from Dan Eicher (dna) - image.gl_load() was using GL_UNSIGNED_INT rather then GL_UNSIGNED_BYTE channel type. - changed image.gl_load() to use the image clamp settings (patch had this as an extra argument).