On 10/14/07, Thomas Mayer thomas@dergrossebruder.org wrote:

...

[gemhead] | [pix_(movie|film|image)] <-- reg_exp for human reading only | [pix_texture] | [gemhead] | | | [pix_texture] | [square]

I can't figure out what this diagram is showing. Why are there two gemheads and pix_textures? Does it work if the second gemhead and pix_texture objects are removed?

zmoelnig@iem.at wrote:

the diagram shows: one [gemhead] is uploading an image to the gfx-card
via [pix_texture]. another gemhead uses this texture and displays it onto a square.

the 2nd outlet of [pix_texture] should output enough information, so
that another [pix_texture] (without a pix in the chain), can re-use
the same texture when it get's the info in its 2nd inlet.

Yes, that describes my setup exactly.

thomas: what is the problem? "not working" is a rather vague description. i recently discovered that the output of the 1st pix-texture would
give wrong texture coordinates when using normalized textures
(power-of-two sized). the w32-binaries certainly still have this bug.

The square is not visible, neither with or without texture, no error output in the Pd console.

for now i can think of two workarounds:

• if your machine/build supports it, try using rectangle-textures and

see if it works.

From what I gather from the [pix_texture] help patch,

rectangle-texturing is the default. IIRC while playing around with the environment mode for [pix_texture] (sending an [env $mode( to inlet 1), GL_Blend or GL_Add (2 or 3) made the square visible without the texture.

• if it is not supported, try manually setting the texture-coordinates

to 1,1 (e.g. use [pix_coordinate]) (you could also manually set the
texcoords by modifying the messages that comes out of [pix_texture];
it is human-readable and of the form (iirc): "<texid> <width> <height>
<texmode>"; you want to modify <width> and <height>)

I'll try that when I have my hand on the Vista machine again.

Thanks, Thomas