manim/manimlib/shaders/get_gl_Position.glsl

// Assumes the following uniforms exist in the surrounding context:
// uniform vec2 frame_shape;
// uniform float focal_distance;
// uniform float is_fixed_in_frame;

const vec2 DEFAULT_FRAME_SHAPE = vec2(8.0 * 16.0 / 9.0, 8.0);

vec4 get_gl_Position(vec3 point){
    vec4 result = vec4(point, 1.0);
    if(!bool(is_fixed_in_frame)){
        result.x *= 2.0 / frame_shape.x;
        result.y *= 2.0 / frame_shape.y;
        result.z *= 2.0 / frame_shape.y;  // Should we give the frame a z shape?  Does that make sense?
        result.z /= focal_distance;
        result.xy /= max(1.0 - result.z, 0.0);
        // Todo, does this discontinuity add weirdness?  Theoretically, by this result,
        // the z-coordiante of gl_Position only matter for z-indexing.  The reason
        // for thie line is to avoid agressive clipping of distant points.
        if(result.z < 0) result.z *= 0.1;
    } else{
        result.x *= 2.0 / DEFAULT_FRAME_SHAPE.x;
        result.y *= 2.0 / DEFAULT_FRAME_SHAPE.y;
    }
    result.z *= -1;
    return result;
}