree2_id, tree3_id]
            >>> radiation.interpolateSpectrumFromObjectData(
            ...     object_ids=tree_ids,
            ...     spectra_labels=["healthy_tree", "stressed_tree", "diseased_tree"],
            ...     values=[1.0, 0.5, 0.0],  # Health index
            ...     object_data_query_label="health_index",
            ...     primitive_data_radprop_label="reflectance"
            ... )
        """
        if not isinstance(object_ids, (list, tuple)) or not object_ids:
            raise ValueError("Object IDs must be a non-empty list")
        if not isinstance(spectra_labels, (list, tuple)) or not spectra_labels:
            raise ValueError("Spectra labels must be a non-empty list")
        if not isinstance(values, (list, tuple)) or not values:
            raise ValueError("Values must be a non-empty list")
        if len(spectra_labels) != len(values):
            raise ValueError(f"Number of spectra ({len(spectra_labels)}) must match number of values ({len(values)})")
 
        self._check_context_alive()
        radiation_wrapper.interpolateSpectrumFromObjectData(
            self.radiation_model, object_ids, spectra_labels, values,
            object_data_query_label, primitive_data_radprop_label
        )
        logger.debug(f"Interpolated spectra for {len(object_ids)} objects")
 
    @require_plugin('radiation', 'set ray count')
    def setDirectRayCount(self, band_label: str, ray_count: int):
        """Set direct ray count for radiation band."""
        validate_band_label(band_label, "band_label", "setDirectRayCount")
        validate_ray_count(ray_count, "ray_count", "setDirectRayCount")
        self._check_context_alive()
        radiation_wrapper.setDirectRayCount(self.radiation_model, band_label, ray_count)
    
    @require_plugin('radiation', 'set ray count')
    def setDiffuseRayCount(self, band_label: str, ray_count: int):
        """Set diffuse ray count for radiation band."""
        validate_band_label(band_label, "band_label", "setDiffuseRayCount")
        validate_ray_count(ray_count, "ray_count", "setDiffuseRayCount")
        self._check_context_alive()
        radiation_wrapper.setDiffuseRayCount(self.radiation_model, band_label, ray_count)
    
    @require_plugin('radiation', 'set radiation flux')
    def setDiffuseRadiationFlux(self, label: str, flux: float):
        """Set diffuse radiation flux for band."""
        validate_band_label(label, "label", "setDiffuseRadiationFlux")
        validate_flux_value(flux, "flux", "setDiffuseRadiationFlux")
        self._check_context_alive()
        radiation_wrapper.setDiffuseRadiationFlux(self.radiation_model, label, flux)
 
    @require_plugin('radiation', 'configure diffuse radiation')
    def setDiffuseRadiationExtinctionCoeff(self, label: str, K: float, peak_direction):
        """
        Set diffuse radiation extinction coefficient with directional bias.
 
        Models directionally-biased diffuse radiation (e.g., sky radiation with zenith peak).
 
        Args:
            label: Band label
            K: Extinction coefficient
            peak_direction: Peak direction as vec3, SphericalCoord, or list [x, y, z]
 
        Example:
            >>> from pyhelios.types import vec3
            >>> radiation.setDiffuseRadiationExtinctionCoeff("SW", 0.5, vec3(0, 0, 1))
        """
        validate_band_label(label, "label", "setDiffuseRadiationExtinctionCoeff")
        if K < 0:
            raise ValueError(f"Extinction coefficient must be non-negative, got {K}")
        validate_direction_like(peak_direction, "peak_direction", "setDiffuseRadiationExtinctionCoeff")
        self._check_context_alive()
        radiation_wrapper.setDiffuseRadiationExtinctionCoeff(self.radiation_model, label, K, peak_direction)
        logger.debug(f"Set diffuse extinction coefficient for band '{label}': K={K}")
 
    @require_plugin('radiation', 'query diffuse flux')
    def getDiffuseFlux(self, band_label: str) -> float:
        """
        Get diffuse flux for band.
 
        Args:
            band_label: Band label
 
        Returns:
            Diffuse flux value
 
        Example:
            >>> flux = radiation.getDiffuseFlux("SW")
        """
        validate_band_label(band_label, "band_label", "getDiffuseFlux")
        self._check_context_alive()
        return radiation_wrapper.getDiffuseFlux(self.radiation_model, band_label)
 
    @require_plugin('radiation', 'configure diffuse spectrum')
    def setDiffuseSpectrum(self, band_label, spectrum_label: str):
        """
        Set diffuse spectrum from global data label.
 
        Args:
            band_label: Band label (string) or list of band labels
            spectrum_label: Spectrum global data label
 
        Example:
            >>> radiation.setDiffuseSpectrum("SW", "sky_spectrum")
            >>> radiation.setDiffuseSpectrum(["SW", "NIR"], "sky_spectrum")
        """
        if isinstance(band_label, str):
            validate_band_label(band_label, "band_label", "setDiffuseSpectrum")
        else:
            for label in band_label:
                validate_band_label(label, "band_label", "setDiffuseSpectrum")
        if not isinstance(spectrum_label, str) or not spectrum_label.strip():
            raise ValueError("Spectrum label must be a non-empty string")
 
        self._check_context_alive()
        radiation_wrapper.setDiffuseSpectrum(self.radiation_model, band_label, spectrum_label)
        logger.debug(f"Set diffuse spectrum for band(s) {band_label}")
 
    @require_plugin('radiation', 'configure diffuse spectrum')
    def setDiffuseSpectrumIntegral(self, spectrum_integral: float, wavelength_min: float = None,
                                   wavelength_max: float = None, band_label: str = None):
        """
        Set diffuse spectrum integral.
 
        Args:
            spectrum_integral: Integral value
            wavelength_min: Optional minimum wavelength
            wavelength_max: Optional maximum wavelength
            band_label: Optional specific band label (None for all bands)
 
        Example:
            >>> radiation.setDiffuseSpectrumIntegral(1000.0)  # All bands
            >>> radiation.setDiffuseSpectrumIntegral(500.0, 400, 700, band_label="PAR")  # Specific band
        """
        if spectrum_integral < 0:
            raise ValueError(f"Spectrum integral must be non-negative, got {spectrum_integral}")
        if band_label is not None:
            validate_band_label(band_label, "band_label", "setDiffuseSpectrumIntegral")
 
        self._check_context_alive()
        radiation_wrapper.setDiffuseSpectrumIntegral(self.radiation_model, spectrum_integral,
                                                     wavelength_min, wavelength_max, band_label)
        logger.debug(f"Set diffuse spectrum integral: {spectrum_integral}")
 
    @require_plugin('radiation', 'set source flux')
    def setSourceFlux(self, source_id, label: str, flux: float):
        """Set source flux for single source or multiple sources."""
        validate_band_label(label, "label", "setSourceFlux")
        validate_flux_value(flux, "flux", "setSourceFlux")
        
        if isinstance(source_id, (list, tuple)):
            # Multiple sources
            validate_source_id_list(list(source_id), "source_id", "setSourceFlux")
            self._check_context_alive()
            radiation_wrapper.setSourceFluxMultiple(self.radiation_model, source_id, label, flux)
        else:
            # Single source
            validate_source_id(source_id, "source_id", "setSourceFlux")
            self._check_context_alive()
            radiation_wrapper.setSourceFlux(self.radiation_model, source_id, label, flux)
    
    
    @require_plugin('radiation', 'get source flux')
    @validate_get_source_flux_params
    def getSourceFlux(self, source_id: int, label: str) -> float:
        """Get source flux for band."""
        self._check_context_alive()
        return radiation_wrapper.getSourceFlux(self.radiation_model, source_id, label)
    
    @require_plugin('radiation', 'update geometry')
    @validate_update_geometry_params
    def updateGeometry(self, uuids: Optional[List[int]] = None):
        """
        Update geometry in radiation model.
 
        The two call forms differ in more than scope, as of helios-core v1.3.79:
 
        - ``updateGeometry()`` (no argument) is **optional**. :meth:`runBand` builds
          the geometry itself before tracing and rebuilds it whenever primitives have
          been added to or deleted from the Context since the last build. Call it
          explicitly only to control when the cost of the build is paid.
        - ``updateGeometry(uuids)`` restricts the model to a subset of Context
          primitives, and that subset is **never** rebuilt automatically -- doing so
          would silently discard the subset you asked for. You must call it again
          yourself after modifying Context geometry.
 
        Args:
            uuids: Optional list of specific UUIDs to update. If None, updates all geometry.
 
        Note:
            Passing ``uuids`` latches the model into subset mode natively. Calling
            ``updateGeometry()`` with no argument afterwards clears that latch and
            returns the model to tracking the full Context.
        """
        if uuids is None:
            self._check_context_alive()
            radiation_wrapper.updateGeometry(self.radiation_model)
            logger.debug("Updated all geometry in radiation model")
            # Clears the native subset latch, so stop treating the model as a subset.
            self._geometry_subset = False
        else:
            self._check_context_alive()
            radiation_wrapper.updateGeometryUUIDs(self.radiation_model, uuids)
            logger.debug(f"Updated {len(uuids)} geometry UUIDs in radiation model")
            self._geometry_subset = True
        self._geometry_updated = True
    
    @require_plugin('radiation', 'run radiation simulation')
    @validate_run_band_params
    def runBand(self, band_label):
        """
        Run radiation simulation for single band or multiple bands.
        
        PERFORMANCE NOTE: When simulating multiple radiation bands, it is HIGHLY RECOMMENDED
        to run all bands in a single call (e.g., runBand(["PAR", "NIR", "SW"])) rather than
        sequential single-band calls. This provides significant computational efficiency gains
        because:
        
        - GPU ray tracing setup is done once for all bands
        - Scene geometry acceleration structures are reused
        - GPU kernel launches are batched together
        - Memory transfers between CPU/GPU are minimized
        
        Example:
            # EFFICIENT - Single call for multiple bands
            radiation.runBand(["PAR", "NIR", "SW"])
            
            # INEFFICIENT - Sequential single-band calls  
            radiation.runBand("PAR")
            radiation.runBand("NIR") 
            radiation.runBand("SW")
        
        Args:
            band_label: Single band name (str) or list of band names for multi-band simulation
 
        Note:
            Geometry and radiative properties are updated automatically as needed
            (helios-core v1.3.79+), so an explicit :meth:`updateGeometry` call is not
            required first. A subset build from ``updateGeometry(uuids)`` is preserved:
            it is never rebuilt automatically, since that would discard the subset.
        """
        if isinstance(band_label, (list, tuple)):
            # Multiple bands - validate each label
            for lbl in band_label:
                if not isinstance(lbl, str):
                    raise TypeError(f"Band labels must be strings, got {type(lbl).__name__}")
            self._check_context_alive()
            radiation_wrapper.runBandMultiple(self.radiation_model, band_label)
            logger.info(f"Completed radiation simulation for bands: {band_label}")
        else:
            # Single band - validate label type
            if not isinstance(band_label, str):
                raise TypeError(f"Band label must be a string, got {type(band_label).__name__}")
            self._check_context_alive()
            radiation_wrapper.runBand(self.radiation_model, band_label)
            logger.info(f"Completed radiation simulation for band: {band_label}")
        # runBand() builds geometry itself when needed, so geometry is now loaded even
        # if updateGeometry() was never called. Recording that keeps the auto-update in
        # calculateGtheta() from doing a redundant rebuild -- which, after a subset
        # build, would silently widen the model back to the full Context.
        self._geometry_updated = True
    
    
    @require_plugin('radiation', 'get simulation results')
    def getTotalAbsorbedFlux(self) -> List[float]:
        """Get absorbed radiation flux density for all primitives, summed over all bands.
 
        .. warning::
           **The returned values cannot be matched to UUIDs by position.** They are
           ordered by the radiation model's internal primitive ordering (grouped by
           parent object, built during :meth:`updateGeometry`), which is *not* the
           order of ``context.getAllUUIDs()`` -- that iterates an unordered map and
           returns a hash order. Pairing the two by index silently attributes flux
           to the wrong primitive. Use :meth:`getAbsorbedFlux` instead, which is
           keyed by UUID.
 
        .. warning::
           This sums the ``radiation_flux_<band>`` primitive data over **every**
           band registered in the model. For overlapping bands -- e.g. PAR, NIR
           and SW, where SW already spans the other two -- the sum double-counts
           and is not a physically meaningful quantity. Use
           :meth:`getAbsorbedFlux` to get a single band's flux.
 
        Units are **W/m^2** (flux density), not watts. Because it is a density, the
        value does not change when a primitive's size changes: a 1x1 m and a 2x2 m
        patch under the same collimated source both report the same number.
 
        Summing the returned values directly (``sum(flux)``) adds flux densities of
        differently-sized surfaces and is not physically meaningful. To obtain
        absorbed power in watts, weight each primitive by its area -- via
        :meth:`getAbsorbedFlux`, so that flux and area refer to the same primitive::
 
            uuids = context.getAllUUIDs()
            flux = radiation.getAbsorbedFlux("SW")
            power = float((flux * context.getPrimitiveArea(uuids)).sum(dtype="float64"))
 
        Returns:
            Absorbed flux density per primitive in W/m^2, summed over all bands.
 
        See Also:
            :meth:`getAbsorbedFlux`: band-specific, UUID-aligned absorbed flux.
        """
        self._check_context_alive()
        results = radiation_wrapper.getTotalAbsorbedFlux(self.radiation_model)
        logger.debug(f"Retrieved absorbed flux data for {len(results)} primitives")
        return results
 
    @require_plugin('radiation', 'get band-specific simulation results')
    def getAbsorbedFlux(self, band_label: Union[str, List[str]],
                        uuids: Optional[List[int]] = None
                        ) -> Union["np.ndarray", Dict[str, "np.ndarray"]]:
        """Get per-band absorbed radiation flux density, aligned to UUIDs.
 
        ``runBand()`` stores each band's result as ``radiation_flux_<band>``
        primitive data: the **total** absorbed flux density for that band, i.e.
        direct plus diffuse plus scattered contributions. This method reads that
        data back keyed by UUID, so element ``i`` of the result always belongs to
        ``uuids[i]``.
 
        Prefer this over :meth:`getTotalAbsorbedFlux`, which sums over every band
        (double-counting overlapping bands such as PAR/NIR/SW) and whose ordering
        does not correspond to ``context.getAllUUIDs()``.
 
        Args:
            band_label: Band name, or a list of band names to fetch at once.
            uuids: Primitives to query, defaulting to ``context.getAllUUIDs()``.
                Results follow this list's order.
 
        Returns:
            For a single band, a float32 array of absorbed flux density per
            primitive in W/m^2. For a list of bands, a dict mapping each band
            label to that array.
 
        Raises:
            RadiationModelError: If a band does not exist, or if ``runBand()`` has
                not been run for it.
 
        Example:
            >>> radiation.runBand(["PAR", "NIR", "SW"])
            >>> uuids = context.getAllUUIDs()
            >>> par = radiation.getAbsorbedFlux("PAR")          # W/m^2, PAR only
            >>> par_watts = par * context.getPrimitiveArea(uuids)
            >>> all_bands = radiation.getAbsorbedFlux(["PAR", "NIR", "SW"])
            >>> all_bands["SW"][0]                              # doctest: +SKIP
        """
        single_band = isinstance(band_label, str)
        labels = [band_label] if single_band else band_label
 
        if not single_band:
            if not isinstance(labels, (list, tuple)):
                raise TypeError(
                    f"band_label must be a string or list of strings, "
                    f"got {type(band_label).__name__}"
                )
            if not labels:
                raise ValueError("band_label list cannot be empty")
        for label in labels:
            validate_band_label(label, "band_label", "getAbsorbedFlux")
 
        self._check_context_alive()
 
        if uuids is None:
            uuids = self.context.getAllUUIDs()
        else:
            if not isinstance(uuids, (list, tuple)):
                raise TypeError(
                    f"uuids must be a list of primitive UUIDs, "
                    f"got {type(uuids).__name__}"
                )
            uuids = list(uuids)
        if not uuids:
            raise ValueError(
                "No primitives to query: the Context contains no geometry."
            )
 
        results: Dict[str, "np.ndarray"] = {}
        for label in labels:
            if not radiation_wrapper.doesBandExist(self.radiation_model, label):
                raise RadiationModelError(
                    f"Radiation band '{label}' does not exist. "
                    f"Add it with radiation.addRadiationBand('{label}') before "
                    f"querying its absorbed flux."
                )
 
            data_label = f"radiation_flux_{label}"
            # runBand() writes this for every primitive in the Context, so a single
            # probe distinguishes "band never run" from a per-primitive gap.
            if not self.context.doesPrimitiveDataExist(uuids[0], data_label):
                raise RadiationModelError(
                    f"No absorbed flux results for band '{label}': primitive data "
                    f"'{data_label}' has not been written. Call "
                    f"radiation.runBand('{label}') (or include it in a multi-band "
                    f"runBand([...]) call) before querying its absorbed flux."
                )
 
            # One native call for the whole band. Goes straight to the bulk wrapper
            # rather than Context.getPrimitiveDataArray(), which would re-validate
            # every UUID and re-check existence that the probe above already
            # established.
            results[label] = context_wrapper.getPrimitiveDataFloatArray(
                self.context.getNativePtr(), uuids, data_label)
 
        logger.debug(
            f"Retrieved absorbed flux for bands {labels} over {len(uuids)} primitives"
        )
        return results[band_label] if single_band else results
 
    # Band query methods
    @require_plugin('radiation', 'check band existence')
    def doesBandExist(self, label: str) -> bool:
        """
        Check if a radiation band exists.
 
        Args:
            label: Name/label of the radiation band to check
 
        Returns:
            True if band exists, False otherwise
 
        Example:
            >>> radiation.addRadiationBand("SW")
            >>> radiation.doesBandExist("SW")
            True
            >>> radiation.doesBandExist("nonexistent")
            False
        """
        validate_band_label(label, "label", "doesBandExist")
        self._check_context_alive()
        return radiation_wrapper.doesBandExist(self.radiation_model, label)
 
    # Advanced source management methods
    @require_plugin('radiation', 'manage radiation sources')
    def deleteRadiationSource(self, source_id: int):
        """
        Delete a radiation source.
 
        Args:
            source_id: ID of the radiation source to delete
 
        Example:
            >>> source_id = radiation.addCollimatedRadiationSource()
            >>> radiation.deleteRadiationSource(source_id)
        """
        if not isinstance(source_id, int) or source_id < 0:
            raise ValueError(f"Source ID must be a non-negative integer, got {source_id}")
        self._check_context_alive()
        radiation_wrapper.deleteRadiationSource(self.radiation_model, source_id)
        logger.debug(f"Deleted radiation source {source_id}")
 
    @require_plugin('radiation', 'query radiation sources')
    def getSourcePosition(self, source_id: int):
        """
        Get position of a radiation source.
 
        Args:
            source_id: ID of the radiation source
 
        Returns:
            vec3 position of the source
 
        Example:
            >>> source_id = radiation.addCollimatedRadiationSource()
            >>> position = radiation.getSourcePosition(source_id)
            >>> print(f"Source at: {position}")
        """
        if not isinstance(source_id, int) or source_id < 0:
            raise ValueError(f"Source ID must be a non-negative integer, got {source_id}")
        self._check_context_alive()
        position_list = radiation_wrapper.getSourcePosition(self.radiation_model, source_id)
        from .wrappers.DataTypes import vec3
        return vec3(position_list[0], position_list[1], position_list[2])
 
    # Advanced simulation methods
    @require_plugin('radiation', 'get sky energy')
    def getSkyEnergy(self) -> float:
        """
        Get total sky energy.
 
        Returns:
            Total sky energy value
 
        Example:
            >>> energy = radiation.getSkyEnergy()
            >>> print(f"Sky energy: {energy}")
        """
        self._check_context_alive()
        return radiation_wrapper.getSkyEnergy(self.radiation_model)
 
    @require_plugin('radiation', 'calculate G-function')
    def calculateGtheta(self, view_direction) -> float:
        """
        Calculate G-function (geometry factor) for given view direction.
 
        The G-function describes the geometric relationship between leaf area
        distribution and viewing direction, important for canopy radiation modeling.
 
        The G-function is computed from the geometry currently loaded in the radiation
        model. If no geometry has been built yet, this method builds it automatically
        (with a warning) so the query operates on the current context geometry. If the
        result is still undefined (no primitives / zero leaf area), a RuntimeError is
        raised rather than silently returning NaN.
 
        A subset built by ``updateGeometry(uuids)`` is never widened by this method: the
        G-function is reported over the subset the caller selected.
 
        Args:
            view_direction: View direction as vec3 or list/tuple [x, y, z]
 
        Returns:
            G-function value
 
        Raises:
            RuntimeError: If the context has no geometry (or zero total leaf area),
                so the G-function is undefined.
 
        Example:
            >>> from pyhelios.types import vec3
            >>> radiation.updateGeometry()
            >>> g_value = radiation.calculateGtheta(vec3(0, 0, 1))
            >>> print(f"G-function: {g_value}")
        """
        validate_position_like(view_direction, "view_direction", "calculateGtheta")
 
        if not self._geometry_updated:
            # Never auto-build over a user-selected subset: updateGeometry() with no
            # argument clears the native subset latch, which would silently widen the
            # model to the full Context behind the caller's back.
            if self._geometry_subset:
                raise RuntimeError(
                    "calculateGtheta requires built geometry, but the radiation model "
                    "holds a user-specified primitive subset from updateGeometry(uuids) "
                    "and building automatically would discard it. Call "
                    "updateGeometry(uuids) again for the subset you want, or "
                    "updateGeometry() to use the full Context."
                )
            logger.warning(
                "calculateGtheta called before geometry was built; updating radiation "
                "model geometry automatically. Call updateGeometry() explicitly after "
                "building the scene to avoid this."
            )
            self.updateGeometry()
 
        context_ptr = self.context.getNativePtr()
        self._check_context_alive()
        value = radiation_wrapper.calculateGtheta(self.radiation_model, context_ptr, view_direction)
 
        if value is None or math.isnan(value):
            raise RuntimeError(
                "calculateGtheta returned an undefined (NaN) G-function. The radiation "
                "model has no geometry with positive leaf area for this context. Add "
                "primitives to the Context and ensure updateGeometry() succeeds before "
                "calling calculateGtheta()."
            )
        return value
 
    @require_plugin('radiation', 'configure output data')
    def optionalOutputPrimitiveData(self, label: str):
        """
        Enable optional primitive data output.
 
        Args:
            label: Name/label of the primitive data to output
 
        Example:
            >>> radiation.optionalOutputPrimitiveData("temperature")
        """
        validate_band_label(label, "label", "optionalOutputPrimitiveData")
        self._check_context_alive()
        radiation_wrapper.optionalOutputPrimitiveData(self.radiation_model, label)
        logger.debug(f"Enabled optional output for primitive data: {label}")
 
    @require_plugin('radiation', 'configure boundary conditions')
    def enforcePeriodicBoundary(self, boundary: str):
        """
        Enforce periodic boundary conditions.
 
        Periodic boundaries are useful for large-scale simulations to reduce
        edge effects by wrapping radiation at domain boundaries.
 
        Args:
            boundary: Boundary specification string (e.g., "xy", "xyz", "x", "y", "z")
 
        Example:
            >>> radiation.enforcePeriodicBoundary("xy")
        """
        if not isinstance(boundary, str) or not boundary:
            raise ValueError("Boundary specification must be a non-empty string")
        self._check_context_alive()
        radiation_wrapper.enforcePeriodicBoundary(self.radiation_model, boundary)
        logger.debug(f"Enforced periodic boundary: {boundary}")
 
    # Configuration methods
    @require_plugin('radiation', 'configure radiation simulation')
    @validate_scattering_depth_params
    def setScatteringDepth(self, label: str, depth: int):
        """Set scattering depth for radiation band."""
        self._check_context_alive()
        radiation_wrapper.setScatteringDepth(self.radiation_model, label, depth)
    
    @require_plugin('radiation', 'configure radiation simulation')
    @validate_min_scatter_energy_params
    def setMinScatterEnergy(self, label: str, energy: float):
        """Set minimum scatter energy for radiation band."""
        self._check_context_alive()
        radiation_wrapper.setMinScatterEnergy(self.radiation_model, label, energy)
    
    @require_plugin('radiation', 'configure radiation emission')
    def disableEmission(self, label: str):
        """Disable emission for radiation band."""
        validate_band_label(label, "label", "disableEmission")
        self._check_context_alive()
        radiation_wrapper.disableEmission(self.radiation_model, label)
    
    @require_plugin('radiation', 'configure radiation emission')
    def enableEmission(self, label: str):
        """Enable emission for radiation band."""
        validate_band_label(label, "label", "enableEmission")
        self._check_context_alive()
        radiation_wrapper.enableEmission(self.radiation_model, label)
    
    #=============================================================================
    # Camera and Image Functions (v1.3.47)
    #=============================================================================
 
    @require_plugin('radiation', 'add radiation camera')
    def addRadiationCamera(self, camera_label: str, band_labels: List[str], position, lookat_or_direction,
                          camera_properties=None, antialiasing_samples: int = 100):
        """
        Add a radiation camera to the simulation.
 
        Args:
            camera_label: Unique label string for the camera
            band_labels: List of radiation band labels for the camera
            position: Camera position as vec3 object
            lookat_or_direction: Either:
                - Lookat point as vec3 object
                - SphericalCoord for viewing direction
            camera_properties: CameraProperties instance or None for defaults
            antialiasing_samples: Number of antialiasing samples (default: 100)
 
        Raises:
            ValidationError: If parameters are invalid or have wrong types
            RadiationModelError: If camera creation fails
 
        Example:
            >>> from pyhelios import vec3, CameraProperties
            >>> # Create camera looking at origin from above
            >>> camera_props = CameraProperties(camera_resolution=(1024, 1024))
            >>> radiation_model.addRadiationCamera("main_camera", ["red", "green", "blue"],
            ...                                   position=vec3(0, 0, 5), lookat_or_direction=vec3(0, 0, 0),
            ...                                   camera_properties=camera_props)
        """
        # Import here to avoid circular imports
        from .wrappers import URadiationModelWrapper as radiation_wrapper
        from .wrappers.DataTypes import SphericalCoord, vec3, make_vec3
        from .validation.plugins import validate_camera_label, validate_band_labels_list, validate_antialiasing_samples
 
        # Validate basic parameters
        validated_label = validate_camera_label(camera_label, "camera_label", "addRadiationCamera")
        validated_bands = validate_band_labels_list(band_labels, "band_labels", "addRadiationCamera")
        validated_samples = validate_antialiasing_samples(antialiasing_samples, "antialiasing_samples", "addRadiationCamera")
 
        # Validate position (must be vec3)
        if not isinstance(position, vec3):
            raise TypeError("position must be a vec3 object. Use vec3(x, y, z) to create one.")
        validated_position = position
 
        # Validate lookat_or_direction (must be vec3 or SphericalCoord)
        if isinstance(lookat_or_direction, SphericalCoord):
            validated_direction = lookat_or_direction
        elif isinstance(lookat_or_direction, vec3):
            validated_direction = lookat_or_direction
        else:
            raise TypeError("lookat_or_direction must be a vec3 or SphericalCoord object. Use vec3(x, y, z) or SphericalCoord to create one.")
 
        # Set up camera properties
        if camera_properties is None:
            camera_properties = CameraProperties()
 
        # Call appropriate wrapper function based on direction type
        self._check_context_alive()
        try:
            if hasattr(validated_direction, 'radius') and hasattr(validated_direction, 'elevation'):
                # SphericalCoord case
                direction_coords = validated_direction.to_list()
                # SphericalCoord.to_list() returns [radius, elevation, zenith, azimuth]
                # (4 elements). The native API expects azimuth (index 3), not zenith.
                if len(direction_coords) < 4:
                    raise ValueError("SphericalCoord must expose radius, elevation, zenith, and azimuth")
                radius, elevation, azimuth = direction_coords[0], direction_coords[1], direction_coords[3]
 
                radiation_wrapper.addRadiationCameraSpherical(
                    self.radiation_model,
                    validated_label,
                    validated_bands,
                    validated_position.x, validated_position.y, validated_position.z,
                    radius, elevation, azimuth,
                    camera_properties.to_array(),
                    validated_samples,
                    camera_properties.exposure,
                    camera_properties_obj=camera_properties
                )
            else:
                # vec3 case
                radiation_wrapper.addRadiationCameraVec3(
                    self.radiation_model,
                    validated_label,
                    validated_bands,
                    validated_position.x, validated_position.y, validated_position.z,
                    validated_direction.x, validated_direction.y, validated_direction.z,
                    camera_properties.to_array(),
                    validated_samples,
                    camera_properties.exposure,
                    camera_properties_obj=camera_properties
                )
 
        except Exception as e:
            raise RadiationModelError(f"Failed to add radiation camera '{validated_label}': {e}")
 
    @require_plugin('radiation', 'add SIF camera')
    def addSIFCamera(self, camera_label: str, emission_band_labels: List[str], position,
                     lookat_or_direction, camera_properties=None, antialiasing_samples: int = 100):
        """
        Add a solar-induced chlorophyll fluorescence (SIF) camera.
 
        Each band in ``emission_band_labels`` must already exist (added via
        :meth:`addRadiationBand`); those bands are flagged internally as SIF-emitting and
        use the Fluspect-B leaf-fluorescence kernel for emission instead of Stefan-Boltzmann.
        Helios auto-creates internal radiation bands covering 400-750 nm at the resolution
        specified by ``camera_properties.excitation_bin_width_nm``.
 
        Args:
            camera_label: Unique label for the camera.
            emission_band_labels: List of pre-existing radiation band labels to drive
                with SIF emission.
            position: Camera position as a ``vec3``.
            lookat_or_direction: Either a ``vec3`` lookat point or a ``SphericalCoord``
                viewing direction.
            camera_properties: :class:`SIFCameraProperties` instance. If ``None`` defaults
                are used (10 nm excitation bins, no excitation scattering).
            antialiasing_samples: Antialiasing samples per pixel (>= 1, default 100).
 
        Raises:
            RadiationModelError: If the underlying SIF camera cannot be added (e.g.,
                an emission band was already bound to a different excitation bin width).
            NotImplementedError: If running against helios-core older than v1.3.72.
        """
        from .wrappers import URadiationModelWrapper as radiation_wrapper
        from .wrappers.DataTypes import SphericalCoord, vec3
        from .validation.plugins import (
            validate_camera_label, validate_band_labels_list, validate_antialiasing_samples
        )
 
        validated_label = validate_camera_label(camera_label, "camera_label", "addSIFCamera")
        validated_bands = validate_band_labels_list(emission_band_labels, "emission_band_labels", "addSIFCamera")
        validated_samples = validate_antialiasing_samples(antialiasing_samples, "antialiasing_samples", "addSIFCamera")
 
        if not isinstance(position, vec3):
            raise TypeError("position must be a vec3 object. Use vec3(x, y, z) to create one.")
 
        if not isinstance(lookat_or_direction, (vec3, SphericalCoord)):
            raise TypeError("lookat_or_direction must be a vec3 or SphericalCoord object.")
 
        if camera_properties is None:
            camera_properties = SIFCameraProperties()
        elif not isinstance(camera_properties, SIFCameraProperties):
            raise TypeError(
                "camera_properties must be a SIFCameraProperties instance "
                "(use SIFCameraProperties(...) — not the plain CameraProperties)."
            )
 
        self._check_context_alive()
        try:
            if isinstance(lookat_or_direction, SphericalCoord):
                # SphericalCoord.to_list() is [radius, elevation, zenith, azimuth];
                # the C wrapper expects azimuth (index 3), not zenith.
                direction_coords = lookat_or_direction.to_list()
                if len(direction_coords) < 4:
                    raise ValueError("SphericalCoord must expose radius, elevation, zenith, and azimuth")
                radius, elevation, azimuth = direction_coords[0], direction_coords[1], direction_coords[3]
                radiation_wrapper.addSIFCameraSpherical(
                    self.radiation_model,
                    validated_label,
                    validated_bands,
                    position.x, position.y, position.z,
                    radius, elevation, azimuth,
                    camera_properties.to_array(),
                    camera_properties.excitation_bin_width_nm,
                    camera_properties.excitation_scattering_depth,
                    validated_samples,
                    camera_properties_obj=camera_properties,
                )
            else:
                radiation_wrapper.addSIFCameraVec3(
                    self.radiation_model,
                    validated_label,
                    validated_bands,
                    position.x, position.y, position.z,
                    lookat_or_direction.x, lookat_or_direction.y, lookat_or_direction.z,
                    camera_properties.to_array(),
                    camera_properties.excitation_bin_width_nm,
                    camera_properties.excitation_scattering_depth,
                    validated_samples,
                    camera_properties_obj=camera_properties,
                )
        except Exception as e:
            raise RadiationModelError(f"Failed to add SIF camera '{validated_label}': {e}")
 
    @require_plugin('radiation', 'check SIF camera registration')
    def isSIFCamera(self, camera_label: str) -> bool:
        """
        Return True if the camera was registered via :meth:`addSIFCamera` (vs. ``addRadiationCamera``).
        """
        from .wrappers import URadiationModelWrapper as radiation_wrapper
        if not isinstance(camera_label, str) or not camera_label.strip():
            raise ValueError("Camera label must be a non-empty string")
        self._check_context_alive()
        return radiation_wrapper.isSIFCamera(self.radiation_model, camera_label)
 
    @require_plugin('radiation', 'enable camera flux smoothing')
    def enableCameraFluxSmoothing(self, crease_angle_degrees: float = 30.0) -> None:
        """
        Reconstruct camera images by interpolating outgoing flux across each facet.
 
        Averages each primitive's outgoing flux onto the mesh vertices it shares with its
        neighbours and interpolates it back across the facet, so a coarsely tessellated
        curved surface reads as a curve rather than as a set of flat panels. This applies
        to Tube, Sphere, Cone, Polymesh, Tile and AdaptiveTile objects. A Box, a Disk, and
        any primitive belonging to no object are left alone, since their faces are genuinely
        flat. A polymesh must carry the face table that :meth:`Context.loadOBJ` and
        :meth:`Context.loadPLY` retain; one assembled from loose primitives by
        :meth:`Context.addPolymeshObject` has no topology and is left alone.
 
        Smoothing is off by default and should be enabled deliberately: it changes what the
        camera reports. A pixel no longer carries the outgoing flux of the primitive behind
        it, but a blend of that primitive's value with its neighbours', so pixel values no
        longer correspond exactly to the ``radiation_flux_*`` primitive data. The radiation
        solve, those primitive data values, and the pixel-label and depth images are all
        unchanged.
 
        Args:
            crease_angle_degrees: Angle between adjacent facet normals above which a shared
                edge is treated as a hard crease, so flux is not averaged across it. Applies
                to polymesh objects only; the curved object types are smooth by construction
                and are never creased. Must be between 0 and 180 degrees. Defaults to 30.
 
        Raises:
            RadiationModelError: If the crease angle is out of range or the call fails
            RuntimeError: If the native library predates helios-core v1.3.84
 
        Note:
            If geometry has already been built, this re-uploads it and rebuilds the
            acceleration structure, which is not cheap. Prefer calling it before
            :meth:`updateGeometry`.
 
        Example:
            >>> with RadiationModel(context) as radiation:
            ...     radiation.enableCameraFluxSmoothing(crease_angle_degrees=45.0)
            ...     radiation.updateGeometry()
        """
        from .wrappers import URadiationModelWrapper as radiation_wrapper
        if not isinstance(crease_angle_degrees, (int, float)) or isinstance(crease_angle_degrees, bool):
            raise ValueError(
                f"crease_angle_degrees must be a number, got {type(crease_angle_degrees).__name__}"
            )
        self._check_context_alive()
        try:
            radiation_wrapper.enableCameraFluxSmoothing(self.radiation_model, float(crease_angle_degrees))
        except RuntimeError:
            raise
        except Exception as e:
            raise RadiationModelError(f"Failed to enable camera flux smoothing: {e}")
 
    @require_plugin('radiation', 'disable camera flux smoothing')
    def disableCameraFluxSmoothing(self) -> None:
        """
        Reconstruct camera images by holding the outgoing flux constant across each facet.
 
        This is the default. Each pixel reports the outgoing flux of the primitive behind it.
 
        Raises:
            RadiationModelError: If the call fails
            RuntimeError: If the native library predates helios-core v1.3.84
 
        Note:
            If geometry has already been built, this re-uploads it and rebuilds the
            acceleration structure.
        """
        from .wrappers import URadiationModelWrapper as radiation_wrapper
        self._check_context_alive()
        try:
            radiation_wrapper.disableCameraFluxSmoothing(self.radiation_model)
        except RuntimeError:
            raise
        except Exception as e:
            raise RadiationModelError(f"Failed to disable camera flux smoothing: {e}")
 
    @require_plugin('radiation', 'query camera flux smoothing state')
    def isCameraFluxSmoothingEnabled(self) -> bool:
        """
        Return True if camera images are reconstructed by interpolating flux across each facet.
 
        Raises:
            RuntimeError: If the native library predates helios-core v1.3.84
        """
        from .wrappers import URadiationModelWrapper as radiation_wrapper
        self._check_context_alive()
        return radiation_wrapper.isCameraFluxSmoothingEnabled(self.radiation_model)
 
    @require_plugin('radiation', 'query camera flux smoothing crease angle')
    def getCameraFluxSmoothingCreaseAngle(self) -> float:
        """
        Return the crease angle in degrees used by camera flux smoothing.
 
        This is the value last passed to :meth:`enableCameraFluxSmoothing`, or 30.0 if it
        has never been called.
 
        Raises:
            RuntimeError: If the native library predates helios-core v1.3.84
        """
        from .wrappers import URadiationModelWrapper as radiation_wrapper
        self._check_context_alive()
        return radiation_wrapper.getCameraFluxSmoothingCreaseAngle(self.radiation_model)
 
    @require_plugin('radiation', 'manage camera position')
    def setCameraPosition(self, camera_label: str, position):
        """
        Set camera position.
 
        Allows dynamic camera repositioning during simulation, useful for
        time-series captures or multi-view imaging.
 
        Args:
            camera_label: Camera label string
            position: Camera position as vec3 or list [x, y, z]
 
        Example:
            >>> radiation.setCameraPosition("cam1", [0, 0, 10])
            >>> from pyhelios.types import vec3
            >>> radiation.setCameraPosition("cam1", vec3(5, 5, 10))
        """
        if not isinstance(camera_label, str) or not camera_label.strip():
            raise ValueError("Camera label must be a non-empty string")
        validate_position_like(position, "position", "setCameraPosition")
        self._check_context_alive()
        radiation_wrapper.setCameraPosition(self.radiation_model, camera_label, position)
        logger.debug(f"Updated camera '{camera_label}' position")
 
    @require_plugin('radiation', 'query camera position')
    def getCameraPosition(self, camera_label: str):
        """
        Get camera position.
 
        Args:
            camera_label: Camera label string
 
        Returns:
            vec3 position of the camera
 
        Example:
            >>> position = radiation.getCameraPosition("cam1")
            >>> print(f"Camera at: {position}")
        """
        if not isinstance(camera_label, str) or not camera_label.strip():
            raise ValueError("Camera label must be a non-empty string")
        self._check_context_alive()
        position_list = radiation_wrapper.getCameraPosition(self.radiation_model, camera_label)
        from .wrappers.DataTypes import vec3
        return vec3(position_list[0], position_list[1], position_list[2])
 
    @require_plugin('radiation', 'manage camera lookat')
    def setCameraLookat(self, camera_label: str, lookat):
        """
        Set camera lookat point.
 
        Args:
            camera_label: Camera label string
            lookat: Lookat point as vec3 or list [x, y, z]
 
        Example:
            >>> radiation.setCameraLookat("cam1", [0, 0, 0])
        """
        if not isinstance(camera_label, str) or not camera_label.strip():
            raise ValueError("Camera label must be a non-empty string")
        validate_position_like(lookat, "lookat", "setCameraLookat")
        self._check_context_alive()
        radiation_wrapper.setCameraLookat(self.radiation_model, camera_label, lookat)
        logger.debug(f"Updated camera '{camera_label}' lookat point")
 
    @require_plugin('radiation', 'query camera lookat')
    def getCameraLookat(self, camera_label: str):
        """
        Get camera lookat point.
 
        Args:
            camera_label: Camera label string
 
        Returns:
            vec3 lookat point
 
        Example:
            >>> lookat = radiation.getCameraLookat("cam1")
            >>> print(f"Camera looking at: {lookat}")
        """
        if not isinstance(camera_label, str) or not camera_label.strip():
            raise ValueError("Camera label must be a non-empty string")
        self._check_context_alive()
        lookat_list = radiation_wrapper.getCameraLookat(self.radiation_model, camera_label)
        from .wrappers.DataTypes import vec3
        return vec3(lookat_list[0], lookat_list[1], lookat_list[2])
 
    @require_plugin('radiation', 'manage camera orientation')
    def setCameraOrientation(self, camera_label: str, direction):
        """
        Set camera orientation.
 
        Args:
            camera_label: Camera label string
            direction: View direction as vec3, SphericalCoord, or list [x, y, z]
 
        Example:
            >>> radiation.setCameraOrientation("cam1", [0, 0, 1])
            >>> from pyhelios.types import SphericalCoord
            >>> radiation.setCameraOrientation("cam1", SphericalCoord(1.0, 45.0, 90.0))
        """
        if not isinstance(camera_label, str) or not camera_label.strip():
            raise ValueError("Camera label must be a non-empty string")
        validate_direction_like(direction, "direction", "setCameraOrientation")
        self._check_context_alive()
        radiation_wrapper.setCameraOrientation(self.radiation_model, camera_label, direction)
        logger.debug(f"Updated camera '{camera_label}' orientation")
 
    @require_plugin('radiation', 'query camera orientation')
    def getCameraOrientation(self, camera_label: str):
        """
        Get camera orientation.
 
        Args:
            camera_label: Camera label string
 
        Returns:
            SphericalCoord orientation [radius, elevation, azimuth]
 
        Example:
            >>> orientation = radiation.getCameraOrientation("cam1")
            >>> print(f"Camera orientation: {orientation}")
        """
        if not isinstance(camera_label, str) or not camera_label.strip():
            raise ValueError("Camera label must be a non-empty string")
        self._check_context_alive()
        orientation_list = radiation_wrapper.getCameraOrientation(self.radiation_model, camera_label)
        from .wrappers.DataTypes import SphericalCoord
        return SphericalCoord(orientation_list[0], orientation_list[1], orientation_list[2])
 
    @require_plugin('radiation', 'query cameras')
    def getAllCameraLabels(self) -> List[str]:
        """
        Get all camera labels.
 
        Returns:
            List of all camera label strings
 
        Example:
            >>> cameras = radiation.getAllCameraLabels()
            >>> print(f"Available cameras: {cameras}")
        """
        self._check_context_alive()
        return radiation_wrapper.getAllCameraLabels(self.radiation_model)
 
    @require_plugin('radiation', 'configure camera spectral response')
    def setCameraSpectralResponse(self, camera_label: str, band_label: str, global_data: str):
        """
        Set camera spectral response from global data.
 
        Args:
            camera_label: Camera label
            band_label: Band label
            global_data: Global data label for spectral response curve
 
        Example:
            >>> radiation.setCameraSpectralResponse("cam1", "red", "sensor_red_response")
        """
        if not isinstance(camera_label, str) or not camera_label.strip():
            raise ValueError("Camera label must be a non-empty string")
        validate_band_label(band_label, "band_label", "setCameraSpectralResponse")
        if not isinstance(global_data, str) or not global_data.strip():
            raise ValueError("Global data label must be a non-empty string")
 
        self._check_context_alive()
        radiation_wrapper.setCameraSpectralResponse(self.radiation_model, camera_label, band_label, global_data)
        logger.debug(f"Set spectral response for camera '{camera_label}', band '{band_label}'")
        self._warn_if_bounded_bands_with_camera_response(
            camera_label, [band_label], "setCameraSpectralResponse")
 
    @require_plugin('radiation', 'configure camera from library')
    def setCameraSpectralResponseFromLibrary(self, camera_label: str, camera_library_name: str):
        """
        Set camera spectral response from standard camera library.
 
        Uses pre-defined spectral response curves for common cameras.
 
        Args:
            camera_label: Camera label
            camera_library_name: Standard camera name (e.g., "iPhone13", "NikonD850", "CanonEOS5D")
 
        Example:
            >>> radiation.setCameraSpectralResponseFromLibrary("cam1", "iPhone13")
        """
        if not isinstance(camera_label, str) or not camera_label.strip():
            raise ValueError("Camera label must be a non-empty string")
        if not isinstance(camera_library_name, str) or not camera_library_name.strip():
            raise ValueError("Camera library name must be a non-empty string")
 
        self._check_context_alive()
        radiation_wrapper.setCameraSpectralResponseFromLibrary(self.radiation_model, camera_label, camera_library_name)
        logger.debug(f"Set camera '{camera_label}' response from library: {camera_library_name}")
 
    @require_plugin('radiation', 'get camera pixel data')
    def getCameraPixelData(self, camera_label: str, band_label: str) -> List[float]:
        """
        Get camera pixel data for specific band.
 
        Retrieves raw pixel values for programmatic access and analysis.
 
        Args:
            camera_label: Camera label
            band_label: Band label
 
        Returns:
            List of pixel values
 
        Example:
            >>> pixels = radiation.getCameraPixelData("cam1", "red")
            >>> print(f"Mean pixel value: {sum(pixels)/len(pixels)}")
        """
        if not isinstance(camera_label, str) or not camera_label.strip():
            raise ValueError("Camera label must be a non-empty string")
        validate_band_label(band_label, "band_label", "getCameraPixelData")
 
        self._check_context_alive()
        return radiation_wrapper.getCameraPixelData(self.radiation_model, camera_label, band_label)
 
    @require_plugin('radiation', 'set camera pixel data')
    def setCameraPixelData(self, camera_label: str, band_label: str, pixel_data: List[float]):
        """
        Set camera pixel data for specific band.
 
        Allows programmatic modification of pixel values.
 
        Args:
            camera_label: Camera label
            band_label: Band label
            pixel_data: List of pixel values
 
        Example:
            >>> pixels = radiation.getCameraPixelData("cam1", "red")
            >>> modified_pixels = [p * 1.2 for p in pixels]  # Brighten by 20%
            >>> radiation.setCameraPixelData("cam1", "red", modified_pixels)
        """
        if not isinstance(camera_label, str) or not camera_label.strip():
            raise ValueError("Camera label must be a non-empty string")
        validate_band_label(band_label, "band_label", "setCameraPixelData")
        if not isinstance(pixel_data, (list, tuple)):
            raise ValueError("Pixel data must be a list or tuple")
 
        self._check_context_alive()
        radiation_wrapper.setCameraPixelData(self.radiation_model, camera_label, band_label, pixel_data)
        logger.debug(f"Set pixel data for camera '{camera_label}', band '{band_label}': {len(pixel_data)} pixels")
 
    # =========================================================================
    # Camera Library Functions (v1.3.58+)
    # =========================================================================
 
    @require_plugin('radiation', 'add camera from library')
    def addRadiationCameraFromLibrary(self, camera_label: str, library_camera_label: str,
                                       position, lookat, antialiasing_samples: int = 1,
                                       band_labels: Optional[List[str]] = None):
        """
        Add radiation camera loading all properties from camera library.
 
        Loads camera intrinsic parameters (resolution, FOV, sensor size) and spectral
        response data from the camera library XML file. This is the recommended way to
        create realistic cameras with proper spectral responses.
 
        Args:
            camera_label: Label for the camera instance
            library_camera_label: Label of camera in library (e.g., "Canon_20D", "iPhone11", "NikonD700")
            position: Camera position as vec3 or (x, y, z) tuple
            lookat: Lookat point as vec3 or (x, y, z) tuple
            antialiasing_samples: Number of ray samples per pixel. Default: 1
            band_labels: Optional custom band labels. If None, uses library defaults.
 
        Raises:
            RadiationModelError: If operation fails
            ValueError: If parameters are invalid
 
        Note:
            Available cameras in plugins/radiation/camera_library/camera_library.xml include:
            - Canon_20D, Nikon_D700, Nikon_D50
            - iPhone11, iPhone12ProMAX
            - Additional cameras available in library
 
        Example:
            >>> radiation.addRadiationCameraFromLibrary(
            ...     camera_label="cam1",
            ...     library_camera_label="iPhone11",
            ...     position=(0, -5, 1),
            ...     lookat=(0, 0, 0.5),
            ...     antialiasing_samples=10
            ... )
        """
        validate_band_label(camera_label, "camera_label", "addRadiationCameraFromLibrary")
        validate_position_like(position, "position", "addRadiationCameraFromLibrary")
        validate_position_like(lookat, "lookat", "addRadiationCameraFromLibrary")
 
        self._check_context_alive()
        try:
            radiation_wrapper.addRadiationCameraFromLibrary(
                self.radiation_model, camera_label, library_camera_label,
                position, lookat, antialiasing_samples, band_labels
            )
            logger.info(f"Added camera '{camera_label}' from library '{library_camera_label}'")
            # A library camera always brings measured spectral response curves.
            if band_labels:
                self._warn_if_bounded_bands_with_camera_response(
                    camera_label, band_labels, "addRadiationCameraFromLibrary")
        except Exception as e:
            raise RadiationModelError(f"Failed to add camera from library: {e}")
 
    @require_plugin('radiation', 'update camera parameters')
    def updateCameraParameters(self, camera_label: str, camera_properties: CameraProperties):
        """
        Update camera parameters for an existing camera.
 
        Allows modification of camera properties after creation while preserving
        position, lookat direction, and spectral band configuration.
 
        Args:
            camera_label: Label for the camera to update
            camera_properties: CameraProperties instance with new parameters
 
        Raises:
            RadiationModelError: If operation fails or camera doesn't exist
            ValueError: If parameters are invalid
 
        Note:
            FOV_aspect_ratio is automatically recalculated from camera_resolution.
            Camera position and lookat are preserved.
 
        Warning:
            **Changing camera_resolution discards the camera's existing image data.**
            The per-pixel buffers are sized to the resolution the camera was rendered
            at, so helios-core v1.3.79+ clears them rather than reinterpret them at the
            new size. The camera must be re-rendered with :meth:`runBand` before its
            image can be written again; :meth:`writeCameraImage` will otherwise report
            that the camera has no rendered pixel data. Changing any other parameter
            leaves existing image data intact.
 
        Example:
            >>> props = CameraProperties(
            ...     camera_resolution=(1920, 1080),
            ...     HFOV=35.0,
            ...     lens_focal_length=0.085  # 85mm lens
            ... )
            >>> radiation.updateCameraParameters("cam1", props)
            >>> # Resolution changed, so re-render before writing an image.
            >>> radiation.runBand(["red", "green", "blue"])
            >>> radiation.writeCameraImage("cam1", ["red", "green", "blue"], "out")
        """
        validate_band_label(camera_label, "camera_label", "updateCameraParameters")
 
        if not isinstance(camera_properties, CameraProperties):
            raise ValueError("camera_properties must be a CameraProperties instance")
 
        self._check_context_alive()
        try:
            radiation_wrapper.updateCameraParameters(self.radiation_model, camera_label, camera_properties)
            logger.debug(f"Updated parameters for camera '{camera_label}'")
        except Exception as e:
            raise RadiationModelError(f"Failed to update camera parameters: {e}")
 
    @require_plugin('radiation', 'enable camera metadata')
    def enableCameraMetadata(self, camera_labels):
        """
        Enable automatic JSON metadata file writing for camera(s).
 
        When enabled, writeCameraImage() automatically creates a JSON metadata file
        alongside the image containing comprehensive camera and scene information.
 
        Args:
            camera_labels: Single camera label (str) or list of camera labels (List[str])
 
        Raises:
            RadiationModelError: If operation fails
            ValueError: If parameters are invalid
 
        Note:
            Metadata includes:
            - Camera properties (model, lens, sensor specs)
            - Geographic location (latitude, longitude)
            - Acquisition settings (date, time, exposure, white balance)
            - Agronomic data (plant species, heights, phenology stages)
 
        Example:
            >>> # Enable for single camera
            >>> radiation.enableCameraMetadata("cam1")
            >>>
            >>> # Enable for multiple cameras
            >>> radiation.enableCameraMetadata(["cam1", "cam2", "cam3"])
        """
        self._check_context_alive()
        try:
            radiation_wrapper.enableCameraMetadata(self.radiation_model, camera_labels)
            if isinstance(camera_labels, str):
                logger.info(f"Enabled metadata for camera '{camera_labels}'")
            else:
                logger.info(f"Enabled metadata for {len(camera_labels)} cameras")
        except Exception as e:
            raise RadiationModelError(f"Failed to enable camera metadata: {e}")
 
    @require_plugin('radiation', 'write camera images')
    def writeCameraImage(self, camera: str, bands: List[str], imagefile_base: str,
                          image_path: str = "./", frame: int = -1,
                          flux_to_pixel_conversion: float = 1.0) -> str:
        """
        Write camera image to file and return output filename.
        
        Args:
            camera: Camera label
            bands: List of band labels to include in the image
            imagefile_base: Base filename for output
            image_path: Output directory path (default: current directory)
            frame: Frame number to write (-1 for all frames)
            flux_to_pixel_conversion: Conversion factor from flux to pixel values
            
        Returns:
            Output filename string
 
        Raises:
            RadiationModelError: If camera image writing fails
            TypeError: If parameters have incorrect types
 
        Note:
            helios-core v1.3.79+ reports a failed write by returning an empty
            filename rather than raising. PyHelios converts that into a
            RadiationModelError so a write that produced no file can never be
            mistaken for a success.
        """
        # Validate inputs
        if not isinstance(camera, str) or not camera.strip():
            raise TypeError("Camera label must be a non-empty string")
        if not isinstance(bands, list) or not bands:
            raise TypeError("Bands must be a non-empty list of strings")
        if not all(isinstance(band, str) and band.strip() for band in bands):
            raise TypeError("All band labels must be non-empty strings")
        if not isinstance(imagefile_base, str) or not imagefile_base.strip():
            raise TypeError("Image file base must be a non-empty string")
        if not isinstance(image_path, str):
            raise TypeError("Image path must be a string")
        if not isinstance(frame, int):
            raise TypeError("Frame must be an integer")
        if not isinstance(flux_to_pixel_conversion, (int, float)) or flux_to_pixel_conversion <= 0:
            raise TypeError("Flux to pixel conversion must be a positive number")
 
        self._check_context_alive()
        self._check_camera_has_pixel_data(camera, bands, "write camera image")
        filename = radiation_wrapper.writeCameraImage(
            self.radiation_model, camera, bands, imagefile_base,
            image_path, frame, flux_to_pixel_conversion)
 
        self._check_image_was_written(filename, camera, bands, image_path,
                                      "write camera image")
        logger.info(f"Camera image written to: {filename}")
        return filename
    
    @require_plugin('radiation', 'write normalized camera images')
    def writeNormCameraImage(self, camera: str, bands: List[str], imagefile_base: str,
                               image_path: str = "./", frame: int = -1) -> str:
        """
        Write normalized camera image to file and return output filename.
        
        Args:
            camera: Camera label
            bands: List of band labels to include in the image
            imagefile_base: Base filename for output
            image_path: Output directory path (default: current directory)
            frame: Frame number to write (-1 for all frames)
            
        Returns:
            Output filename string
 
        Raises:
            RadiationModelError: If normalized camera image writing fails
            TypeError: If parameters have incorrect types
 
        Note:
            helios-core v1.3.79+ reports a failed write by returning an empty
            filename rather than raising. PyHelios converts that into a
            RadiationModelError so a write that produced no file can never be
            mistaken for a success.
        """
        # Validate inputs
        if not isinstance(camera, str) or not camera.strip():
            raise TypeError("Camera label must be a non-empty string")
        if not isinstance(bands, list) or not bands:
            raise TypeError("Bands must be a non-empty list of strings")
        if not all(isinstance(band, str) and band.strip() for band in bands):
            raise TypeError("All band labels must be non-empty strings")
        if not isinstance(imagefile_base, str) or not imagefile_base.strip():
            raise TypeError("Image file base must be a non-empty string")
        if not isinstance(image_path, str):
            raise TypeError("Image path must be a string")
        if not isinstance(frame, int):
            raise TypeError("Frame must be an integer")
        
        self._check_context_alive()
        self._check_camera_has_pixel_data(camera, bands, "write normalized camera image")
        filename = radiation_wrapper.writeNormCameraImage(
            self.radiation_model, camera, bands, imagefile_base, image_path, frame)
 
        self._check_image_was_written(filename, camera, bands, image_path,
                                      "write normalized camera image")
        logger.info(f"Normalized camera image written to: {filename}")
        return filename
    
    @require_plugin('radiation', 'write camera image data')
    def writeCameraImageData(self, camera: str, band: str, imagefile_base: str,
                               image_path: str = "./", frame: int = -1):
        """
        Write camera image data to file (ASCII format).
        
        Args:
            camera: Camera label
            band: Band label
            imagefile_base: Base filename for output
            image_path: Output directory path (default: current directory)
            frame: Frame number to write (-1 for all frames)
            
        Raises:
            RadiationModelError: If camera image data writing fails
            TypeError: If parameters have incorrect types
        """
        # Validate inputs
        if not isinstance(camera, str) or not camera.strip():
            raise TypeError("Camera label must be a non-empty string")
        if not isinstance(band, str) or not band.strip():
            raise TypeError("Band label must be a non-empty string")
        if not isinstance(imagefile_base, str) or not imagefile_base.strip():
            raise TypeError("Image file base must be a non-empty string")
        if not isinstance(image_path, str):
            raise TypeError("Image path must be a string")
        if not isinstance(frame, int):
            raise TypeError("Frame must be an integer")
        
        self._check_context_alive()
        radiation_wrapper.writeCameraImageData(
            self.radiation_model, camera, band, imagefile_base, image_path, frame)
 
        logger.info(f"Camera image data written for camera {camera}, band {band}")
 
    @require_plugin('radiation', 'write primitive data label map')
    def writePrimitiveDataLabelMap(self, camera: str, primitive_data_label: str, imagefile_base: str,
                                   image_path: str = "./", frame: int = -1, padvalue: float = float('nan')):
        """
        Write a per-pixel primitive-data label map for a camera to a text file.
 
        For each camera pixel, writes the value of ``primitive_data_label`` on the primitive
        seen at that pixel. Pixels that hit no geometry (or a primitive lacking the data) are
        written as ``padvalue`` (NaN by default). The primitive data must be of type
        float, double, uint, or int. The radiation model must have been run
        (``updateGeometry`` + ``runBand``) so that per-pixel primitive labels exist.
 
        The output file is written row-by-row (one line per image row), so it loads directly
        into a 2D ``(height, width)`` array. See :meth:`getPrimitiveDataLabelMap` for a
        convenience that returns a NumPy array instead of a file on disk.
 
        Output filename: ``{camera}_{imagefile_base}.txt`` when ``frame < 0`` (default), or
        ``{camera}_{imagefile_base}_{frame:05d}.txt`` when ``frame >= 0``.
 
        Args:
            camera: Camera label
            primitive_data_label: Primitive data label to map (float/double/uint/int)
            imagefile_base: Base filename for output
            image_path: Output directory path (default: current directory)
            frame: Frame number to write (-1 to omit the frame suffix)
            padvalue: Value written for empty/background pixels (default: NaN)
 
        Raises:
            RadiationModelError: If the label map writing fails
            TypeError: If parameters have incorrect types
 
        Example:
            >>> radiation.writePrimitiveDataLabelMap(
            ...     camera="main_cam", primitive_data_label="leaf_id",
            ...     imagefile_base="leaf_labels", image_path="./output")
        """
        # Validate inputs
        if not isinstance(camera, str) or not camera.strip():
            raise TypeError("Camera label must be a non-empty string")
        if not isinstance(primitive_data_label, str) or not primitive_data_label.strip():
            raise TypeError("Primitive data label must be a non-empty string")
        if not isinstance(imagefile_base, str) or not imagefile_base.strip():
            raise TypeError("Image file base must be a non-empty string")
        if not isinstance(image_path, str):
            raise TypeError("Image path must be a string")
        if not isinstance(frame, int):
            raise TypeError("Frame must be an integer")
        if not isinstance(padvalue, (int, float)) or isinstance(padvalue, bool):
            raise TypeError("Pad value must be a numeric type")
 
        self._check_context_alive()
        radiation_wrapper.writePrimitiveDataLabelMap(
            self.radiation_model, camera, primitive_data_label, imagefile_base,
            image_path, frame, float(padvalue))
 
        logger.info(f"Primitive data label map written for camera {camera}, label {primitive_data_label}")
 
    @require_plugin('radiation', 'write object data label map')
    def writeObjectDataLabelMap(self, camera: str, object_data_label: str, imagefile_base: str,
                                image_path: str = "./", frame: int = -1, padvalue: float = float('nan')):
        """
        Write a per-pixel object-data label map for a camera to a text file.
 
        Identical to :meth:`writePrimitiveDataLabelMap` but maps the value of an object-data
        label (compound-object data) rather than primitive data. The object data must be of
        type float, double, uint, or int. The radiation model must have been run
        (``updateGeometry`` + ``runBand``) so that per-pixel labels exist.
 
        Output filename: ``{camera}_{imagefile_base}.txt`` when ``frame < 0`` (default), or
        ``{camera}_{imagefile_base}_{frame:05d}.txt`` when ``frame >= 0``.
 
        Args:
            camera: Camera label
            object_data_label: Object data label to map (float/double/uint/int)
            imagefile_base: Base filename for output
            image_path: Output directory path (default: current directory)
            frame: Frame number to write (-1 to omit the frame suffix)
            padvalue: Value written for empty/background pixels (default: NaN)
 
        Raises:
            RadiationModelError: If the label map writing fails
            TypeError: If parameters have incorrect types
        """
        # Validate inputs
        if not isinstance(camera, str) or not camera.strip():
            raise TypeError("Camera label must be a non-empty string")
        if not isinstance(object_data_label, str) or not object_data_label.strip():
            raise TypeError("Object data label must be a non-empty string")
        if not isinstance(imagefile_base, str) or not imagefile_base.strip():
            raise TypeError("Image file base must be a non-empty string")
        if not isinstance(image_path, str):
            raise TypeError("Image path must be a string")
        if not isinstance(frame, int):
            raise TypeError("Frame must be an integer")
        if not isinstance(padvalue, (int, float)) or isinstance(padvalue, bool):
            raise TypeError("Pad value must be a numeric type")
 
        self._check_context_alive()
        radiation_wrapper.writeObjectDataLabelMap(
            self.radiation_model, camera, object_data_label, imagefile_base,
            image_path, frame, float(padvalue))
 
        logger.info(f"Object data label map written for camera {camera}, label {object_data_label}")
 
    @require_plugin('radiation', 'read primitive data label map')
    def getPrimitiveDataLabelMap(self, camera: str, primitive_data_label: str,
                                 padvalue: float = float('nan')) -> 'np.ndarray':
        """
        Return a per-pixel primitive-data label map for a camera as a NumPy array.
 
        Convenience wrapper around :meth:`writePrimitiveDataLabelMap` for the common use case
        of per-pixel masking in Python: the label map is written to a temporary file, loaded
        with ``numpy.loadtxt``, and returned as a 2D array. The file does not persist.
 
        Args:
            camera: Camera label
            primitive_data_label: Primitive data label to map (float/double/uint/int)
            padvalue: Value used for empty/background pixels (default: NaN)
 
        Returns:
            2D NumPy array of shape ``(height, width)`` (row-major) holding the primitive-data
            value at each pixel, with ``padvalue`` (NaN by default) where no labelled geometry
            was seen.
 
        Raises:
            RadiationModelError: If the label map generation fails
            TypeError: If parameters have incorrect types
 
        Example:
            >>> labels = radiation.getPrimitiveDataLabelMap("main_cam", "leaf_id")
            >>> mask = labels == 3          # per-pixel mask for primitive label 3
            >>> background = np.isnan(labels)
        """
        with tempfile.TemporaryDirectory() as tmpdir:
            imagefile_base = "labelmap"
            self.writePrimitiveDataLabelMap(
                camera, primitive_data_label, imagefile_base,
                image_path=os.path.join(tmpdir, ""), frame=-1, padvalue=padvalue)
            # frame < 0 => filename has no frame suffix
            filepath = os.path.join(tmpdir, f"{camera}_{imagefile_base}.txt")
            labels = np.loadtxt(filepath)
 
        # np.loadtxt collapses a single-row file to 1D; keep a consistent 2D shape.
        if labels.ndim == 1:
            labels = labels.reshape(1, -1)
        return labels
 
    @require_plugin('radiation', 'read object data label map')
    def getObjectDataLabelMap(self, camera: str, object_data_label: str,
                              padvalue: float = float('nan')) -> 'np.ndarray':
        """
        Return a per-pixel object-data label map for a camera as a NumPy array.
 
        Convenience wrapper around :meth:`writeObjectDataLabelMap`; see
        :meth:`getPrimitiveDataLabelMap` for behaviour. The label map is written to a temporary
        file, loaded with ``numpy.loadtxt``, and returned as a 2D ``(height, width)`` array with
        ``padvalue`` (NaN by default) for background pixels. The file does not persist.
 
        Args:
            camera: Camera label
            object_data_label: Object data label to map (float/double/uint/int)
            padvalue: Value used for empty/background pixels (default: NaN)
 
        Returns:
            2D NumPy array of shape ``(height, width)`` (row-major).
 
        Raises:
            RadiationModelError: If the label map generation fails
            TypeError: If parameters have incorrect types
        """
        with tempfile.TemporaryDirectory() as tmpdir:
            imagefile_base = "labelmap"
            self.writeObjectDataLabelMap(
                camera, object_data_label, imagefile_base,
                image_path=os.path.join(tmpdir, ""), frame=-1, padvalue=padvalue)
            filepath = os.path.join(tmpdir, f"{camera}_{imagefile_base}.txt")
            labels = np.loadtxt(filepath)
 
        if labels.ndim == 1:
            labels = labels.reshape(1, -1)
        return labels
 
    @require_plugin('radiation', 'write image bounding boxes')
    def writeImageBoundingBoxes(self, camera_label: str,
                                  primitive_data_labels=None, object_data_labels=None,
                                  object_class_ids=None, image_file: str = "",
                                  classes_txt_file: str = "classes.txt",
                                  image_path: str = "./"):
        """
        Write image bounding boxes for object detection training.
        
        Supports both single and multiple data labels. Either provide primitive_data_labels
        or object_data_labels, not both.
        
        Args:
            camera_label: Camera label
            primitive_data_labels: Single primitive data label (str) or list of primitive data labels
            object_data_labels: Single object data label (str) or list of object data labels  
            object_class_ids: Single class ID (int) or list of class IDs (must match data labels)
            image_file: Image filename
            classes_txt_file: Classes definition file (default: "classes.txt")
            image_path: Image output path (default: current directory)
            
        Raises:
            RadiationModelError: If bounding box writing fails
            TypeError: If parameters have incorrect types
            ValueError: If both primitive and object data labels are provided, or neither
        """
        # Validate exclusive parameter usage
        if primitive_data_labels is not None and object_data_labels is not None:
            raise ValueError("Cannot specify both primitive_data_labels and object_data_labels")
        if primitive_data_labels is None and object_data_labels is None:
            raise ValueError("Must specify either primitive_data_labels or object_data_labels")
        
        # Validate common parameters
        if not isinstance(camera_label, str) or not camera_label.strip():
            raise TypeError("Camera label must be a non-empty string")
        if not isinstance(image_file, str) or not image_file.strip():
            raise TypeError("Image file must be a non-empty string")
        if not isinstance(classes_txt_file, str):
            raise TypeError("Classes txt file must be a string")
        if not isinstance(image_path, str):
            raise TypeError("Image path must be a string")
        
        # Handle primitive data labels
        if primitive_data_labels is not None:
            if isinstance(primitive_data_labels, str):
                # Single label
                if not isinstance(object_class_ids, int):
                    raise TypeError("For single primitive data label, object_class_ids must be an integer")
                self._check_context_alive()
                radiation_wrapper.writeImageBoundingBoxes(
                    self.radiation_model, camera_label, primitive_data_labels, 
                    object_class_ids, image_file, classes_txt_file, image_path)
                logger.info(f"Image bounding boxes written for primitive data: {primitive_data_labels}")
            
            elif isinstance(primitive_data_labels, list):
                # Multiple labels
                if not isinstance(object_class_ids, list):
                    raise TypeError("For multiple primitive data labels, object_class_ids must be a list")
                if len(primitive_data_labels) != len(object_class_ids):
                    raise ValueError("primitive_data_labels and object_class_ids must have the same length")
                if not all(isinstance(lbl, str) and lbl.strip() for lbl in primitive_data_labels):
                    raise TypeError("All primitive data labels must be non-empty strings")
                if not all(isinstance(cid, int) for cid in object_class_ids):
                    raise TypeError("All object class IDs must be integers")
                
                self._check_context_alive()
                radiation_wrapper.writeImageBoundingBoxesVector(
                    self.radiation_model, camera_label, primitive_data_labels, 
                    object_class_ids, image_file, classes_txt_file, image_path)
                logger.info(f"Image bounding boxes written for {len(primitive_data_labels)} primitive data labels")
            else:
                raise TypeError("primitive_data_labels must be a string or list of strings")
        
        # Handle object data labels  
        elif object_data_labels is not None:
            if isinstance(object_data_labels, str):
                # Single label
                if not isinstance(object_class_ids, int):
                    raise TypeError("For single object data label, object_class_ids must be an integer")
                self._check_context_alive()
                radiation_wrapper.writeImageBoundingBoxes_ObjectData(
                    self.radiation_model, camera_label, object_data_labels, 
                    object_class_ids, image_file, classes_txt_file, image_path)
                logger.info(f"Image bounding boxes written for object data: {object_data_labels}")
            
            elif isinstance(object_data_labels, list):
                # Multiple labels
                if not isinstance(object_class_ids, list):
                    raise TypeError("For multiple object data labels, object_class_ids must be a list")
                if len(object_data_labels) != len(object_class_ids):
                    raise ValueError("object_data_labels and object_class_ids must have the same length")
                if not all(isinstance(lbl, str) and lbl.strip() for lbl in object_data_labels):
                    raise TypeError("All object data labels must be non-empty strings")
                if not all(isinstance(cid, int) for cid in object_class_ids):
                    raise TypeError("All object class IDs must be integers")
                
                self._check_context_alive()
                radiation_wrapper.writeImageBoundingBoxes_ObjectDataVector(
                    self.radiation_model, camera_label, object_data_labels, 
                    object_class_ids, image_file, classes_txt_file, image_path)
                logger.info(f"Image bounding boxes written for {len(object_data_labels)} object data labels")
            else:
                raise TypeError("object_data_labels must be a string or list of strings")
    
    @require_plugin('radiation', 'write image segmentation masks')
    def writeImageSegmentationMasks(self, camera_label: str,
                                      primitive_data_labels=None, object_data_labels=None,
                                      object_class_ids=None, json_filename: str = "",
                                      image_file: str = "", append_file: bool = False):
        """
        Write image segmentation masks in COCO JSON format.
        
        Supports both single and multiple data labels. Either provide primitive_data_labels
        or object_data_labels, not both.
        
        Args:
            camera_label: Camera label
            primitive_data_labels: Single primitive data label (str) or list of primitive data labels
            object_data_labels: Single object data label (str) or list of object data labels
            object_class_ids: Single class ID (int) or list of class IDs (must match data labels)
            json_filename: JSON output filename
            image_file: Image filename
            append_file: Whether to append to existing JSON file
            
        Raises:
            RadiationModelError: If segmentation mask writing fails
            TypeError: If parameters have incorrect types
            ValueError: If both primitive and object data labels are provided, or neither
        """
        # Validate exclusive parameter usage
        if primitive_data_labels is not None and object_data_labels is not None:
            raise ValueError("Cannot specify both primitive_data_labels and object_data_labels")
        if primitive_data_labels is None and object_data_labels is None:
            raise ValueError("Must specify either primitive_data_labels or object_data_labels")
        
        # Validate common parameters
        if not isinstance(camera_label, str) or not camera_label.strip():
            raise TypeError("Camera label must be a non-empty string")
        if not isinstance(json_filename, str) or not json_filename.strip():
            raise TypeError("JSON filename must be a non-empty string")
        if not isinstance(image_file, str) or not image_file.strip():
            raise TypeError("Image file must be a non-empty string")
        if not isinstance(append_file, bool):
            raise TypeError("append_file must be a boolean")
        
        # Handle primitive data labels
        if primitive_data_labels is not None:
            if isinstance(primitive_data_labels, str):
                # Single label
                if not isinstance(object_class_ids, int):
                    raise TypeError("For single primitive data label, object_class_ids must be an integer")
                self._check_context_alive()
                radiation_wrapper.writeImageSegmentationMasks(
                    self.radiation_model, camera_label, primitive_data_labels, 
                    object_class_ids, json_filename, image_file, append_file)
                logger.info(f"Image segmentation masks written for primitive data: {primitive_data_labels}")
            
            elif isinstance(primitive_data_labels, list):
                # Multiple labels
                if not isinstance(object_class_ids, list):
                    raise TypeError("For multiple primitive data labels, object_class_ids must be a list")
                if len(primitive_data_labels) != len(object_class_ids):
                    raise ValueError("primitive_data_labels and object_class_ids must have the same length")
                if not all(isinstance(lbl, str) and lbl.strip() for lbl in primitive_data_labels):
                    raise TypeError("All primitive data labels must be non-empty strings")
                if not all(isinstance(cid, int) for cid in object_class_ids):
                    raise TypeError("All object class IDs must be integers")
                
                self._check_context_alive()
                radiation_wrapper.writeImageSegmentationMasksVector(
                    self.radiation_model, camera_label, primitive_data_labels, 
                    object_class_ids, json_filename, image_file, append_file)
                logger.info(f"Image segmentation masks written for {len(primitive_data_labels)} primitive data labels")
            else:
                raise TypeError("primitive_data_labels must be a string or list of strings")
        
        # Handle object data labels
        elif object_data_labels is not None:
            if isinstance(object_data_labels, str):
                # Single label
                if not isinstance(object_class_ids, int):
                    raise TypeError("For single object data label, object_class_ids must be an integer")
                self._check_context_alive()
                radiation_wrapper.writeImageSegmentationMasks_ObjectData(
                    self.radiation_model, camera_label, object_data_labels, 
                    object_class_ids, json_filename, image_file, append_file)
                logger.info(f"Image segmentation masks written for object data: {object_data_labels}")
            
            elif isinstance(object_data_labels, list):
                # Multiple labels
                if not isinstance(object_class_ids, list):
                    raise TypeError("For multiple object data labels, object_class_ids must be a list")
                if len(object_data_labels) != len(object_class_ids):
                    raise ValueError("object_data_labels and object_class_ids must have the same length")
                if not all(isinstance(lbl, str) and lbl.strip() for lbl in object_data_labels):
                    raise TypeError("All object data labels must be non-empty strings")
                if not all(isinstance(cid, int) for cid in object_class_ids):
                    raise TypeError("All object class IDs must be integers")
                
                self._check_context_alive()
                radiation_wrapper.writeImageSegmentationMasks_ObjectDataVector(
                    self.radiation_model, camera_label, object_data_labels, 
                    object_class_ids, json_filename, image_file, append_file)
                logger.info(f"Image segmentation masks written for {len(object_data_labels)} object data labels")
            else:
                raise TypeError("object_data_labels must be a string or list of strings")
    
    @require_plugin('radiation', 'auto-calibrate camera image')
    def autoCalibrateCameraImage(self, camera_label: str, red_band_label: str,
                                   green_band_label: str, blue_band_label: str,
                                   output_file_path: str, print_quality_report: bool = False,
                                   algorithm: str = "MATRIX_3X3_AUTO",
                                   ccm_export_file_path: str = "") -> str:
        """
        Auto-calibrate camera image with color correction and return output filename.
        
        Args:
            camera_label: Camera label
            red_band_label: Red band label
            green_band_label: Green band label  
            blue_band_label: Blue band label
            output_file_path: Output file path
            print_quality_report: Whether to print quality report
            algorithm: Color correction algorithm ("DIAGONAL_ONLY", "MATRIX_3X3_AUTO", "MATRIX_3X3_FORCE")
            ccm_export_file_path: Path to export color correction matrix (optional)
            
        Returns:
            Output filename string
            
        Raises:
            RadiationModelError: If auto-calibration fails
            TypeError: If parameters have incorrect types
            ValueError: If algorithm is not valid
        """
        # Validate inputs
        if not isinstance(camera_label, str) or not camera_label.strip():
            raise TypeError("Camera label must be a non-empty string")
        if not isinstance(red_band_label, str) or not red_band_label.strip():
            raise TypeError("Red band label must be a non-empty string")
        if not isinstance(green_band_label, str) or not green_band_label.strip():
            raise TypeError("Green band label must be a non-empty string")
        if not isinstance(blue_band_label, str) or not blue_band_label.strip():
            raise TypeError("Blue band label must be a non-empty string")
        if not isinstance(output_file_path, str) or not output_file_path.strip():
            raise TypeError("Output file path must be a non-empty string")
        if not isinstance(print_quality_report, bool):
            raise TypeError("print_quality_report must be a boolean")
        if not isinstance(ccm_export_file_path, str):
            raise TypeError("ccm_export_file_path must be a string")
        
        # Map algorithm string to integer (using MATRIX_3X3_AUTO = 1 as default)
        algorithm_map = {
            "DIAGONAL_ONLY": 0,
            "MATRIX_3X3_AUTO": 1,
            "MATRIX_3X3_FORCE": 2
        }
        
        if algorithm not in algorithm_map:
            raise ValueError(f"Invalid algorithm: {algorithm}. Must be one of: {list(algorithm_map.keys())}")
        
        algorithm_int = algorithm_map[algorithm]
        
        self._check_context_alive()
        filename = radiation_wrapper.autoCalibrateCameraImage(
            self.radiation_model, camera_label, red_band_label, green_band_label,
            blue_band_label, output_file_path, print_quality_report, 
            algorithm_int, ccm_export_file_path)
        
        logger.info(f"Auto-calibrated camera image written to: {filename}")
        return filename
    
    def getPluginInfo(self) -> dict:
        """Get information about the radiation plugin."""
        registry = get_plugin_registry()
        return registry.get_plugin_capabilities('radiation')
 
    # =========================================================================
    # EXR Image Export (v1.3.66+)
    # =========================================================================
 
    def writeCameraImageDataEXR(self, camera: str, band, imagefile_base: str,
                                image_path: str = "./", frame: int = -1):
        """
        Write camera pixel data to an EXR file with lossless float compression.
 
        Preserves full floating-point precision unlike JPEG/PNG exports.
 
        Args:
            camera: Camera label
            band: Band label (str) for single-band, or list of band labels for multi-band
            imagefile_base: Base filename for output
            image_path: Output directory path (default: current directory)
            frame: Frame number to append to filename (-1 to omit)
 
        Raises:
            RadiationModelError: If writing fails
            TypeError: If parameters have incorrect types
        """
        if not isinstance(camera, str) or not camera.strip():
            raise TypeError("Camera label must be a non-empty string")
        if not isinstance(imagefile_base, str) or not imagefile_base.strip():
            raise TypeError("Image file base must be a non-empty string")
        if not isinstance(image_path, str):
            raise TypeError("Image path must be a string")
        if not isinstance(frame, int):
            raise TypeError("Frame must be an integer")
 
        if isinstance(band, str):
            if not band.strip():
                raise TypeError("Band label must be a non-empty string")
        elif isinstance(band, (list, tuple)):
            if not band:
                raise ValueError("Band list cannot be empty")
            for b in band:
                if not isinstance(b, str) or not b.strip():
                    raise TypeError("Each band label must be a non-empty string")
        else:
            raise TypeError("band must be a string or list of strings")
 
        self._check_context_alive()
        radiation_wrapper.writeCameraImageDataEXR(
            self.radiation_model, camera, band, imagefile_base, image_path, frame)
 
    def writeDepthImageData(self, camera_label: str, imagefile_base: str,
                            image_path: str = "./", frame: int = -1):
        """
        Write depth image data to an ASCII text file.
 
        Args:
            camera_label: Camera label
            imagefile_base: Base filename for output
            image_path: Output directory path (default: current directory)
            frame: Frame number to append to filename (-1 to omit)
 
        Raises:
            RadiationModelError: If writing fails
            TypeError: If parameters have incorrect types
        """
        if not isinstance(camera_label, str) or not camera_label.strip():
            raise TypeError("Camera label must be a non-empty string")
        if not isinstance(imagefile_base, str) or not imagefile_base.strip():
            raise TypeError("Image file base must be a non-empty string")
        if not isinstance(image_path, str):
            raise TypeError("Image path must be a string")
        if not isinstance(frame, int):
            raise TypeError("Frame must be an integer")
 
        self._check_context_alive()
        radiation_wrapper.writeDepthImageData(
            self.radiation_model, camera_label, imagefile_base, image_path, frame)
 
    def writeDepthImageDataEXR(self, camera_label: str, imagefile_base: str,
                               image_path: str = "./", frame: int = -1):
        """
        Write depth image data to an EXR file with lossless float compression.
 
        Preserves full floating-point depth precision unlike ASCII or JPEG exports.
 
        Args:
            camera_label: Camera label
            imagefile_base: Base filename for output
            image_path: Output directory path (default: current directory)
            frame: Frame number to append to filename (-1 to omit)
 
        Raises:
            RadiationModelError: If writing fails
            TypeError: If parameters have incorrect types
        """
        if not isinstance(camera_label, str) or not camera_label.strip():
            raise TypeError("Camera label must be a non-empty string")
        if not isinstance(imagefile_base, str) or not imagefile_base.strip():
            raise TypeError("Image file base must be a non-empty string")
        if not isinstance(image_path, str):
            raise TypeError("Image path must be a string")
        if not isinstance(frame, int):
            raise TypeError("Frame must be an integer")
 
        self._check_context_alive()
        radiation_wrapper.writeDepthImageDataEXR(
            self.radiation_model, camera_label, imagefile_base, image_path, frame)
 
    def writeNormDepthImage(self, camera_label: str, imagefile_base: str, max_depth: float,
                            image_path: str = "./", frame: int = -1):
        """
        Write normalized depth image as grayscale JPEG.
 
        Depth values are normalized to the range [0, max_depth] for visualization.
 
        Args:
            camera_label: Camera label
            imagefile_base: Base filename for output
            max_depth: Maximum depth value for normalization (e.g., sky depth)
            image_path: Output directory path (default: current directory)
            frame: Frame number to append to filename (-1 to omit)
 
        Raises:
            RadiationModelError: If writing fails
            TypeError: If parameters have incorrect types
            ValueError: If max_depth is not positive
        """
        if not isinstance(camera_label, str) or not camera_label.strip():
            raise TypeError("Camera label must be a non-empty string")
        if not isinstance(imagefile_base, str) or not imagefile_base.strip():
            raise TypeError("Image file base must be a non-empty string")
        if not isinstance(max_depth, (int, float)):
            raise TypeError("max_depth must be a number")
        if max_depth <= 0:
            raise ValueError("max_depth must be positive")
        if not isinstance(image_path, str):
            raise TypeError("Image path must be a string")
        if not isinstance(frame, int):
            raise TypeError("Frame must be an integer")
 
        self._check_context_alive()
        radiation_wrapper.writeNormDepthImage(
            self.radiation_model, camera_label, imagefile_base, float(max_depth), image_path, frame)
 
    # =========================================================================
    # Backend Query (v1.3.67+)
    # =========================================================================
 
    def getBackendName(self) -> str:
        """
        Get the name of the active ray tracing backend.
 
        Returns:
            Backend name string (e.g., "OptiX 8.1", "Vulkan Compute")
        """
        self._check_context_alive()
        return radiation_wrapper.getBackendName(self.radiation_model)
 
    @staticmethod
    def probeAnyGPUBackend() -> bool:
        """
        Probe whether any compiled-in GPU backend is available on this system.
 
        Probes backends in priority order (OptiX 8 -> OptiX 6 -> Vulkan) without
        constructing a full backend. Useful for checking GPU availability before
        creating a RadiationModel.
 
        This is the PyHelios equivalent of the native
        ``RadiationModel::isGPUBackendAvailable()``, which as of helios-core v1.3.79
        is a pure delegate to the same underlying probe.
 
        Returns False when ``HELIOS_NO_GPU`` is set to anything other than ``"0"``,
        whatever hardware is actually present -- see
        :meth:`gpuBackendsDisabledByEnvironment` to distinguish an environment veto
        from genuinely absent hardware.
 
        Note:
            The probe runs at most once per process and the result is cached
            (helios-core v1.3.79+), so repeated calls are cheap and never re-enter
            the GPU driver. The flip side is that a driver which becomes usable
            after the first probe is not picked up until the process restarts, so a
            test that enables a device and re-probes in the same process still sees
            the cached answer.
 
        Returns:
            True if at least one GPU backend is available
 
        Example:
            >>> if RadiationModel.probeAnyGPUBackend():
            ...     with RadiationModel(context) as radiation:
            ...         radiation.addRadiationBand("SW")
        """
        return radiation_wrapper.probeAnyGPUBackend()
 
    @staticmethod
    def gpuBackendsDisabledByEnvironment() -> bool:
        """
        Check whether GPU backends are vetoed by the ``HELIOS_NO_GPU`` environment variable.
 
        True when ``HELIOS_NO_GPU`` is set to any value other than ``"0"``. The veto
        makes a GPU-equipped machine behave exactly like one with no compatible
        hardware: :meth:`probeAnyGPUBackend` returns False and constructing a
        RadiationModel with the automatic backend fails. Requesting a backend by name
        bypasses the veto.
 
        Use this to tell an intentional veto apart from missing hardware when
        reporting why the GPU path is unavailable.
 
        Note:
            The environment is read once and cached for the process lifetime, so
            changing ``HELIOS_NO_GPU`` after the first call has no effect.
 
        Returns:
            True if GPU backend probing is disabled by the environment
 
        Raises:
            RuntimeError: If the native library predates helios-core v1.3.79
 
        Example:
            >>> if not RadiationModel.probeAnyGPUBackend():
            ...     if RadiationModel.gpuBackendsDisabledByEnvironment():
            ...         print("GPU disabled via HELIOS_NO_GPU")
            ...     else:
            ...         print("No compatible GPU found")
        """
        return radiation_wrapper.gpuBackendsDisabledByEnvironment()