SORRY. I forgot about a parameter (which has not solved my problem!) and apparently I almost found the problem.

Here is a new zip.

First, output_texture with jit.movie (I mean using textures instead of matrices) doesn't seem to work fine with poly~ Second, without using (output_texture 1), it seems to work (even if I have jit.gl.pix within the poly~) BUT, and it makes sense: This is only the latest voice of my poly~ that seems to render a result. We can "prove" that by sending (target 3, ) with n as a value in my patch.

I'd need to ... add, or blend or multiply all voices actually..

The idea is, again, I'd need to make matrices output to static output of my poly~... I guess. With a matrix router.. OR multiplex a lot of planes ?

Hi, Julian. I don't think your problem has anything to do with jit.gl.pix. In the main patcher, after you send the multiplying factors message (multislider/listfunnel/target x), you also change the destination of the matrix or gl texture to that voice target from that moment onwards. The poly~object focuses on the voice of the last target message received. So, when you send the movie's matrix or texture, you only get results on voice 3 (the last target message sent).

Regarding the composition of the result, it obviously cannot be done inside the poly~, because each instance only has it's own information. You could do a "sort of" accumulation buffer outside the poly~, where each voice's texture would be added in every frame and then cleared, but why not use OpenGL blending modes instead?