Already applied your modifications on the code.

thanks again!

byte[] line = new byte[width * 3];
int srcN = 0;
for (int y = height; y > 0; y–) {
int dstN = 0;
for (int x = width; x > 0; x–) {
int rgb = conversionBuffer[srcN++];
line[dstN++] = (byte) (rgb >> 16);
line[dstN++] = (byte) (rgb >> 8);
line[dstN++] = (byte) (rgb >> 0);
}
byteBuffer.put(line);
}

I get go from 50% CPU usage to 30% with this.

The problem is that you haven’t registered the custom renderer I’ve created on JMF, so when you are executing the getRenderer method it gets the default JMF renderer (Java2DRenderer) that obviously cannot be cast to my custom IGLTextureRenderer interface.

To fix this, simply follow the steps above on the paragraph that contains “register the render on JMF using the JMFRegistry”

Drop me by if you have any more problems.

Thanks for passing by!

Something got wrong!
java.lang.ClassCastException: com.omnividea.media.renderer.video.Java2DRenderer cannot be cast to org.pirelenito.multimedia.jmf.plugin.IGLTextureRenderer
at org.pirelenito.multimedia.jmf.MoviePlayer.getRenderer(MoviePlayer.java:48)
at org.pirelenito.movieGL.Main.(Main.java:104)
at org.pirelenito.movieGL.Main.main(Main.java:74)

I don’t understand what the problem is…

That conversion part is something I want to avoid by using popper pixel types, that is why I didn’t look on doing it through JNI (:p), I hope I can get rid of it.

Check latter for more updates on the code… I am still messing with it!

Interesting way to solve the problem =]

Using PBO would certainly increase performance, since you wouldn’t need to ‘draw frames to a texture’, but would be writing directly to the buffer. One thing I think (THINK) that would still make my knees shake is the conversion from the media data format to RGB texture format (that ‘for’ statement made me feel a huge fear… look at http://nehe.gamedev.net/data/lessons/lesson.asp?lesson=35, the ‘for’ thing is also needed but was done through assembly code [REAL PROGRAMMERS CODE IN BINARY, aeouahaeo, or NOT!], don’t know if Java’s JIT is capable of increasing the ‘for’ perfomance), it (format conversion, if you got lost at the last ‘parenthesis’ statement.. hehe) is needed in both methods.

Congratulations! It’s a pretty code chunk!!!

aeuaeohaouhea

Kinda hilarious stuff.

Thanks for pointing it out!

Guess I’m gonna to copy take some inspiration… BTW, I’ve gotta post something =~

See ya, dude!