raytraverse.evaluate

BaseMetricSet

class raytraverse.evaluate.BaseMetricSet(vec, omega, lum, vm, metricset=None, scale=179.0, omega_as_view_area=True, guth=True, warnings=False, **kwargs)

Bases: object

object for calculating metrics based on a view direction, and rays consisting on direction, solid angle and luminance information

by encapsulating these calculations within a class, metrics with redundant calculations can take advantage of cached results, for example dgp does not need to recalculate illuminance when it has been directly requested. all metrics can be accessed as properties (and are calculated just in time) or the object can be called (no arguments) to return a np.array of all metrics defined in “metricset”

Parameters

• vm (raytraverse.mapper.ViewMapper) – the view direction

• vec (np.array) – (N, 3) directions of all rays in view

• omega (np.array) – (N,) solid angle of all rays in view

• lum (np.array) – (N,) luminance of all rays in view (multiplied by “scale”)

• metricset (list, optional) – keys of metrics to return, same as property names

• scale (float, optional) – scalefactor for luminance

• omega_as_view_area (bool, optional) – take sum(omega) as view area. if false corrects omega to vm.area

• warnings (bool, optional) – if False, suppresses numpy warnings (zero div, etc…) when accessed via __call__

• kwargs – additional arguments that may be required by additional properties

allmetrics = ['illum', 'avglum', 'loggcr', 'gcr', 'pwgcr', 'logpwgcr', 'density', 'avgraylum', 'pwavglum', 'maxlum']

safe2sum = {'avglum', 'density', 'illum'}

defaultmetrics = ['illum', 'avglum', 'loggcr']

available metrics (and the default return set)

classmethod check_metrics(metrics, raise_error=False)

returns list of valid metric names from argument if raise_error is True, raises an Atrribute Error

classmethod check_safe2sum(metrics)

checks if list if metrics is safe to compute for seperate sources before adding

property vec

property lum

property omega

property ctheta

cos angle between ray and view

property radians

cos angle between ray and view

property pos_idx

property pweight

property pweighted_area

property illum

illuminance

property avglum

average luminance

property maxlum

average luminance

property pwavglum

position weighted average luminance

property avgraylum

average luminance (not weighted by omega

property gcr

a unitless measure of relative contrast defined as the average of the squared luminances divided by the average luminance squared

property pwgcr

a unitless measure of relative contrast defined as the average of the squared luminances divided by the average luminance squared weighted by a position index

property logpwgcr

a unitless measure of relative contrast defined as the log of gcr

property loggcr

a unitless measure of relative contrast defined as the log of gcr

property density

MultiLumMetricSet

class raytraverse.evaluate.MultiLumMetricSet(vec, omega, lum, vm, metricset=None, scale=179.0, omega_as_view_area=True, **kwargs)

Bases: raytraverse.evaluate.basemetricset.BaseMetricSet

object for calculating metrics based on a view direction, and rays consisting on direction, solid angle and luminance information

by encapsulating these calculations within a class, metrics with redundant calculations can take advantage of cached results, for example dgp does not need to recalculate illuminance when it has been directly requested. all metrics can be accessed as properties (and are calculated just in time) or the object can be called (no arguments) to return a np.array of all metrics defined in “metricset”

Parameters

• vm (raytraverse.mapper.ViewMapper) – the view direction

• vec (np.array) – (N, 3) directions of all rays in view

• omega (np.array) – (N,) solid angle of all rays in view

• lum (np.array) – (N, M) luminance of all rays in view (multiplied by “scale”)

• metricset (list, optional) – keys of metrics to return, same as property names

• scale (float, optional) – scalefactor for luminance

• kwargs – additional arguments that may be required by additional properties

property illum

illuminance

property avglum

average luminance

property avgraylum

average luminance (not weighted by omega)

property gcr

a unitless measure of relative contrast defined as the average of the squared luminances divided by the average luminance squared

MetricSet

class raytraverse.evaluate.MetricSet(vec, omega, lum, vm, metricset=None, scale=179.0, threshold=2000.0, guth=True, tradius=30.0, omega_as_view_area=False, lowlight=False, **kwargs)

Bases: raytraverse.evaluate.basemetricset.BaseMetricSet

object for calculating metrics based on a view direction, and rays consisting on direction, solid angle and luminance information

by encapsulating these calculations within a class, metrics with redundant calculations can take advantage of cached results, for example dgp does not need to recalculate illuminance when it has been directly requested. all metrics can be accessed as properties (and are calculated just in time) or the object can be called (no arguments) to return a np.array of all metrics defined in “metricset”

Parameters

• vm (raytraverse.mapper.ViewMapper) – the view direction

• vec (np.array) – (N, 3) directions of all rays in view

• omega (np.array) – (N,) solid angle of all rays in view

• lum (np.array) – (N,) luminance of all rays in view (multiplied by “scale”)

• metricset (list, optional) – keys of metrics to return, same as property names

• scale (float, optional) – scalefactor for luminance

• threshold (float, optional) – threshold for glaresource/background similar behavior to evalglare ‘-b’ paramenter. if greater than 100 used as a fixed luminance threshold. otherwise used as a factor times the task luminance (defined by ‘tradius’)

• guth (bool, optional) – if True, use Guth for the upper field of view and iwata for the lower if False, use Kim

• tradius (float, optional) – radius in degrees for task luminance calculation

• kwargs – additional arguments that may be required by additional properties

defaultmetrics = ['illum', 'avglum', 'loggcr', 'ugp', 'dgp']

available metrics (and the default return set)

allmetrics = ['illum', 'avglum', 'loggcr', 'gcr', 'pwgcr', 'logpwgcr', 'density', 'avgraylum', 'pwavglum', 'maxlum', 'ugp', 'dgp', 'tasklum', 'backlum', 'dgp_t1', 'log_gc', 'dgp_t2', 'ugr', 'threshold', 'pwsl2', 'view_area', 'backlum_true', 'srcillum', 'srcarea', 'maxlum']

property src_mask

boolean mask for filtering source/background rays

property task_mask

property sources

vec, omega, lum of rays above threshold

property background

vec, omega, lum of rays below threshold

property source_pos_idx

property threshold

threshold for glaresource/background similar behavior to evalglare ‘-b’ paramenter

property pwsl2

position weighted source luminance squared, used by dgp, ugr, etc sum(Ls^2*omega/Ps^2)

property srcillum

source illuminance

property srcarea

total source area

property maxlum

peak luminance

property backlum

average background luminance CIE estimate (official for some metrics)

property backlum_true

average background luminance mathematical

property tasklum

average task luminance

property dgp

property dgp_t1

property log_gc

property dgp_t2

property ugr

property ugp

//dx.doi.org/10.1016/j.buildenv.2016.08.005

Type

http

FieldMetric

class raytraverse.evaluate.FieldMetric(vec, omega, lum, vm=None, scale=1.0, npts=360, close=True, sigma=0.05, omega_as_view_area=True, **kwargs)

Bases: raytraverse.evaluate.basemetricset.BaseMetricSet

calculate metrics on full spherical point clouds rather than view based metrics.

Parameters

• vec (np.array) – (N, 3) directions of all rays

• omega (np.array) – (N,) solid angle of all rays

• lum (np.array) – (N,) luminance of all rays (multiplied by “scale”)

• metricset (list, optional) – keys of metrics to return, same as property names

• scale (float, optional) – scalefactor for luminance

• npts (int, optional) – for equatorial metrics, the number of points to interpolate

• close (bool, optional) – include npts+1 duplicate to draw closed curve

• sigma (float, optional) – scale parameter of gaussian for kernel estimated metrics

• omega_as_view_area (bool, optional) – set to true when vectors either represent a whole sphere or a subset that does not match the viewmapper. if False, corrects boundary omega to properly trim to correct size.

• kwargs – additional arguments that may be required by additional properties

property tp

vectors in spherical coordinates

property phi

interpolated output phi values

property eq_xyz

interpolated output xyz vectors

property avg

overall vector (with magnitude)

property peak

overall vector (with magnitude)

property eq_lum

luminance along an interpolated equator with a bandwidth=sigma

property eq_density

ray density along an interpolated equator

property eq_illum

illuminiance along an interpolated equator

property eq_gcr

cosine weighted gcr along an interpolated equator

property eq_loggc

property eq_dgp

SamplingMetrics

class raytraverse.evaluate.SamplingMetrics(vec, omega, lum, vm, scale=1.0, peakthreshold=0.0, lmin=0, **kwargs)

Bases: raytraverse.evaluate.basemetricset.BaseMetricSet

default metricset for areasampler

defaultmetrics = ['avglum', 'loggcr', 'xpeak', 'ypeak']

available metrics (and the default return set)

allmetrics = ['avglum', 'loggcr', 'xpeak', 'ypeak']

property peakvec

average vector (with magnitude) for peak rays

property xpeak

x-component of avgvec as positive number (in range 0-2)

property ypeak

y-component of avgvec as positive number (in range 0-2)

property loggcr

log of global contrast ratio

PositionIndex

class raytraverse.evaluate.PositionIndex(guth=True)

Bases: object

calculate position index according to guth/iwata or kim

Parameters

guth (bool) – if True, use Guth for the upper field of view and iwata for the lower if False, use Kim

positions(vm, vec)

calculate position indices for a set of vectors

Parameters

• vm (raytraverse.mapper.ViewMapper) – the view/analysis point, should have 180 degree field of view

• vec (np.array) – shape (N,3) the view vectors to calculate

Returns

posidx – shape (N,) the position indices

Return type

np.arrray

positions_vec(viewvec, srcvec, up=0, 0, 1)

retina

raytraverse.evaluate.retina.hpsf(x, fwhm=0.183333)

estimate of human eye point-spread function

from: Yang, Yr., Wanek, J. & Shahidi, M. Representing the retinal line spread shape with mathematical functions. J. Zhejiang Univ. Sci. B 9, 996–1002 (2008). https://doi.org/10.1631/jzus.B0820184

raytraverse.evaluate.retina.inv_hpsf(y, fwhm=0.183333)

inverse of hpsf

raytraverse.evaluate.retina.blur_sun(omega, lmax, lmin=279.33, fwhm=0.183333)

calculate source correction to small bright source

returned value should be multiplied by omega and divides luminance

Parameters

• omega (Union[float, np.arrray]) – solid angle in steradians of source

• lmax (Union[float, np.arrray]) – maximum radiance in source (cd/m^2)/179

• lmin (Union[float, np.arrray], optional) – minimum radiance value to gather after spread (mimic peak extraction of evalglare, but note the different units (cd/m^2)/179

• fwhm (Union[float, np.arrray], optional) – full width half max of Lorentzian curve (radius in degrees) default is 11 arcmin.

Returns

correction factor – value should be multiplied by omega and divides luminance

Return type

Union[float, np.arrray]

raytraverse.evaluate.retina.rgcf_density_on_meridian(deg, mi)

retinal ganlgion cell field density along a meridian as a functional best fit.

the field density accounts for the input region of the ganglion cell to account for displaced ganglion cells. This value is estimate from cone density and the inferred density of midget ganglion cells. see Watson (2014) for important caveats.

Parameters

• deg (np.array) – eccentricity in degrees along merdian

• mi (int) – meridian index. [0, 1, 2, 3] for Temporal, Superior, Nasal, Inferior.

Returns

1d array of retinal ganglion cell density along a merdian

Return type

np.array

raytraverse.evaluate.retina.rgc_density_on_meridian(deg, mi)

retinal ganglion cell density along a merdian as a linear interpolation between non-zero measurements

As opposed to the field density this estimate the actual location of ganglion cells, which could be important to consider for intrinsically photosensitive cells. These are (partially?) responsible for pupillary response. However, even iprgc (may?) receive signals from rods/cones

Parameters

• deg (np.array) – eccentricity in degrees along merdian

• mi (int) – meridian index. [0, 1, 2, 3] for Temporal, Superior, Nasal, Inferior.

Returns

1d array of retinal ganglion cell density along a merdian

Return type

np.array

raytraverse.evaluate.retina.rgcf_density_xy(xy, func=<function rgcf_density_on_meridian>)

interpolate density between meridia, selected by quadrant

Parameters

• xy (np.array) – xy visual field coordinates on a disk in degrees (eccentricity 0-90 from fovea)

• func (callable) – density function along a meridian, takes r in degrees and an axes index: [0, 1, 2, 3] for Temporal, Superior, Nasal, Inferior.

Returns

1d array of single eye densities

Return type

np.array

raytraverse.evaluate.retina.binocular_density(xy, func=<function rgcf_density_on_meridian>)

average denisty between both eyes.

Parameters

• xy (np.array) – xy visual field coordinates on a disk (eccentricity 0-1 from fovea)

• func (callable) – density function along a meridian, takes r in degrees and an axes index: [0, 1, 2, 3] for Temporal, Superior, Nasal, Inferior. coordinates are for the visual field.

Returns

1d array of average binocular densities

Return type

np.array

raytraverse.evaluate.retina.rgcf_density(xy)

retinal ganglion cell field density

Parameters

xy (np.array) – xy visual field coordinates on a disk (eccentricity 0-1 from fovea)

Returns

1d array retinal ganglion cell field density according to model by Watson

Return type

np.array

raytraverse.evaluate.retina.rgc_density(xy)

retinal ganglion cell density (includes displaced ganglion cells)

Parameters

xy (np.array) – xy visual field coordinates on a disk (eccentricity 0-1 from fovea)

Returns

1d array retinal ganglion cell density according to measurements by Curcio

Return type

np.array