If you had a matrix and you combined all the anodes or cathodes into one resistor, you could make it work under the following optimum requirements:
-
The number of LEDs that are 'on' at any given time per resistor remains the same - whether it's one LED or 4, 5, 10, 40 or whatever. You have to switch them correctly in order to do this and it probably would be very complicated. Also, slight timing differences in your switching could lead to extreme transient spikes between switching times, which could damage LEDs and would get worse with more LEDs.
-
All LEDs connected to one resistor must have the same voltage drop rating. (Even then the brightness of each will vary due to manufacturing differences)
While this is all doable, it's really not very practical on a large scale. The risks of something going wrong are pretty high, especially with large groups of LEDs. If you're doing small groups and can live with differences in brightness, it can be done outside of optimum practices.
ive got 16 right now running off of 4 resistors in a matrix, and the visible difference is negligible
Can you post a schematic of you physics defying circuit?
This actually doesn't surprise me. As long as there is never a condition where the current in one LED or resistor is beyond spec, you're OK. For example, at 5V you can run 1-4 LEDs off of the same 220 ohm, 1/4w resistor. But as each LED turns on, the others dim slightly - maybe not enough to be a concern. In this case, when one LED is on, it's running at about 22mA, two LEDs -> 11mA each, three -> 7.3mA, and 4 -> 5.5mA each. Depending on the LEDs, the brightness may not vary enough to bother the designer.