OpenSceneGraph/include/osgText/Text

433 lines
19 KiB
Plaintext
Raw Normal View History

/* -*-c++-*- OpenSceneGraph - Copyright (C) 1998-2006 Robert Osfield
*
* This library is open source and may be redistributed and/or modified under
* the terms of the OpenSceneGraph Public License (OSGPL) version 0.0 or
* (at your option) any later version. The full license is in LICENSE file
* included with this distribution, and on the openscenegraph.org website.
*
* This library is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* OpenSceneGraph Public License for more details.
*/
#ifndef OSGTEXT_TEXT
#define OSGTEXT_TEXT 1
#include <osg/Drawable>
#include <osg/Quat>
#include <osgText/TextBase>
#include <osgText/Font>
namespace osgText {
class OSGTEXT_EXPORT Text : public osgText::TextBase
{
public:
Text();
Text(const Text& text,const osg::CopyOp& copyop=osg::CopyOp::SHALLOW_COPY);
virtual osg::Object* cloneType() const { return new Text(); }
virtual osg::Object* clone(const osg::CopyOp& copyop) const { return new Text(*this,copyop); }
virtual bool isSameKindAs(const osg::Object* obj) const { return dynamic_cast<const Text*>(obj)!=NULL; }
virtual const char* className() const { return "Text"; }
virtual const char* libraryName() const { return "osgText"; }
virtual void setFont(Font* font=0) { setFont(osg::ref_ptr<Font>(font)); };
/** Set the Font to use to render the text.*/
virtual void setFont(osg::ref_ptr<Font> font);
/** Set the font, loaded from the specified front file, to use to render the text,
* setFont("") sets the use of the default font.
* See the osgText::readFontFile function for how the font file will be located. */
virtual void setFont(const std::string& fontfile) { TextBase::setFont(fontfile); }
From March Sciabica, "Here is the solution I coded up over the weekend. For improved performance option, I added a flag to control whether the depth writing pass is performed. Since text is alpha-blended when rendering, it is placed in the transparent bin and rendered back to front. Writing to the depth buffer should therefore be unnecessary. Indeed, rendering something behind text (or any blended object) after that object is drawn will give incorrect results whether the depth buffer is written or not. I therefore think it is safe to keep this option off by default. Users can turn it on for any special needs they have. I did not fix the existing backdrop implementations to work with the new code since this new method of rendering intrinsically handles backdrops correctly. Its results are more accurate than all of the existing backdrop implementations. Its only downside is that it requires two passes if depth buffer updates are desired, whereas DEPTH_RANGE and POLYGON_OFFSET achieve their (less accurate) results in one pass. The NO_DEPTH_BUFFER method also only uses one pass, but it disables depth tests and not depth writes so will have serious problems if anything is drawn in front of the text before OR after the text is drawn. Given the better all-around behavior of the new method, I believe the other backdrop implementations can be safely removed. Code that adjusts the backdrop implementation will of course be broken if the member functions are removed. For this reason I left them in, but set the new rendering method as the default backdrop implementation. At the very least I think the old backdrop implementations should be deprecated and removed at a later date. " Note from Robert Osfield, testing this submission with osgtext I found that the text would not render correctly when different text labels were overlapping in deth and screen space. I change _enableDepthWrites to default to true and found the that which artifacts still occurred around the alpha blended edges the artifacts where better than issue with occlusion of nearer pixels that was happening with _enableDepthWrites set to false.I therefore set the _enableDepthWrites to true as I feel it's the lesser of the two artefacts.
2010-09-16 17:09:43 +08:00
/**
* Turns off writing to the depth buffer when rendering text. This only affects text
* with no backdrop or text using the DELAYED_DEPTH_WRITES implementation, since
* the other backdrop implementations are really only useful for backwards
* compatibility and are not worth updating to utilize this flag.
*/
void setEnableDepthWrites(bool enable) { _enableDepthWrites = enable; }
2010-10-27 21:49:16 +08:00
bool getEnableDepthWrites() const { return _enableDepthWrites; }
From March Sciabica, "Here is the solution I coded up over the weekend. For improved performance option, I added a flag to control whether the depth writing pass is performed. Since text is alpha-blended when rendering, it is placed in the transparent bin and rendered back to front. Writing to the depth buffer should therefore be unnecessary. Indeed, rendering something behind text (or any blended object) after that object is drawn will give incorrect results whether the depth buffer is written or not. I therefore think it is safe to keep this option off by default. Users can turn it on for any special needs they have. I did not fix the existing backdrop implementations to work with the new code since this new method of rendering intrinsically handles backdrops correctly. Its results are more accurate than all of the existing backdrop implementations. Its only downside is that it requires two passes if depth buffer updates are desired, whereas DEPTH_RANGE and POLYGON_OFFSET achieve their (less accurate) results in one pass. The NO_DEPTH_BUFFER method also only uses one pass, but it disables depth tests and not depth writes so will have serious problems if anything is drawn in front of the text before OR after the text is drawn. Given the better all-around behavior of the new method, I believe the other backdrop implementations can be safely removed. Code that adjusts the backdrop implementation will of course be broken if the member functions are removed. For this reason I left them in, but set the new rendering method as the default backdrop implementation. At the very least I think the old backdrop implementations should be deprecated and removed at a later date. " Note from Robert Osfield, testing this submission with osgtext I found that the text would not render correctly when different text labels were overlapping in deth and screen space. I change _enableDepthWrites to default to true and found the that which artifacts still occurred around the alpha blended edges the artifacts where better than issue with occlusion of nearer pixels that was happening with _enableDepthWrites set to false.I therefore set the _enableDepthWrites to true as I feel it's the lesser of the two artefacts.
2010-09-16 17:09:43 +08:00
enum BackdropType
{
DROP_SHADOW_BOTTOM_RIGHT = 0, // usually the type of shadow you see
DROP_SHADOW_CENTER_RIGHT,
DROP_SHADOW_TOP_RIGHT,
DROP_SHADOW_BOTTOM_CENTER,
DROP_SHADOW_TOP_CENTER,
DROP_SHADOW_BOTTOM_LEFT,
DROP_SHADOW_CENTER_LEFT,
DROP_SHADOW_TOP_LEFT,
OUTLINE,
NONE
};
enum BackdropImplementation
{
/* POLYGON_OFFSET:
* This uses glPolygonOffset to draw the text multiple times to
* create the drop-shadow and outline effects. glPolygonOffset
* is used to prevent z-fighting of the overlapping text.
* This probably should have been the best option, but all the ATI
* cards we have encountered so far have serious problems with this.
* We see little white holes/artifacts in the rendered glyph textures
* which move around depending on the viewing angle. For moving text,
* the moving holes give an extremely unpleasant flickering effect.
* Pumping up the "units" parameter in glPolygonOffset can minimize
* this problem, but two other bad side-effects occur if you do this.
* First, high values will cause problems with clipping, particularly
* when there are objects behind the text. The drop-shadows or outline
* may be culled because their computed offset is behind the object or
* z-far plane. Second, there is an additional problem associated with
* the Z-slope. High values can make large chunks of the backdrop
* suddenly disappear. This can be reduced by the "factor" parameter.
* Making the "factor" value small, can help, but experimentally, we've
* found that it creates a new, different kind of z-fighting problem.
* So there is no perfect solution. With units, you trade off the 'holes'
* for the large-section clipping.
* Experimentally, we have found units values from 150-512 to be tolerable
* to acceptable with respect to the 'holes'. A factor of .1 seems to
* bring down the large clipping problem without creating a new z-fighting
* problem.
* (You can experiment with these numbers by playing with the
* osg:PolygonOffset multipliers which this backend tries to respect.)
*
* If ATI ever fixes their cards/drivers, then this might become the
* best option.*/
POLYGON_OFFSET = 0,
/* NO_DEPTH_BUFFER
* Instead of using glPolygonOffset to prevent z-fighting, this mode
* just disables the depth buffer when rendering the text. This allows
* the text to be rendered without any z-fighting. The downside to this
* mode is that render order begins to matter and the text will not
* necessarily correctly appear above or behind other objects in the
* scene based on depth values.
* This mode is best for text that only needs to be ontop and
* not obscured by any objects.*/
NO_DEPTH_BUFFER,
/* DEPTH_RANGE
* This mode is inspired by Paul Martz's OpenGL FAQ, item 13.050.
* This uses glDepthRange as a substitute for glPolygonOffset.
* Strangely, experiments on ATI cards seem to produce cleaner results
* than when using glPolygonOffset. The trade-off for this is that the
* backdrop still may be placed too far back and might be culled by objects
* directly behind the object or by the far z-plane. If ATI ever fixes
* the glPolygonOffset problem, polygon offset is probably a slightly
* better solution because you can use smaller offsets. But with the
* current ATI problem, this option may be preferable.*/
DEPTH_RANGE,
/* STENCIL_BUFFER
* (Assuming the backend is written correctly,) the Stencil Buffer is
* the most "correct" and reliable way of producing backdrop text.
* The stencil buffer is a multipass system that allows writing to the
* same z-values without needing to resort to offsets. This implementation
* should not have any of the problems associated with the 3 previous
* implementations. But the trade-off for this mode is that without
* hardware acceleration for the stencil buffer, rendering will be
* extremely slow. (There is also potentially more overhead for this
* algorithm so it could be slower than the other implementations.
* Benchmarking would be required to determine if the speed differences
* are significant on your particular hardware.) This mode is best for
* when quality is important and stencil buffer hardware acceleration
* is available.*/
From March Sciabica, "Here is the solution I coded up over the weekend. For improved performance option, I added a flag to control whether the depth writing pass is performed. Since text is alpha-blended when rendering, it is placed in the transparent bin and rendered back to front. Writing to the depth buffer should therefore be unnecessary. Indeed, rendering something behind text (or any blended object) after that object is drawn will give incorrect results whether the depth buffer is written or not. I therefore think it is safe to keep this option off by default. Users can turn it on for any special needs they have. I did not fix the existing backdrop implementations to work with the new code since this new method of rendering intrinsically handles backdrops correctly. Its results are more accurate than all of the existing backdrop implementations. Its only downside is that it requires two passes if depth buffer updates are desired, whereas DEPTH_RANGE and POLYGON_OFFSET achieve their (less accurate) results in one pass. The NO_DEPTH_BUFFER method also only uses one pass, but it disables depth tests and not depth writes so will have serious problems if anything is drawn in front of the text before OR after the text is drawn. Given the better all-around behavior of the new method, I believe the other backdrop implementations can be safely removed. Code that adjusts the backdrop implementation will of course be broken if the member functions are removed. For this reason I left them in, but set the new rendering method as the default backdrop implementation. At the very least I think the old backdrop implementations should be deprecated and removed at a later date. " Note from Robert Osfield, testing this submission with osgtext I found that the text would not render correctly when different text labels were overlapping in deth and screen space. I change _enableDepthWrites to default to true and found the that which artifacts still occurred around the alpha blended edges the artifacts where better than issue with occlusion of nearer pixels that was happening with _enableDepthWrites set to false.I therefore set the _enableDepthWrites to true as I feel it's the lesser of the two artefacts.
2010-09-16 17:09:43 +08:00
STENCIL_BUFFER,
/* DELAYED_DEPTH_WRITES
* This mode renders all text with depth writes turned off, then
* again with depth writes on, but with the color buffer disabled.
* This should render text accurately for all graphics cards. The
* only downside is the additional pass to render to the depth
* buffer. But if you don't need the depth buffer updated for
* your, this extra pass can be disabled by calling
* enableDepthWrites(false).*/
DELAYED_DEPTH_WRITES
};
/**
* BackdropType gives you a background shadow text behind your regular
* text. This helps give text extra contrast which can be useful when
* placing text against noisy backgrounds.
* The color of the background shadow text is specified by setBackdropColor().
* DROP_SHADOW_BOTTOM_RIGHT will draw backdrop text to the right and down of
* the normal text. Other DROW_SHADOW_* modes do the same for their repective directions.
* OUTLINE will draw backdrop text so that it appears the text has an outline
* or border around the normal text. This mode is particularly useful against
* really noisy backgrounds that may put text on top of things that have
* all types of colors which you don't have control over.
* Some real world examples of this general technique in use that I know of
* are Google Earth, Sid Meier's Pirates (2004 Remake), and Star Control 2 (PC 1993).
* The default is NONE.
*/
void setBackdropType(BackdropType type);
BackdropType getBackdropType() const { return _backdropType; }
/**
* Sets the amount text is offset to create the backdrop/shadow effect.
* Set the value too high and for example, in OUTLINE mode you will get a "Brady Bunch"
* effect where you see duplicates of the text in a 3x3 grid.
* Set the value too small and you won't see anything.
* The values represent percentages. 1.0 means 100% so a value of 1.0
* in DROW_SHADOW_LEFT_CENTER mode would cause each glyph to be echoed
* next to it self. So the letter 'e' might look like 'ee'.
* Good values tend to be in the 0.03 to 0.10 range (but will be subject
* to your specific font and display characteristics).
* Note that the text bounding boxes are updated to include backdrop offsets.
* However, other metric information such as getCharacterHeight() are unaffected
* by this. This means that individual glyph spacing (kerning?) are unchanged
* even when this mode is used.
* The default is 0.07 (7% offset).
*/
void setBackdropOffset(float offset = 0.07f);
/**
* This overloaded version lets you specify the offset for the horizontal
* and vertical components separately.
*/
void setBackdropOffset(float horizontal, float vertical);
2006-11-08 19:49:35 +08:00
float getBackdropHorizontalOffset() const { return _backdropHorizontalOffset; }
float getBackdropVerticalOffset() const { return _backdropVerticalOffset; }
/**
* This specifies the color of the backdrop text.
* The default is black.
*/
void setBackdropColor(const osg::Vec4& color);
const osg::Vec4& getBackdropColor() const { return _backdropColor; }
/**
* This specifies the underlying backdrop rendering implementation.
* Unfortunately, at this time, there is no "perfect" rendering solution
* so this function is provided to let you 'pick your poison'. Each
* implementation has trade-offs. See BackdropImplementation enum
* docs for details.*/
void setBackdropImplementation(BackdropImplementation implementation);
BackdropImplementation getBackdropImplementation() const { return _backdropImplementation; }
enum ColorGradientMode
{
SOLID = 0, // a.k.a. ColorGradients off
PER_CHARACTER,
OVERALL
};
/**
* This sets different types of text coloring modes.
* When the coloring mode is not set to SOLID, the
* colors specified in setColorGradientCorners() determine
* the colors for the text.
* When the gradient mode is OVERALL, the coloring scheme
* attempts to approximate the effect as if the entire text box/region
* were a single polygon and you had applied colors to each of the four
* corners with GL_SMOOTH enabled. In this mode, OpenGL interpolates
* the colors across the polygon, and this is what OVERALL tries to
* emulate. This can be used to give nice embellishments on things
* like logos and names.
* PER_CHARACTER is similar to OVERALL except that it applies the
* color interpolation to the four corners of each character instead
* of across the overall text box.
* The default is SOLID (a.k.a. off).
*/
void setColorGradientMode(ColorGradientMode mode);
ColorGradientMode getColorGradientMode() const { return _colorGradientMode; }
/**
* Used only for gradient mode, let's you specify the colors of the 4 corners.
* If ColorGradients are off, these values are ignored (and the value from setColor()
* is the only one that is relevant.
*/
void setColorGradientCorners(const osg::Vec4& topLeft, const osg::Vec4& bottomLeft, const osg::Vec4& bottomRight, const osg::Vec4& topRight);
const osg::Vec4& getColorGradientTopLeft() const { return _colorGradientTopLeft; }
const osg::Vec4& getColorGradientBottomLeft() const { return _colorGradientBottomLeft; }
const osg::Vec4& getColorGradientBottomRight() const { return _colorGradientBottomRight; }
const osg::Vec4& getColorGradientTopRight() const { return _colorGradientTopRight; }
/** Draw the text.*/
virtual void drawImplementation(osg::RenderInfo& renderInfo) const;
2003-03-11 00:40:26 +08:00
/** return false, osgText::Text does not support accept(AttributeFunctor&).*/
virtual bool supports(const osg::Drawable::AttributeFunctor&) const { return false; }
2003-03-11 00:40:26 +08:00
/** return true, osgText::Text does support accept(ConstAttributeFunctor&).*/
virtual bool supports(const osg::Drawable::ConstAttributeFunctor&) const { return true; }
2003-03-11 00:40:26 +08:00
/** accept an ConstAttributeFunctor and call its methods to tell it about the internal attributes that this Drawable has.*/
2003-03-11 00:40:26 +08:00
virtual void accept(osg::Drawable::ConstAttributeFunctor& af) const;
/** return true, osgText::Text does support accept(PrimitiveFunctor&) .*/
virtual bool supports(const osg::PrimitiveFunctor&) const { return true; }
2003-03-11 00:40:26 +08:00
/** accept a PrimtiveFunctor and call its methods to tell it about the internal primitives that this Drawable has.*/
virtual void accept(osg::PrimitiveFunctor& pf) const;
2003-03-11 00:40:26 +08:00
/** Set whether to use a mutex to ensure ref() and unref() are thread safe.*/
virtual void setThreadSafeRefUnref(bool threadSafe);
/** Resize any per context GLObject buffers to specified size. */
virtual void resizeGLObjectBuffers(unsigned int maxSize);
/** If State is non-zero, this function releases OpenGL objects for
* the specified graphics context. Otherwise, releases OpenGL objexts
* for all graphics contexts. */
virtual void releaseGLObjects(osg::State* state=0) const;
2003-03-11 00:40:26 +08:00
public:
// internal structures, variable and methods used for rendering of characters.
struct OSGTEXT_EXPORT GlyphQuads
{
typedef std::vector<Glyph*> Glyphs;
typedef std::vector<unsigned int> LineNumbers;
typedef osg::ref_ptr<osg::Vec2Array> Coords2;
typedef osg::ref_ptr<osg::Vec3Array> Coords3;
typedef osg::ref_ptr<osg::Vec2Array> TexCoords;
typedef osg::ref_ptr<osg::Vec4Array> ColorCoords;
Glyphs _glyphs;
Coords2 _coords;
osg::buffered_object<Coords3> _transformedCoords;
TexCoords _texcoords;
LineNumbers _lineNumbers;
osg::buffered_object<Coords3> _transformedBackdropCoords[8];
ColorCoords _colorCoords;
osg::ref_ptr<osg::DrawElements> _quadIndices;
void updateQuadIndices();
GlyphQuads();
GlyphQuads(const GlyphQuads& gq);
void initGlyphQuads();
void initGPUBufferObjects();
Glyphs& getGlyphs() { return _glyphs; }
const Glyphs& getGlyphs() const { return _glyphs; }
Coords2& getCoords() { return _coords; }
const Coords2& getCoords() const { return _coords; }
Coords3& getTransformedCoords(unsigned int contexID) { return _transformedCoords[contexID]; }
const Coords3& getTransformedCoords(unsigned int contexID) const { return _transformedCoords[contexID]; }
TexCoords& getTexCoords() { return _texcoords; }
const TexCoords& getTexCoords() const { return _texcoords; }
LineNumbers& getLineNumbers() { return _lineNumbers; }
const LineNumbers& getLineNumbers() const { return _lineNumbers; }
/** Resize any per context GLObject buffers to specified size. */
void resizeGLObjectBuffers(unsigned int maxSize);
/** If State is non-zero, this function releases OpenGL objects for
* the specified graphics context. Otherwise, releases OpenGL objexts
* for all graphics contexts. */
void releaseGLObjects(osg::State* state=0) const;
private:
GlyphQuads& operator = (const GlyphQuads&) { return *this; }
};
typedef std::map<osg::ref_ptr<GlyphTexture>,GlyphQuads> TextureGlyphQuadMap;
/** Direct Access to GlyphQuads */
const GlyphQuads* getGlyphQuads(GlyphTexture* texture) const
{
TextureGlyphQuadMap::iterator itGlyphQuad = _textureGlyphQuadMap.find(texture);
if (itGlyphQuad == _textureGlyphQuadMap.end()) return NULL;
return &itGlyphQuad->second;
}
const TextureGlyphQuadMap& getTextureGlyphQuadMap() const
{
return _textureGlyphQuadMap;
}
protected:
virtual ~Text();
Font* getActiveFont();
const Font* getActiveFont() const;
String::iterator computeLastCharacterOnLine(osg::Vec2& cursor, String::iterator first,String::iterator last);
// members which have public access.
// iternal map used for rendering. Set up by the computeGlyphRepresentation() method.
2003-06-24 23:40:09 +08:00
mutable TextureGlyphQuadMap _textureGlyphQuadMap;
void computeGlyphRepresentation();
// internal caches of the positioning of the text.
bool computeAverageGlyphWidthAndHeight(float& avg_width, float& avg_height) const;
virtual void computePositions(unsigned int contextID) const;
void computeBackdropPositions(unsigned int contextID) const;
void computeBackdropBoundingBox() const;
void computeBoundingBoxMargin() const;
void computeColorGradients() const;
void computeColorGradientsOverall() const;
void computeColorGradientsPerCharacter() const;
void drawImplementation(osg::State& state, const osg::Vec4& colorMultiplier) const;
void drawForegroundText(osg::State& state, const GlyphQuads& glyphquad, const osg::Vec4& colorMultiplier) const;
From March Sciabica, "Here is the solution I coded up over the weekend. For improved performance option, I added a flag to control whether the depth writing pass is performed. Since text is alpha-blended when rendering, it is placed in the transparent bin and rendered back to front. Writing to the depth buffer should therefore be unnecessary. Indeed, rendering something behind text (or any blended object) after that object is drawn will give incorrect results whether the depth buffer is written or not. I therefore think it is safe to keep this option off by default. Users can turn it on for any special needs they have. I did not fix the existing backdrop implementations to work with the new code since this new method of rendering intrinsically handles backdrops correctly. Its results are more accurate than all of the existing backdrop implementations. Its only downside is that it requires two passes if depth buffer updates are desired, whereas DEPTH_RANGE and POLYGON_OFFSET achieve their (less accurate) results in one pass. The NO_DEPTH_BUFFER method also only uses one pass, but it disables depth tests and not depth writes so will have serious problems if anything is drawn in front of the text before OR after the text is drawn. Given the better all-around behavior of the new method, I believe the other backdrop implementations can be safely removed. Code that adjusts the backdrop implementation will of course be broken if the member functions are removed. For this reason I left them in, but set the new rendering method as the default backdrop implementation. At the very least I think the old backdrop implementations should be deprecated and removed at a later date. " Note from Robert Osfield, testing this submission with osgtext I found that the text would not render correctly when different text labels were overlapping in deth and screen space. I change _enableDepthWrites to default to true and found the that which artifacts still occurred around the alpha blended edges the artifacts where better than issue with occlusion of nearer pixels that was happening with _enableDepthWrites set to false.I therefore set the _enableDepthWrites to true as I feel it's the lesser of the two artefacts.
2010-09-16 17:09:43 +08:00
void drawTextWithBackdrop(osg::State& state, const osg::Vec4& colorMultiplier) const;
void renderOnlyForegroundText(osg::State& state, const osg::Vec4& colorMultiplier) const;
void renderWithPolygonOffset(osg::State& state, const osg::Vec4& colorMultiplier) const;
void renderWithNoDepthBuffer(osg::State& state, const osg::Vec4& colorMultiplier) const;
void renderWithDepthRange(osg::State& state, const osg::Vec4& colorMultiplier) const;
void renderWithStencilBuffer(osg::State& state, const osg::Vec4& colorMultiplier) const;
From March Sciabica, "Here is the solution I coded up over the weekend. For improved performance option, I added a flag to control whether the depth writing pass is performed. Since text is alpha-blended when rendering, it is placed in the transparent bin and rendered back to front. Writing to the depth buffer should therefore be unnecessary. Indeed, rendering something behind text (or any blended object) after that object is drawn will give incorrect results whether the depth buffer is written or not. I therefore think it is safe to keep this option off by default. Users can turn it on for any special needs they have. I did not fix the existing backdrop implementations to work with the new code since this new method of rendering intrinsically handles backdrops correctly. Its results are more accurate than all of the existing backdrop implementations. Its only downside is that it requires two passes if depth buffer updates are desired, whereas DEPTH_RANGE and POLYGON_OFFSET achieve their (less accurate) results in one pass. The NO_DEPTH_BUFFER method also only uses one pass, but it disables depth tests and not depth writes so will have serious problems if anything is drawn in front of the text before OR after the text is drawn. Given the better all-around behavior of the new method, I believe the other backdrop implementations can be safely removed. Code that adjusts the backdrop implementation will of course be broken if the member functions are removed. For this reason I left them in, but set the new rendering method as the default backdrop implementation. At the very least I think the old backdrop implementations should be deprecated and removed at a later date. " Note from Robert Osfield, testing this submission with osgtext I found that the text would not render correctly when different text labels were overlapping in deth and screen space. I change _enableDepthWrites to default to true and found the that which artifacts still occurred around the alpha blended edges the artifacts where better than issue with occlusion of nearer pixels that was happening with _enableDepthWrites set to false.I therefore set the _enableDepthWrites to true as I feel it's the lesser of the two artefacts.
2010-09-16 17:09:43 +08:00
void renderWithDelayedDepthWrites(osg::State& state, const osg::Vec4& colorMultiplier) const;
From March Sciabica, "Here is the solution I coded up over the weekend. For improved performance option, I added a flag to control whether the depth writing pass is performed. Since text is alpha-blended when rendering, it is placed in the transparent bin and rendered back to front. Writing to the depth buffer should therefore be unnecessary. Indeed, rendering something behind text (or any blended object) after that object is drawn will give incorrect results whether the depth buffer is written or not. I therefore think it is safe to keep this option off by default. Users can turn it on for any special needs they have. I did not fix the existing backdrop implementations to work with the new code since this new method of rendering intrinsically handles backdrops correctly. Its results are more accurate than all of the existing backdrop implementations. Its only downside is that it requires two passes if depth buffer updates are desired, whereas DEPTH_RANGE and POLYGON_OFFSET achieve their (less accurate) results in one pass. The NO_DEPTH_BUFFER method also only uses one pass, but it disables depth tests and not depth writes so will have serious problems if anything is drawn in front of the text before OR after the text is drawn. Given the better all-around behavior of the new method, I believe the other backdrop implementations can be safely removed. Code that adjusts the backdrop implementation will of course be broken if the member functions are removed. For this reason I left them in, but set the new rendering method as the default backdrop implementation. At the very least I think the old backdrop implementations should be deprecated and removed at a later date. " Note from Robert Osfield, testing this submission with osgtext I found that the text would not render correctly when different text labels were overlapping in deth and screen space. I change _enableDepthWrites to default to true and found the that which artifacts still occurred around the alpha blended edges the artifacts where better than issue with occlusion of nearer pixels that was happening with _enableDepthWrites set to false.I therefore set the _enableDepthWrites to true as I feel it's the lesser of the two artefacts.
2010-09-16 17:09:43 +08:00
bool _enableDepthWrites;
BackdropType _backdropType;
BackdropImplementation _backdropImplementation;
float _backdropHorizontalOffset;
float _backdropVerticalOffset;
osg::Vec4 _backdropColor;
ColorGradientMode _colorGradientMode;
osg::Vec4 _colorGradientTopLeft;
osg::Vec4 _colorGradientBottomLeft;
osg::Vec4 _colorGradientBottomRight;
osg::Vec4 _colorGradientTopRight;
// Helper function for color interpolation
float bilinearInterpolate(float x1, float x2, float y1, float y2, float x, float y, float q11, float q12, float q21, float q22) const;
};
}
#endif