Whoa, we might have something else going on here. Sounds like the power to the LEDs may be a little starved. In a perfectly working system, all segments would be equally bright, no matter if one or all sixteen segments are lit. So I have a few questions:
1) are you using NPN transistors to drive the two cathodes (5,10) to ground? If so, what is the resistor value on the bases of those resistors?
2) are you using a power supply with an excess of available current? What is the maximum it can deliver, and are there sufficiently large electrolytic capacitors on it?
3) do you know if the raspberry is able to drive the current required by the LED segments, or are there any buffers/drivers for the outputs from the raspberry to the LEDs?
It would be helpful to see a schematic drawing of the circuit with part details, even if just drawn on a napkin.
My understanding of the raspberry is that the DIO pins are driven from 3.3 volts (not a lot), and is limited to 50mA for all outputs at most. (50/16=3mA each if all are on).
Furthermore, I think each output pin is limited to about 15mA. I could be wrong here, but if that's close to the truth then the raspberry isn't a very good current source.
If any of the above is the source of the varying brightness, we (humans) have the technology to correct that.