GLG2D is intended to speed up the most common painting operations. Not all Graphics2D methods are implemented according to the standard. In general, we try to give priority to the most commonly-used functions. If you have a request for a feature or want to report a bug, you can submit an issue on GitHub or send an email to borkhobs (at) gmail.
Below are a list of all the abstract functions in Graphics2D. Most methods are fully implemented, and some still have work left in order to behave like the Sun implementation.
| draw(Shape) | yes |
| drawString(String, int, int) | yes |
| drawString(String, float, float) | yes |
| drawString(AttributedCharacterIterator, int, int) | does not draw attributes |
| drawString(AttributedCharacterIterator, float, float) | does not draw attributes |
| drawGlyphVector(GlyphVector, float, float) | not yet |
| fill(Shape) | yes |
| hit(Rectangle, Shape, boolean) | yes |
| getDeviceConfiguration() | yes |
| getComposite() | yes |
| setComposite(Composite) | not all composite types are supported |
| setPaint(Paint) | only accepts Color |
| setRenderingHint(Key, Object) | accepts a subset of the rendering hints (e.g. anti-aliasing) |
| getRenderingHint(Key) | accepts a subset of the rendering hints (e.g. anti-aliasing) |
| setRenderingHints(Map) | accepts a subset of the rendering hints (e.g. anti-aliasing) |
| addRenderingHints(Map) | accepts a subset of the rendering hints (e.g. anti-aliasing) |
| getRenderingHints() | yes |
| translate(int, int) | yes |
| translate(double, double) | yes |
| rotate(double) | yes |
| rotate(double, double, double) | yes |
| scale(double, double) | yes |
| shear(double, double) | yes |
| transform(AffineTransform) | yes |
| setTransform(AffineTransform) | yes |
| getTransform() | yes |
| getPaint() | yes |
| getColor() | yes |
| setColor(Color) | yes |
| setBackground(Color) | yes |
| getBackground() | yes |
| getStroke() | yes |
| setStroke(Stroke) | yes |
| setPaintMode() | yes |
| setXORMode(Color) | not yet |
| getFont() | yes |
| setFont(Font) | yes |
| getFontMetrics(Font) | yes |
| getFontRenderContext() | yes |
| getClipBounds() | yes |
| clip(Shape s) | yes |
| clipRect(int, int, int, int) | yes |
| setClip(int, int, int, int) | yes |
| getClip() | yes |
| setClip(Shape clipShape) | yes |
| copyArea(int, int, int, int, int, int) | yes |
| drawLine(int, int, int, int) | yes |
| fillRect(int, int, int, int) | yes |
| clearRect(int, int, int, int) | yes |
| drawRoundRect(int, int, int, int, int, int) | yes |
| fillRoundRect(int, int, int, int, int, int) | yes |
| drawOval(int, int, int, int) | yes |
| fillOval(int, int, int, int) | yes |
| drawArc(int, int, int, int, int, int) | yes |
| fillArc(int, int, int, int, int, int) | yes |
| drawPolyline(int[], int[], int) | yes |
| drawPolygon(int[], int[], int) | yes |
| fillPolygon(int[], int[], int) | yes |
| drawImage(Image, AffineTransform, ImageObserver) | yes |
| drawImage(BufferedImage, BufferedImageOp, int, int) | yes |
| drawRenderedImage(RenderedImage, AffineTransform) | not yet |
| drawRenderableImage(RenderableImage, AffineTransform) | not yet |
| drawImage(Image, int, int, ImageObserver) | yes |
| drawImage(Image, int, int, Color, ImageObserver) | yes |
| drawImage(Image, int, int, int, int, ImageObserver) | yes |
| drawImage(Image, int, int, int, int, Color, ImageObserver) | yes |
| drawImage(Image, int, int, int, int, int, int, int, int, ImageObserver) | yes |
| drawImage(Image, int, int, int, int, int, int, int, int, Color) | yes |
| create() | yes |
| dispose() | yes |
The initial implementation of GLG2D used just the fixed-function OpenGL pipeline. However, there are several methods in Graphics2D that need shaders. The most obvious one is Graphics2D.setPaint(Paint). A fragment shader could paint a gradient or any other kind of paint, whereas the fixed function pipeline can only paint a solid color (easily). This is why the current implementation status of setPaint(Paint) is partial.
Another interesting use of shaders is rendering a shape with a BasicStroke. The Stroke interface specifies only one method, createStrokedShape(Shape). That means theoretically you would have to tessellate every shape outline — a very expensive task! Instead, GLG2D determines if you're using a BasicStroke and uses a specialized renderer that draws the stroke using GL_QUADS. But all the vertex calculation is done on the CPU. This could be offloaded to the GPU with a geometry shader.
GLG2D — fast Java2D painting using OpenGL © Brandon Borkholder 2013