In my last question I asked how to render an (ortho) object behind everything. However, the more general problem to the that post is:
Is there any way to render exactly into some pixel on the screen? The JMECanvas I use has to placed "invisibly" within its environment, so especially to the border regions some bitmaps have to be drawn very precisely.
Could using a non power-of-two texture (which also had to be updated every few seconds) of a size of about 1600x1000 pixel work for that issue, or would it get distorted, e.g. by interpolation effects?
Is there any way to directly draw 2D or at least setPixel in the canvas' video-memory or image buffer or whatever?
Probably the easiest and most precise way to do it, not sure if it gets accelerated by hardware though.
Thanks for the info. I tried that yesterday, but the frame rate dropped dramatically. I guess I'll have to try the texture-approach lateron.
In that case the best way is probably to render in ortho mode with a textured quad applied to the whole screen. If the screen size & texture are of the same resolution you should get it to be pixel-accurate. Also I suggest you enable NEAREST filtering and disable mipmaps, those shouldn't really change anything but they can increase performance.
why don't you render the ortho quad alphatested texture first?
In that case the best way is probably to render in ortho mode with a textured quad applied to the whole screen. If the screen size & texture are of the same resolution you should get it to be pixel-accurate.
Yeah, I guess that's the fastest solution. I'll try that as soon as my BackgroundToTextureBridge is finished and post the results. Thanks.