sceneDelegate.h

Go to the documentation of this file.

//
// Copyright 2016 Pixar
//
// Licensed under the Apache License, Version 2.0 (the "Apache License")
// with the following modification; you may not use this file except in
// compliance with the Apache License and the following modification to it:
// Section 6. Trademarks. is deleted and replaced with:
//
// 6. Trademarks. This License does not grant permission to use the trade
//    names, trademarks, service marks, or product names of the Licensor
//    and its affiliates, except as required to comply with Section 4(c) of
//    the License and to reproduce the content of the NOTICE file.
//
// You may obtain a copy of the Apache License at
//
//     http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the Apache License with the above modification is
// distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
// KIND, either express or implied. See the Apache License for the specific
// language governing permissions and limitations under the Apache License.
//
#ifndef PXR_IMAGING_HD_SCENE_DELEGATE_H
#define PXR_IMAGING_HD_SCENE_DELEGATE_H

#include "pxr/pxr.h"
#include "pxr/imaging/hd/api.h"
#include "pxr/imaging/hd/version.h"

#include "pxr/imaging/hd/aov.h"
#include "pxr/imaging/hd/basisCurvesTopology.h"
#include "pxr/imaging/hd/enums.h"
#include "pxr/imaging/hd/meshTopology.h"
#include "pxr/imaging/hd/renderIndex.h"
#include "pxr/imaging/hd/repr.h"
#include "pxr/imaging/hd/timeSampleArray.h"

#include "pxr/imaging/pxOsd/subdivTags.h"

#include "pxr/base/vt/array.h"
#include "pxr/base/vt/dictionary.h"
#include "pxr/base/vt/value.h"
#include "pxr/usd/sdf/assetPath.h"
#include "pxr/usd/sdf/path.h"

#include "pxr/base/gf/vec2i.h"
#include "pxr/base/tf/hash.h"

#include <memory>
#include <vector>

PXR_NAMESPACE_OPEN_SCOPE

class HdExtComputationContext;

/// A shared pointer to a vector of id's.
typedef std::shared_ptr<SdfPathVector> HdIdVectorSharedPtr;

/// Instancer context: a pair of instancer paths and instance indices.
typedef std::vector<std::pair<SdfPath, int>> HdInstancerContext;

/// \class HdSyncRequestVector
///
/// The SceneDelegate is requested to synchronize prims as the result of
/// executing a specific render pass, the following data structure is passed
/// back to the delegate to drive synchronization.
///
struct HdSyncRequestVector {
    // The Prims to synchronize in this request.
    SdfPathVector IDs;

    // The HdChangeTracker::DirtyBits that are set for each Prim.
    std::vector<HdDirtyBits> dirtyBits;
};

/// \struct HdDisplayStyle
///
/// Describes how the geometry of a prim should be displayed.
///
struct HdDisplayStyle {
    /// The prim refine level, in the range [0, 8].
    int refineLevel;
    
    /// Is the prim flat shaded.
    bool flatShadingEnabled;
    
    /// Is the prim displacement shaded.
    bool displacementEnabled;

    /// Does the prim act "transparent" to allow occluded selection to show
    /// through?
    bool occludedSelectionShowsThrough;

    /// Should the prim's points get shaded like surfaces, as opposed to 
    /// constant shaded?
    bool pointsShadingEnabled;

    /// Is this prim exempt from having its material disabled or overridden,
    /// for example, when a renderer chooses to ignore all scene materials?
    bool materialIsFinal;
    
    /// Creates a default DisplayStyle.
    /// - refineLevel is 0.
    /// - flatShading is disabled.
    /// - displacement is enabled.
    /// - occludedSelectionShowsThrough is disabled.
    /// - pointsShading is disabled.
    HdDisplayStyle()
        : refineLevel(0)
        , flatShadingEnabled(false)
        , displacementEnabled(true)
        , occludedSelectionShowsThrough(false)
        , pointsShadingEnabled(false)
        , materialIsFinal(false)
    { }
    
    /// Creates a DisplayStyle.
    /// \param refineLevel_ the refine level to display.
    ///        Valid range is [0, 8].
    /// \param flatShading enables flat shading, defaults to false.
    /// \param displacement enables displacement shading, defaults to false.
    /// \param occludedSelectionShowsThrough controls whether the prim lets
    ///        occluded selection show through it, defaults to false.
    /// \param pointsShadingEnabled controls whether the prim's points 
    ///        are shaded as surfaces or constant-shaded, defaults to false.
    /// \param materialisFinal controls whether the prim's material should be 
    ///        exempt from override or disabling, such as when a renderer 
    ///        wants to ignore all scene materials.
    HdDisplayStyle(int refineLevel_,
                   bool flatShading = false,
                   bool displacement = true,
                   bool occludedSelectionShowsThrough_ = false,
                   bool pointsShadingEnabled_ = false,
                   bool materialIsFinal_ = false)
        : refineLevel(std::max(0, refineLevel_))
        , flatShadingEnabled(flatShading)
        , displacementEnabled(displacement)
        , occludedSelectionShowsThrough(occludedSelectionShowsThrough_)
        , pointsShadingEnabled(pointsShadingEnabled_)
        , materialIsFinal(materialIsFinal_)
    {
        if (refineLevel_ < 0) {
            TF_CODING_ERROR("negative refine level is not supported");
        } else if (refineLevel_ > 8) {
            TF_CODING_ERROR("refine level > 8 is not supported");
        }
    }
    
    HdDisplayStyle(HdDisplayStyle const& rhs) = default;
    ~HdDisplayStyle() = default;
    
    bool operator==(HdDisplayStyle const& rhs) const {
        return refineLevel == rhs.refineLevel
            && flatShadingEnabled == rhs.flatShadingEnabled
            && displacementEnabled == rhs.displacementEnabled
            && occludedSelectionShowsThrough ==
                rhs.occludedSelectionShowsThrough
            && pointsShadingEnabled == rhs.pointsShadingEnabled
            && materialIsFinal == rhs.materialIsFinal;
    }
    bool operator!=(HdDisplayStyle const& rhs) const {
        return !(*this == rhs);
    }
};

/// \struct HdPrimvarDescriptor
///
/// Describes a primvar.
struct HdPrimvarDescriptor {
    /// Name of the primvar.
    TfToken name;
    /// Interpolation (data-sampling rate) of the primvar.
    HdInterpolation interpolation;
    /// Optional "role" indicating a desired interpretation --
    /// for example, to distinguish color/vector/point/normal.
    /// See HdPrimvarRoleTokens; default is HdPrimvarRoleTokens->none.
    TfToken role;
    /// Optional bool, true if primvar is indexed. This value should be checked
    /// before calling "GetIndexedPrimvar"
    bool indexed;

    HdPrimvarDescriptor()
    : interpolation(HdInterpolationConstant)
    , role(HdPrimvarRoleTokens->none)
    , indexed(false)
    {}
    HdPrimvarDescriptor(TfToken const& name_,
                        HdInterpolation interp_,
                        TfToken const& role_=HdPrimvarRoleTokens->none,
                        bool indexed_=false)
        : name(name_), interpolation(interp_), role(role_), indexed(indexed_)
    { }
    bool operator==(HdPrimvarDescriptor const& rhs) const {
        return name == rhs.name && role == rhs.role
            && interpolation == rhs.interpolation;
    }
    bool operator!=(HdPrimvarDescriptor const& rhs) const {
        return !(*this == rhs);
    }
};

typedef std::vector<HdPrimvarDescriptor> HdPrimvarDescriptorVector;

/// \struct HdModelDrawMode
///
/// Describes optional alternative imaging behavior for prims.
/// 
/// Some scene delegates, like the UsdImagingDelegate, will pre-flatten this
/// data, but other scene delegates may wish to use this to pipe the data 
/// through to a draw mode resolving scene index (see 
/// UsdImagingDrawModeSceneIndex as an example of such a scene index).
/// 
/// There is currently no plan to add emulation support for this information,
/// such as via HdLegacyPrimSceneIndex or HdSceneIndexAdapterSceneDelegate.
/// 
struct HdModelDrawMode {
    // Alternate imaging mode. Options are origin, bounds, cards, default, and 
    // inherited.
    TfToken drawMode;
    // Specifies whether to apply the alternative imaging mode or not.
    bool applyDrawMode;
    // The color in which to draw the geometry.
    GfVec3f drawModeColor;
    // The specific geometry to use in cards mode. Options are cross, box, and 
    // fromTexture. 
    TfToken cardGeometry;
    // The textures applied to the respective quads in cards mode.
    SdfAssetPath cardTextureXPos;
    SdfAssetPath cardTextureYPos;
    SdfAssetPath cardTextureZPos;
    SdfAssetPath cardTextureXNeg;
    SdfAssetPath cardTextureYNeg;
    SdfAssetPath cardTextureZNeg;

    HdModelDrawMode()
    : drawMode(HdModelDrawModeTokens->inherited)
    , applyDrawMode(false)
    , drawModeColor(GfVec3f(0.18))
    , cardGeometry(HdModelDrawModeTokens->cross)
    {}

    HdModelDrawMode(
        TfToken const& drawMode_,
        bool applyDrawMode_=false,
        GfVec3f drawModeColor_=GfVec3f(0.18),
        TfToken const& cardGeometry_=HdModelDrawModeTokens->cross,
        SdfAssetPath cardTextureXPos_=SdfAssetPath(),
        SdfAssetPath cardTextureYPos_=SdfAssetPath(),
        SdfAssetPath cardTextureZPos_=SdfAssetPath(),
        SdfAssetPath cardTextureXNeg_=SdfAssetPath(),
        SdfAssetPath cardTextureYNeg_=SdfAssetPath(),
        SdfAssetPath cardTextureZNeg_=SdfAssetPath())
        : drawMode(drawMode_), applyDrawMode(applyDrawMode_),
          drawModeColor(drawModeColor_), cardGeometry(cardGeometry_),
          cardTextureXPos(cardTextureXPos_), cardTextureYPos(cardTextureYPos_),
          cardTextureZPos(cardTextureZPos_), cardTextureXNeg(cardTextureXNeg_),
          cardTextureYNeg(cardTextureYNeg_), cardTextureZNeg(cardTextureZNeg_)
    {}

    bool operator==(HdModelDrawMode const& rhs) const {
        return drawMode == rhs.drawMode && 
               applyDrawMode == rhs.applyDrawMode &&
               drawModeColor == rhs.drawModeColor &&
               cardGeometry == rhs.cardGeometry &&
               cardTextureXPos == rhs.cardTextureXPos &&
               cardTextureYPos == rhs.cardTextureYPos &&
               cardTextureZPos == rhs.cardTextureZPos &&
               cardTextureXNeg == rhs.cardTextureXNeg &&
               cardTextureYNeg == rhs.cardTextureYNeg &&
               cardTextureZNeg == rhs.cardTextureZNeg;
    }
    bool operator!=(HdModelDrawMode const& rhs) const {
        return !(*this == rhs);
    }
};

/// \struct HdExtComputationPrimvarDescriptor
///
/// Extends HdPrimvarDescriptor to describe a primvar that takes
/// data from the output of an ExtComputation.
///
/// The structure contains the id of the source ExtComputation in the
/// render index, the name of an output from that computation from which
/// the primvar will take data along with a valueType which describes
/// the type of the expected data.
struct HdExtComputationPrimvarDescriptor : public HdPrimvarDescriptor {
    SdfPath sourceComputationId;
    TfToken sourceComputationOutputName;
    HdTupleType valueType;

    HdExtComputationPrimvarDescriptor() {}
    HdExtComputationPrimvarDescriptor(
        TfToken const& name_,
        HdInterpolation interp_,
        TfToken const & role_,
        SdfPath const & sourceComputationId_,
        TfToken const & sourceComputationOutputName_,
        HdTupleType const & valueType_)
        : HdPrimvarDescriptor(name_, interp_, role_, false)
        , sourceComputationId(sourceComputationId_)
        , sourceComputationOutputName(sourceComputationOutputName_)
        , valueType(valueType_)
    { }
    bool operator==(HdExtComputationPrimvarDescriptor const& rhs) const {
        return HdPrimvarDescriptor::operator==(rhs) &&
            sourceComputationId == rhs.sourceComputationId &&
            sourceComputationOutputName == rhs.sourceComputationOutputName &&
            valueType == rhs.valueType;
    }
    bool operator!=(HdExtComputationPrimvarDescriptor const& rhs) const {
        return !(*this == rhs);
    }
};

typedef std::vector<HdExtComputationPrimvarDescriptor>
        HdExtComputationPrimvarDescriptorVector;

/// \struct HdExtComputationInputDescriptor
///
/// Describes an input to an ExtComputation that takes data from
/// the output of another ExtComputation.
///
/// The structure contains the name of the input and the id of the
/// source ExtComputation in the render index, and which output of
/// that computation to bind the input to.
struct HdExtComputationInputDescriptor {
    TfToken name;
    SdfPath sourceComputationId;
    TfToken sourceComputationOutputName;

    HdExtComputationInputDescriptor() {}
    HdExtComputationInputDescriptor(
        TfToken const & name_,
        SdfPath const & sourceComputationId_,
        TfToken const & sourceComputationOutputName_)
    : name(name_), sourceComputationId(sourceComputationId_)
    , sourceComputationOutputName(sourceComputationOutputName_)
    { }

    bool operator==(HdExtComputationInputDescriptor const& rhs) const {
        return name == rhs.name &&
               sourceComputationId == rhs.sourceComputationId &&
               sourceComputationOutputName == rhs.sourceComputationOutputName;
    }
    bool operator!=(HdExtComputationInputDescriptor const& rhs) const {
        return !(*this == rhs);
    }
};

typedef std::vector<HdExtComputationInputDescriptor>
        HdExtComputationInputDescriptorVector;

/// \struct HdExtComputationOutputDescriptor
///
/// Describes an output of an ExtComputation.
///
/// The structure contains the name of the output along with a valueType
/// which describes the type of the computation output data.
struct HdExtComputationOutputDescriptor {
    TfToken name;
    HdTupleType valueType;

    HdExtComputationOutputDescriptor() {}
    HdExtComputationOutputDescriptor(
        TfToken const & name_,
        HdTupleType const & valueType_)
    : name(name_), valueType(valueType_)
    { }

    bool operator==(HdExtComputationOutputDescriptor const& rhs) const {
        return name == rhs.name &&
               valueType == rhs.valueType;
    }
    bool operator!=(HdExtComputationOutputDescriptor const& rhs) const {
        return !(*this == rhs);
    }
};

typedef std::vector<HdExtComputationOutputDescriptor>
        HdExtComputationOutputDescriptorVector;

/// \struct HdVolumeFieldDescriptor
///
/// Description of a single field related to a volume primitive.
///
struct HdVolumeFieldDescriptor {
    TfToken fieldName;
    TfToken fieldPrimType;
    SdfPath fieldId;

    HdVolumeFieldDescriptor() {}
    HdVolumeFieldDescriptor(
        TfToken const & fieldName_,
        TfToken const & fieldPrimType_,
        SdfPath const & fieldId_)
    : fieldName(fieldName_), fieldPrimType(fieldPrimType_), fieldId(fieldId_)
    { }
};

typedef std::vector<HdVolumeFieldDescriptor>
        HdVolumeFieldDescriptorVector;

/// \class HdSceneDelegate
///
/// Adapter class providing data exchange with the client scene graph.
///
class HdSceneDelegate {
public:
    /// Constructor used for nested delegate objects which share a RenderIndex.
    HD_API
    HdSceneDelegate(HdRenderIndex *parentIndex,
                    SdfPath const& delegateID);

    HD_API
    virtual ~HdSceneDelegate();

    /// Returns the RenderIndex owned by this delegate.
    HdRenderIndex& GetRenderIndex() { return *_index; }

    /// Returns the ID of this delegate, which is used as a prefix for all
    /// objects it creates in the RenderIndex. 
    ///
    /// The default value is SdfPath::AbsoluteRootPath().
    SdfPath const& GetDelegateID() const { return _delegateID; }

    /// Synchronizes the delegate state for the given request vector.
    HD_API
    virtual void Sync(HdSyncRequestVector* request);

    /// Opportunity for the delegate to clean itself up after
    /// performing parallel work during sync phase
    HD_API
    virtual void PostSyncCleanup();

    // -----------------------------------------------------------------------//
    /// \name Options
    // -----------------------------------------------------------------------//

    /// Returns true if the named option is enabled by the delegate.
    HD_API
    virtual bool IsEnabled(TfToken const& option) const;


    // -----------------------------------------------------------------------//
    /// \name Rprim Aspects
    // -----------------------------------------------------------------------//

    /// Gets the topological mesh data for a given prim.
    HD_API
    virtual HdMeshTopology GetMeshTopology(SdfPath const& id);

    /// Gets the topological curve data for a given prim.
    HD_API
    virtual HdBasisCurvesTopology GetBasisCurvesTopology(SdfPath const& id);

    /// Gets the subdivision surface tags (sharpness, holes, etc).
    HD_API
    virtual PxOsdSubdivTags GetSubdivTags(SdfPath const& id);


    /// Gets the axis aligned bounds of a prim.
    /// The returned bounds are in the local space of the prim
    /// (transform is yet to be applied) and should contain the
    /// bounds of any child prims.
    ///
    /// The returned bounds does not include any displacement that
    /// might occur as the result of running shaders on the prim.
    HD_API
    virtual GfRange3d GetExtent(SdfPath const & id);

    /// Returns the object space transform, including all parent transforms.
    HD_API
    virtual GfMatrix4d GetTransform(SdfPath const & id);

    /// Returns the authored visible state of the prim.
    HD_API
    virtual bool GetVisible(SdfPath const & id);

    /// Returns the doubleSided state for the given prim.
    HD_API
    virtual bool GetDoubleSided(SdfPath const & id);

    /// Returns the cullstyle for the given prim.
    HD_API
    virtual HdCullStyle GetCullStyle(SdfPath const &id);

    /// Returns the shading style for the given prim.
    HD_API
    virtual VtValue GetShadingStyle(SdfPath const &id);

    /// Returns the refinement level for the given prim in the range [0,8].
    ///
    /// The refinement level indicates how many iterations to apply when
    /// subdividing subdivision surfaces or other refinable primitives.
    HD_API
    virtual HdDisplayStyle GetDisplayStyle(SdfPath const& id);
    
    /// Returns a named value.
    HD_API
    virtual VtValue Get(SdfPath const& id, TfToken const& key);

    /// Returns a named primvar value. If \a *outIndices is not nullptr and the 
    /// primvar has indices, it will return the unflattened primvar and set 
    /// \a *outIndices to the primvar's associated indices, clearing the array
    /// if the primvar is not indexed.
    HD_API
    virtual VtValue GetIndexedPrimvar(SdfPath const& id, 
                                      TfToken const& key, 
                                      VtIntArray *outIndices);

    /// Returns the authored repr (if any) for the given prim.
    HD_API
    virtual HdReprSelector GetReprSelector(SdfPath const &id);

    /// Returns the render tag that will be used to bucket prims during
    /// render pass bucketing.
    HD_API
    virtual TfToken GetRenderTag(SdfPath const& id);

    /// Returns the prim categories.
    HD_API
    virtual VtArray<TfToken> GetCategories(SdfPath const& id);

    /// Returns the categories for all instances in the instancer.
    HD_API
    virtual std::vector<VtArray<TfToken>>
    GetInstanceCategories(SdfPath const &instancerId);

    /// Returns the coordinate system bindings, or a nullptr if none are bound.
    HD_API
    virtual HdIdVectorSharedPtr GetCoordSysBindings(SdfPath const& id);

    /// Returns the model draw mode object for the given prim.
    HD_API
    virtual HdModelDrawMode GetModelDrawMode(SdfPath const& id);

    // -----------------------------------------------------------------------//
    /// \name Motion samples
    // -----------------------------------------------------------------------//

    /// Store up to \a maxSampleCount transform samples in \a *sampleValues.
    /// Returns the union of the authored samples and the boundaries 
    /// of the current camera shutter interval. If this number is greater
    /// than maxSampleCount, you might want to call this function again 
    /// to get all the authored data.
    /// Sample times are relative to the scene delegate's current time.
    /// \see GetTransform()
    HD_API
    virtual size_t
    SampleTransform(SdfPath const & id, 
                    size_t maxSampleCount,
                    float *sampleTimes, 
                    GfMatrix4d *sampleValues);

    /// Convenience form of SampleTransform() that takes an HdTimeSampleArray.
    /// This function returns the union of the authored transform samples 
    /// and the boundaries of the current camera shutter interval.
    template <unsigned int CAPACITY>
    void 
    SampleTransform(SdfPath const & id,
                    HdTimeSampleArray<GfMatrix4d, CAPACITY> *sa) {
        size_t authoredSamples = 
            SampleTransform(id, CAPACITY, sa->times.data(), sa->values.data());
        if (authoredSamples > CAPACITY) {
            sa->Resize(authoredSamples);
            size_t authoredSamplesSecondAttempt = 
                SampleTransform(
                    id, 
                    authoredSamples, 
                    sa->times.data(), 
                    sa->values.data());
            // Number of samples should be consisntent through multiple
            // invokations of the sampling function.
            TF_VERIFY(authoredSamples == authoredSamplesSecondAttempt);
        }
        sa->count = authoredSamples;
    }

    /// Store up to \a maxSampleCount transform samples in \a *sampleValues.
    /// Returns the union of the authored samples and the boundaries 
    /// of the current camera shutter interval. If this number is greater
    /// than maxSampleCount, you might want to call this function again 
    /// to get all the authored data.
    /// Sample times are relative to the scene delegate's current time.
    /// \see GetInstancerTransform()
    HD_API
    virtual size_t
    SampleInstancerTransform(SdfPath const &instancerId,
                             size_t maxSampleCount, 
                             float *sampleTimes,
                             GfMatrix4d *sampleValues);

    /// Convenience form of SampleInstancerTransform()
    /// that takes an HdTimeSampleArray.
    /// This function returns the union of the authored samples 
    /// and the boundaries of the current camera shutter interval.
    template <unsigned int CAPACITY>
    void
    SampleInstancerTransform(SdfPath const &instancerId,
                             HdTimeSampleArray<GfMatrix4d, CAPACITY> *sa) {
        size_t authoredSamples = 
            SampleInstancerTransform(
                instancerId, 
                CAPACITY, 
                sa->times.data(), 
                sa->values.data());
        if (authoredSamples > CAPACITY) {
            sa->Resize(authoredSamples);
            size_t authoredSamplesSecondAttempt = 
                SampleInstancerTransform(
                    instancerId, 
                    authoredSamples, 
                    sa->times.data(), 
                    sa->values.data());
            // Number of samples should be consisntent through multiple
            // invokations of the sampling function.
            TF_VERIFY(authoredSamples == authoredSamplesSecondAttempt);
        }
        sa->count = authoredSamples;
    }

    /// Store up to \a maxSampleCount primvar samples in \a *samplesValues.
    /// Returns the union of the authored samples and the boundaries 
    /// of the current camera shutter interval. If this number is greater
    /// than maxSampleCount, you might want to call this function again 
    /// to get all the authored data.
    ///
    /// Sample values that are array-valued will have a size described
    /// by the HdPrimvarDescriptor as applied to the toplogy.
    ///
    /// For example, this means that a mesh that is fracturing over time
    /// will return samples with the same number of points; the number
    /// of points will change as the scene delegate is resynchronized
    /// to represent the scene at a time with different topology.
    ///
    /// Sample times are relative to the scene delegate's current time.
    ///
    /// \see Get()
    HD_API
    virtual size_t
    SamplePrimvar(SdfPath const& id, 
                  TfToken const& key,
                  size_t maxSampleCount, 
                  float *sampleTimes, 
                  VtValue *sampleValues);

    /// Convenience form of SamplePrimvar() that takes an HdTimeSampleArray.
    /// This function returns the union of the authored samples 
    /// and the boundaries of the current camera shutter interval.
    template <unsigned int CAPACITY>
    void 
    SamplePrimvar(SdfPath const &id, 
                  TfToken const& key,
                  HdTimeSampleArray<VtValue, CAPACITY> *sa);

    /// SamplePrimvar() for getting an unflattened primvar and its indices. If 
    /// \a *sampleIndices is not nullptr and the primvar has indices, it will 
    /// return unflattened primvar samples in \a *sampleValues and the primvar's 
    /// sampled indices in \a *sampleIndices, clearing the \a *sampleIndices 
    /// array if the primvar is not indexed.
    HD_API
    virtual size_t
    SampleIndexedPrimvar(SdfPath const& id, 
                         TfToken const& key,
                         size_t maxSampleCount, 
                         float *sampleTimes, 
                         VtValue *sampleValues,
                         VtIntArray *sampleIndices);

    /// Convenience form of SampleIndexedPrimvar() that takes 
    /// HdTimeSampleArrays. This function returns the union of the authored 
    /// samples and the boundaries of the current camera shutter interval.
    template <unsigned int CAPACITY>
    void 
    SampleIndexedPrimvar(SdfPath const &id, 
                         TfToken const& key,
                         HdIndexedTimeSampleArray<VtValue, CAPACITY> *sa);

    // -----------------------------------------------------------------------//
    /// \name Instancer prototypes
    // -----------------------------------------------------------------------//

    /// Gets the extracted indices array of the prototype id used in the
    /// instancer.
    ///
    /// example
    ///  instances:  0, 1, 2, 3, 4, 5
    ///  protoypes:  A, B, A, A, B, C
    ///
    ///    GetInstanceIndices(A) : [0, 2, 3]
    ///    GetInstanceIndices(B) : [1, 4]
    ///    GetInstanceIndices(C) : [5]
    ///    GetInstanceIndices(D) : []
    ///
    HD_API
    virtual VtIntArray GetInstanceIndices(SdfPath const &instancerId,
                                          SdfPath const &prototypeId);

    /// Returns the instancer transform.
    HD_API
    virtual GfMatrix4d GetInstancerTransform(SdfPath const &instancerId);

    /// Returns the parent instancer of the given rprim or instancer.
    HD_API
    virtual SdfPath GetInstancerId(SdfPath const& primId);

    /// Returns a list of prototypes of this instancer. The intent is to let
    /// renderers cache instance indices by giving them a complete set of prims
    /// to call GetInstanceIndices(instancer, prototype) on.
    /// XXX: This is currently unused, but may be used in the future.
    HD_API
    virtual SdfPathVector GetInstancerPrototypes(SdfPath const& instancerId);

    // -----------------------------------------------------------------------//
    /// \name Path Translation
    // -----------------------------------------------------------------------//

    /// Returns the scene address of the prim corresponding to the given
    /// rprim/instance index. This is designed to give paths in scene namespace,
    /// rather than hydra namespace, so it always strips the delegate ID.
    /// \deprecated use GetScenePrimPaths
    HD_API
    virtual SdfPath GetScenePrimPath(SdfPath const& rprimId,
                                     int instanceIndex,
                                     HdInstancerContext *instancerContext = nullptr);

    /// A vectorized version of GetScenePrimPath that allows the prim adapter
    /// to amortize expensive calculations across a number of path evaluations
    /// in a single call. Note that only a single rprimId is supported. This
    /// allows this call to be forwarded directly to a single prim adapter
    /// rather than requiring a lot of data shuffling.
    HD_API
    virtual SdfPathVector GetScenePrimPaths(SdfPath const& rprimId,
                                     std::vector<int> instanceIndices,
                                     std::vector<HdInstancerContext> *instancerContexts = nullptr);

    /// Returns an identifier that can be used to share data between HdPrims
    /// during a Sync operation. Any number of HdPrims of the same type may
    /// return the same identifier. In this situation, the render delegate
    /// can choose to share information between HdPrims rather than fetching
    /// data for each HdPrim. For the UsdImagingSceneDelegate, this identifier
    /// will be the path of the GetPrototype prim for the UsdPrim correspnding
    /// to this HdPrim. Returns an empty SdfPath if this feature is no
    /// supported or the UsdPrim has no prototype. Render delegates are free
    /// to ignore this method completely if they aren't interested in the
    /// additional level of data sharing enabled by this information.
    HD_API
    virtual SdfPath GetDataSharingId(SdfPath const& primId);

    // -----------------------------------------------------------------------//
    /// \name Material Aspects
    // -----------------------------------------------------------------------//
    
    /// Returns the material ID bound to the rprim \p rprimId.
    HD_API
    virtual SdfPath GetMaterialId(SdfPath const &rprimId);

    // Returns a material resource which contains the information 
    // needed to create a material.
    HD_API 
    virtual VtValue GetMaterialResource(SdfPath const &materialId);

    // -----------------------------------------------------------------------//
    /// \name Renderbuffer Aspects
    // -----------------------------------------------------------------------//

    /// Returns the allocation descriptor for a given render buffer prim.
    HD_API
    virtual HdRenderBufferDescriptor GetRenderBufferDescriptor(SdfPath const& id);

    // -----------------------------------------------------------------------//
    /// \name Light Aspects
    // -----------------------------------------------------------------------//

    // Returns a single value for a given light and parameter.
    HD_API
    virtual VtValue GetLightParamValue(SdfPath const &id, 
                                       TfToken const &paramName);

    // -----------------------------------------------------------------------//
    /// \name Camera Aspects
    // -----------------------------------------------------------------------//

    /// Returns a single value for a given camera and parameter.
    /// See HdCameraTokens for the list of paramters.
    HD_API
    virtual VtValue GetCameraParamValue(SdfPath const& cameraId,
                                        TfToken const& paramName);

    // -----------------------------------------------------------------------//
    /// \name Volume Aspects
    // -----------------------------------------------------------------------//

    HD_API
    virtual HdVolumeFieldDescriptorVector
    GetVolumeFieldDescriptors(SdfPath const &volumeId);

    // -----------------------------------------------------------------------//
    /// \name ExtComputation Aspects
    // -----------------------------------------------------------------------//

    ///
    /// For the given computation id, returns a list of inputs which
    /// will be requested from the scene delegate using the Get() method.
    ///
    /// See GetExtComputationInputDescriptors and
    /// GetExtComputationOutpuDescriptors for descriptions of other
    /// computation inputs and outputs.
    HD_API
    virtual TfTokenVector
    GetExtComputationSceneInputNames(SdfPath const& computationId);

    ///
    /// For the given computation id, returns a list of computation
    /// input descriptors.
    ///
    /// See HdExtComputationInputDecriptor
    HD_API
    virtual HdExtComputationInputDescriptorVector
    GetExtComputationInputDescriptors(SdfPath const& computationId);

    /// For the given computation id, returns a list of computation
    /// output descriptors.
    ///
    /// See HdExtComputationOutputDescriptor
    HD_API
    virtual HdExtComputationOutputDescriptorVector
    GetExtComputationOutputDescriptors(SdfPath const& computationId);


    /// Returns a list of primvar names that should be bound to
    /// a generated output from  an ExtComputation for the given prim id and
    /// interpolation mode.  Binding information is obtained through
    /// GetExtComputationPrimvarDesc()
    /// Returns a structure describing source information for a primvar
    /// that is bound to an ExtComputation.  See HdExtComputationPrimvarDesc
    /// for the expected information to be returned.
    HD_API
    virtual HdExtComputationPrimvarDescriptorVector
    GetExtComputationPrimvarDescriptors(SdfPath const& id,
                                        HdInterpolation interpolationMode);

    /// Returns a single value for a given computation id and input token.
    /// The token may be a computation input or a computation config parameter.
    HD_API
    virtual VtValue GetExtComputationInput(SdfPath const& computationId,
                                           TfToken const& input);

    /// Return up to \a maxSampleCount samples for a given computation id and
    /// input token.
    /// The token may be a computation input or a computation config parameter.
    /// Returns the union of the authored samples and the boundaries
    /// of the current camera shutter interval. If this number is greater
    /// than maxSampleCount, you might want to call this function again
    /// to get all the authored data.
    HD_API
    virtual size_t SampleExtComputationInput(SdfPath const& computationId,
                                             TfToken const& input,
                                             size_t maxSampleCount,
                                             float *sampleTimes,
                                             VtValue *sampleValues);

    /// Convenience form of SampleExtComputationInput() that takes an
    /// HdTimeSampleArray.
    /// Returns the union of the authored samples and the boundaries
    /// of the current camera shutter interval.
    template <unsigned int CAPACITY>
    void SampleExtComputationInput(SdfPath const& computationId,
                                   TfToken const& input,
                                   HdTimeSampleArray<VtValue, CAPACITY> *sa) {
        size_t authoredSamples = SampleExtComputationInput(
                computationId, input, CAPACITY,
                sa->times.data(), sa->values.data());

        if (authoredSamples > CAPACITY) {
            sa->Resize(authoredSamples);
            size_t authoredSamplesSecondAttempt = SampleExtComputationInput(
                    computationId, input, authoredSamples,
                    sa->times.data(), sa->values.data());
            // Number of samples should be consisntent through multiple
            // invokations of the sampling function.
            TF_VERIFY(authoredSamples == authoredSamplesSecondAttempt);
        }
        sa->count = authoredSamples;
    }

    /// Returns the kernel source assigned to the computation at the path id.
    /// If the string is empty the computation has no GPU kernel and the
    /// CPU callback should be used.
    HD_API
    virtual std::string GetExtComputationKernel(SdfPath const& computationId);

    /// Requests the scene delegate run the ExtComputation with the given id.
    /// The context contains the input values that delegate requested through
    /// GetExtComputationInputNames().
    ///
    /// The scene delegate is expected to set each output identified by
    /// GetExtComputationOutputNames() on the context.
    ///
    /// Hydra may invoke the computation on a different thread from
    /// what HdEngine::Execute() was called on.  It may also invoke
    /// many computations in parallel.
    HD_API
    virtual void InvokeExtComputation(SdfPath const& computationId,
                                      HdExtComputationContext *context);

    // -----------------------------------------------------------------------//
    /// \name Primitive Variables
    // -----------------------------------------------------------------------//

    /// Returns descriptors for all primvars of the given interpolation type.
    HD_API
    virtual HdPrimvarDescriptorVector
    GetPrimvarDescriptors(SdfPath const& id, HdInterpolation interpolation);

    // -----------------------------------------------------------------------//
    /// \name Task Aspects
    // -----------------------------------------------------------------------//
    HD_API
    virtual TfTokenVector GetTaskRenderTags(SdfPath const& taskId);

private:
    HdRenderIndex *_index;
    SdfPath _delegateID;

    HdSceneDelegate() = delete;
    HdSceneDelegate(HdSceneDelegate &) =  delete;
    HdSceneDelegate &operator=(HdSceneDelegate &) =  delete;
};

template <unsigned int CAPACITY>
void 
HdSceneDelegate::SamplePrimvar(SdfPath const &id, 
                               TfToken const& key,
                               HdTimeSampleArray<VtValue, CAPACITY> *sa) {
    size_t authoredSamples = 
        SamplePrimvar(
            id, 
            key, 
            CAPACITY, 
            sa->times.data(), 
            sa->values.data());
    if (authoredSamples > CAPACITY) {
        sa->Resize(authoredSamples);
        size_t authoredSamplesSecondAttempt = 
            SamplePrimvar(
                id, 
                key, 
                authoredSamples, 
                sa->times.data(), 
                sa->values.data());
        // Number of samples should be consistent through multiple
        // invocations of the sampling function.
        TF_VERIFY(authoredSamples == authoredSamplesSecondAttempt);
    }
    sa->count = authoredSamples;
}

template <unsigned int CAPACITY>
void 
HdSceneDelegate::SampleIndexedPrimvar(SdfPath const &id, 
                         TfToken const& key,
                         HdIndexedTimeSampleArray<VtValue, CAPACITY> *sa) {
    size_t authoredSamples = 
        SampleIndexedPrimvar(
            id, 
            key, 
            CAPACITY, 
            sa->times.data(), 
            sa->values.data(),
            sa->indices.data());
    if (authoredSamples > CAPACITY) {
        sa->Resize(authoredSamples);
        size_t authoredSamplesSecondAttempt = 
            SampleIndexedPrimvar(
                id, 
                key, 
                authoredSamples, 
                sa->times.data(), 
                sa->values.data(),
                sa->indices.data());
        // Number of samples should be consistent through multiple
        // invocations of the sampling function.
        TF_VERIFY(authoredSamples == authoredSamplesSecondAttempt);
    }
    sa->count = authoredSamples;
}

PXR_NAMESPACE_CLOSE_SCOPE

#endif //PXR_IMAGING_HD_SCENE_DELEGATE_H