Class Viewport

Inheritance Relationships

Base Types

  • public mrpt::serialization::CSerializable

  • public mrpt::system::CObservable

  • public mrpt::viz::CTextMessageCapable (Class CTextMessageCapable)

Class Documentation

class Viewport : public mrpt::serialization::CSerializable, public mrpt::system::CObservable, public mrpt::viz::CTextMessageCapable

A viewport within a Scene, containing a set of OpenGL objects to render. This class has protected constructor, thus it cannot be created by users. Use Scene::createViewport instead. A viewport has these “operation modes”:

  • Normal (default): It renders the contained objects.

  • Cloned: It clones the objects from another viewport. See setCloneView()

  • Image mode: It renders an image (e.g. from a video stream) efficiently using a textured quad. See setImageView().

In any case, the viewport can be resized to only fit a part of the entire parent viewport. There will be always at least one viewport in a Scene named “main”.

This class can be observed (see mrpt::system::CObserver) for the following events (see mrpt::system::mrptEvent):

  • mrpt::viz::mrptEventGLPreRender

  • mrpt::viz::mrptEventGLPostRender

One directional light source is created by default. Multiple lights (directional, point, spot) are supported via lightParameters().lights.

Lighting parameters are accessible via lightParameters().

Refer to mrpt::viz::Scene for further details.

Contained objects set/get/search

using const_iterator = ListVisualObjects::const_iterator
using iterator = ListVisualObjects::iterator
inline const_iterator begin() const
inline const_iterator end() const
inline iterator begin()
inline iterator end()
void clear()

Delete all internal objects *

See also

insert

void insert(const CVisualObject::Ptr &newObject)

Insert a new object into the list. The object MUST NOT be deleted, it will be deleted automatically by this object when not required anymore.

void setCurrentCameraFromPose(mrpt::poses::CPose3D &p)

Changes the point of view of the camera, from a given pose.

See also

getCurrentCameraPose

CVisualObject::Ptr getByName(const std::string &str)

Returns the first object with a given name, or nullptr if not found.

template<typename T>
inline T::Ptr getByClass(size_t ith = 0) const

Returns the i’th object of a given class (or of a descendant class), or nullptr (an empty smart pointer) if not found. Example: 

CSphere::Ptr obs = view.getByClass<CSphere>();
By default (ith=0), the first observation is returned.

void removeObject(const CVisualObject::Ptr &obj)

Removes the given object from the scene (it also deletes the object to free its memory).

inline size_t size() const

Number of objects contained.

This is an overloaded member function, provided for convenience. It differs from the above function only in what argument(s) it accepts.

Requires that the viewport has been rendered at least once (so that updateRenderedViewportSize() has been called by the rendering backend).

inline bool empty() const
inline mrpt::viz::CCamera &getCamera()

Get a reference to the camera associated with this viewport.

inline const mrpt::viz::CCamera &getCamera() const

Get a reference to the camera associated with this viewport.

mrpt::math::TBoundingBox getBoundingBox() const

Viewport “modes”

void setCloneView(const std::string &clonedViewport)

Set this viewport as a clone of some other viewport, given its name - as a side effect, current list of internal OpenGL objects is cleared. By default, only the objects are cloned, not the camera. See

See also

resetCloneView

void setImageView(const mrpt::img::CImage &img, bool transparentBackground = true)

Set this viewport into “image view”-mode, where an image is efficiently drawn (fitting the viewport area) using an OpenGL textured quad. Call this method with the new image to update the displayed image (but recall to first lock the parent Scene’s critical section, then do the update, then release the lock, and then issue a window repaint). Internally, the texture is drawn using a mrpt::viz::CTexturedPlane The viewport can be reverted to behave like a normal viewport by calling setNormalMode()

Parameters:

transparentBackground – [in] This method can also make the viewport transparent (default), so the area not filled with the image still allows seeing an underlying viewport.

void setImageView(mrpt::img::CImage &&img, bool transparentBackground = true)
inline bool isImageViewMode() const

Returns true if setImageView() has been called on this viewport

inline mrpt::viz::CTexturedPlane::Ptr getImageViewPlane() const

Returns the internal image view plane (or nullptr). Used by the renderer.

inline void resetCloneView()

Reset the viewport to normal mode: rendering its own objects.

See also

setCloneView, setNormalMode

void setCloneCamera(bool enable)

If set to true, and setCloneView() has been called, this viewport will be rendered using the camera of the cloned viewport.

inline void setClonedCameraFrom(const std::string &viewPortName)

Use the camera of another viewport. Note this works even for viewports not in “clone” mode, so you can render different scenes but using the same camera.

inline bool isCloned() const
inline bool isClonedCamera() const
inline std::string getClonedViewportName() const

Returns the name of the viewport whose objects are being cloned, or empty if not cloning. Set by setCloneView().

inline std::string isClonedCameraFrom() const
void setNormalMode()

Resets the viewport to a normal 3D viewport

See also

setCloneView, setImageView

inline void setViewportVisibility(bool visible)
inline bool getViewportVisibility() const

OpenGL global settings that affect rendering all objects in the

scene/viewport

inline void enablePolygonNicest(bool enable = true)

Sets glHint(GL_POLYGON_SMOOTH_HINT, GL_NICEST) is enabled, or GL_FASTEST otherwise.

inline bool isPolygonNicestEnabled() const
inline const TLightParameters &lightParameters() const
inline TLightParameters &lightParameters()

Change or read viewport properties (except “viewport modes”)

inline std::string getName() const

Returns the name of the viewport

void setViewportPosition(double x, double y, double width, double height)

Change the viewport position and dimension on the rendering window. X & Y coordinates here can have two interpretations:

  • If in the range [0,1], they are factors with respect to the actual window sizes (i.e. width=1 means the entire width of the rendering window).

  • If >1, they are interpreted as pixels.

width & height can be interpreted as:

  • If >1, they are the size of the viewport in that dimension, in pixels.

  • If in [0,1], they are the size of the viewport in that dimension, in a factor of the width/height.

  • If in [-1,0[, the size is computed such as the right/top border ends up in the given coordinate, interpreted as a factor (e.g. -1: up to the end of the viewport, -0.5: up to the middle of it).

  • If <-1 the size is computed such as the right/top border ends up in the given absolute coordinate (e.g. -200: up to the row/column 200px).

See also

getViewportPosition

Note

(x,y) specify the lower left corner of the viewport rectangle.

void getViewportPosition(double &x, double &y, double &width, double &height) const

Get the current viewport position and dimension on the rendering window. X & Y coordinates here can have two interpretations:

  • If in the range [0,1], they are factors with respect to the actual window sizes (i.e. width=1 means the entire width of the rendering window).

  • If >1, they are interpreted as pixels.

    See also

    setViewportPosition

Note

(x,y) specify the lower left corner of the viewport rectangle.

void setViewportClipDistances(float clip_min, float clip_max)

Set the min/max clip depth distances of the rendering frustum (default: 0.1 - 1000)

See also

getViewportClipDistances

void getViewportClipDistances(float &clip_min, float &clip_max) const

Get the current min/max clip depth distances of the rendering frustum (default: 0.1 - 1000)

See also

setViewportClipDistances

void setLightShadowClipDistances(float clip_min, float clip_max)
void getLightShadowClipDistances(float &clip_min, float &clip_max) const
inline void setBorderSize(unsigned int lineWidth)

Set the border size (“frame”) of the viewport (default=0)

inline unsigned int getBorderSize() const
inline void setBorderColor(const mrpt::img::TColor &c)
inline const mrpt::img::TColor &getBorderColor() const
inline bool isTransparent() const

Return whether the viewport will be rendered transparent over previous viewports.

inline void setTransparent(bool trans)

Set the transparency, that is, whether the viewport will be rendered transparent over previous viewports (default=false).

inline void setCustomBackgroundColor(const mrpt::img::TColorf &color)

Defines the viewport background color

inline mrpt::img::TColorf getCustomBackgroundColor() const
void enableShadowCasting(bool enabled = true, unsigned int SHADOW_MAP_SIZE_X = 0, unsigned int SHADOW_MAP_SIZE_Y = 0)

Enables or disables rendering of shadows cast by the unidirectional light.

Parameters:

  • enabled – Set to true to enable shadow casting (default at ctor=false).

  • SHADOW_MAP_SIZE_X – Width of the shadow cast map (1st pass of rendering with shadows). Larger values are slower but gives more precise shadows. Default=2048x2048. Zero means do not change.

  • SHADOW_MAP_SIZE_Y – Like SHADOW_MAP_SIZE_X but defines the height.

inline bool isShadowCastingEnabled() const
inline void enableSSAO(bool enabled = true)

Enable or disable Screen-Space Ambient Occlusion (SSAO). Parameters are tuned via lightParameters().ssao_* fields. Default: disabled.

inline bool isSSAOEnabled() const

3D ray from pixel coordinates

mrpt::math::TLine3D get3DRayForPixelCoord(const mrpt::img::TPixelCoord &pixelCoord, const mrpt::img::TPixelCoord &viewportSize, mrpt::poses::CPose3D *out_cameraPose = nullptr) const

Computes the 3D ray passing through the viewport pixel (x,y), given explicit viewport dimensions.

See also

getLastRenderedViewportSize

Parameters:

  • pixelCoord – Pixel coordinates (0=left/top, increasing right/down)

  • viewportSize – Viewport size in pixels

  • out_cameraPose – If not nullptr, receives the camera SE(3) pose

Returns:

The 3D ray in the world coordinates, starting at the camera, or nullopt if the viewport has not been rendered at least once (so that updateRenderedViewportSize() has been called by the rendering backend).

std::optional<mrpt::math::TLine3D> get3DRayForPixelCoord(const mrpt::img::TPixelCoord &pixelCoord, mrpt::poses::CPose3D *out_cameraPose = nullptr) const
inline std::pair<uint32_t, uint32_t> getLastRenderedViewportSize() const

Returns the size (width, height) of the viewport as rendered in the last frame. Returns {0,0} if not yet rendered.

inline void updateRenderedViewportSize(uint32_t w, uint32_t h) const

Called by the rendering backend to store the last rendered viewport size (in pixels).

Public Functions

~Viewport() override

Destructor: clears all objects.

Viewport(Scene *parent = nullptr, const std::string &name = std::string(""))

Constructor, invoked from Scene only.

Viewport(const Viewport&) = default
Viewport(Viewport&&) = default
Viewport &operator=(const Viewport&) = default
Viewport &operator=(Viewport&&) = default

Protected Functions

void dumpListOfObjects(std::vector<std::string> &lst) const

Retrieves a list of all objects in text form.

Deprecated:

Prefer asYAML() (since MRPT 2.1.3)

mrpt::containers::yaml asYAML() const

Prints all viewport objects in human-readable YAML form.

Note

(New in MRPT 2.1.3)

const CCamera *internalResolveActiveCamera(const CCamera *forceThisCamera = nullptr) const
void internal_enableImageView(bool transparentBackground)

Protected Attributes

mrpt::viz::CCamera m_camera

The camera associated to the viewport

mrpt::safe_ptr<Scene> m_parent

The scene that contains this viewport.

bool m_isCloned = false

Set by setCloneView

bool m_isClonedCamera = false

Set by setCloneCamera

bool m_isViewportVisible = true
std::string m_clonedViewport

Only if m_isCloned=true

std::string m_clonedCameraViewport

If m_isClonedCamera && !m_isCloned, take the camera from another view, to render a different scene.

std::string m_name

The viewport’s name

bool m_isTransparent = false

Whether to clear color buffer.

uint32_t m_borderWidth = 0

Default=0, the border around the viewport.

mrpt::img::TColor m_borderColor = {255, 255, 255, 255}
double m_view_x = {0}

The viewport position [0,1]

double m_view_y = {0}
double m_view_width = {1}
double m_view_height = {1}
float m_clip_min = 0.01f

The min/max clip depth distances (default: 0.01 - 1000)

float m_clip_max = 1000.0f
float m_lightShadowClipMin = 0.01f

The near/far plane clip distances for unidirectional light shadow casting

float m_lightShadowClipMax = 1000.0f
mrpt::img::TColorf m_background_color = {0.4f, 0.4f, 0.4f}
mrpt::viz::CTexturedPlane::Ptr m_imageViewPlane

The image to display, after calling setImageView()

mutable mrpt::viz::CSetOfLines::Ptr m_borderLines
mrpt::viz::ListVisualObjects m_objects
bool m_OpenGL_enablePolygonNicest = {true}
TLightParameters m_light
bool m_shadowsEnabled = false
uint32_t m_ShadowMapSizeX = 2048
uint32_t m_ShadowMapSizeY = 2048
mutable uint32_t m_lastRenderedWidth = 0

Updated by the rendering backend via updateRenderedViewportSize()

uint32_t m_lastRenderedHeight = 0