I didn't mean to suggest that one *ought* to use SMD LEDs for a cube; I merely took issue with the claim that they have narrower beam patterns. I still disagree with this. It is true that you can see THT LEDs from behind, but if you plot the actual brightness profile in the elevation (?) plane, I think they are still more focused than the wide-angle SMD type. For example, if you looked at the ratio of brightness, say, on axis and 70 degrees off axis, I think the SMD LEDs would still win.
I didn't say they have narrower beam patterns. I said they have a narrower viewing angle. And that's NOT the same thing! Why? Because a non-diffused LED is very nearly a point light source. In a water-clear THT LED (and many SMD ones), the body is a lens that focuses it in some way or another, depending on the design. But in a diffused THT LED, it's not a point source any more, it becomes a diffuse body, with little to no focused beam at all. The light output of a point source LED has a pattern that varies significantly by angle. In a diffused one, the angle is damned near irrelevant, and that's desirable. In an LED cube, you don't
want the light output to be directional. 70˚ off-axis is still looking at the front of the LED. But in this application, you want an LED that is visible from well beyond that. An SMD LED, even wide-angle, will have little to no light output at or beyond 90˚, and even before then, you start to get color shifting as the LED housing blocks your line of sight to one or more of the individual LED dice.
You're thinking flashlights, where brightness is the main criterion. But in this application, the ideal would be a fluorescent sphere emitting all its light equally in all directions.
And again, you want the large diffuse body to provide a place for the colors to blend. Without this, much of the blending doesn't happen until the light beam hits a surface like a wall. Totally opposite of what you want in an LED cube.