Maybe the photos will help, although it's hard for me to visualize how the side of the sphere facing the camera can reflect a point that's 180 degrees opposite the camera (in both polar coordinates). I can't use HDRShop (Mac OS), but if you want to IM me, I'll provide my e-mail address. The question of the sphere polygon count brings up two other questions. 1. Using a sphere's glow channel as the environment light source, would it average all the pixels of the region of the texture map associated in any given polygon, or would each polygon produce a variegated light that mimicked the variations in pixel brightness of that region of the texture map? Meaning that a 60-polygon sphere might produce a more uniform light source than a 360-polygon sphere, using the same glow map. 2. Is the HDRI engine in C3 also resolution-limited? Meaning it might average blocks of pixels in the same way as a sphere would, when used as an environment light in C2. One way to test this is to see if a 1500X3000 HDRI causes a render several orders of magnitude slower than a 150X300 HDRI. If not, then the HDRI engine may be limited to a low resolution, perhaps to avoid the slowdown that would be caused by having to calculate the effects of 4.5 million point sources.