Is there anything preventing the support for a PixelBuffer buffer type and its corresponding PixelFormat.getIntArgbInstance() pixel format? That could make a big difference on slow devices for which copying the image is will include support for IntBuffer in the next rev. I'm curious what others think about this point. One could argue that since we don't actually take advantage of that right now, the callback might be overkill, but it seems safer from an API point of view. The thinking behind the callback is that it allows the implementation to: A) ensure that the image is always updated on the FX Application Thread, and/or B) done in a way that the implementation can know that the buffer is being updated and not to use it. What is the reasoning behind the callback function parameter to PixelBuffer.updateBuffer? Why won't a direct call with a Rectangle2D parameter do the same thing?. Updating a whole WritableImage(1000x1000) using PixelBuffer takes approximately 50% lesser time than compared to using PixelWriter.Ĭhange set includes a test program PixelBufferPerformanceTest, to observe the time taken. On next pulse, WritableImage gets drawn,Ħ.1 If the PixelBuffer was updated using updateBuffer() method only once, then only the dirty rect gets redrawn.Ħ.2 If the PixelBuffer was updated using updateBuffer() method more than once, then whole image gets redrawn.Ħ.3 If the WritableImage was updated using PixelWriter, then whole image gets redrawn(Existing behavior). updateBuffer() marks the Image as dirty and saves dirty rectangle. 
PixelBuffer can be updated using member method updateBuffer() or PixelWriter. The WritableImage uses same Buffer provided by PixelBuffer as pixel data and does not allocate new memory. Added a constructor to WritableImage, which creates an Image from PixelBuffer. 
This class holds the NIO Buffer which stores pixel data. This is first draft version of the change set, sharing here for initial feedback.īelow is the functionality of the change:
0 kommentar(er)
