Update and partial solutions, briefly but happily.

As recommended I re-arranged the code so that textures would be created only once in the setup, and then updated by the loop. It made a true difference, I thought creation of textures with the same variable name would wipe and replace the texture, definetly not (or I don't know how to do that). But it still leaked and was not proper for the installation. Tried billions of funny things and found System.gc() called regularly, it made the situation better. But still...

Then I run my test application on a smaller computer with a Nvidia graphic cards... and the memory ressource graph raises and then stay flated... oO

Here is the test code I used on the two computers
Code:

import codeanticode.glgraphics.*;
import processing.opengl.*;

float count;
PImage img;
GLTexture tex;

void setup() {
  size(1280,720,GLConstants.GLGRAPHICS);
  count = 0;
  tex = new GLTexture(this,1280,720,JPEG);
}

void draw() {

  if(count==0) {
    img = loadImage("img.jpg");
  }

  tex.putImage(img,JPEG);
  image(tex,0,0,width,height);
  count++;

  if(count==20) {
    System.gc();
    background(0);
    count=0; 
  }

}
 

I can't believe it is driver/hardware issue... But I managed to run my photomaton application on this nvidia powered laptop with no outragous memory consumption or leakage, so I might try to borrow one of these for the project's machine.

@ ac : the library update did not make noticeable improvements from what I could see. But I went thru the doc and there are many good things cooking in there.

So this (in my opinion) is not a real solution to my original problem, and I WILL go deeper in GL manipulation to get better at it. And try to gather infos concerning the possible differences between those two manufacturers.

dF