I think I may take some of the pieces of our conversation here and be sure to work into the documentation/examples some discussion of frame rates/calculation, persistence of vision effects, etc… - @Mark_Kriegsman and I came up with a way to abuse POV effects and very high frame rates to cut power consumption of larger LED works drastically. E.g. 100 times the leds, only 10x increase in power draw from an existing project 
You may indeed be limited by the chip’s PWM rate. In fact, trying to do fast updates with 1804 based pixels (I think the PWM spec is only that it’s >= 400HZ), I was getting beat frequency effects with the PWM. It looks like you can only change the color or brightness every couple of msec on those chips. Other chips have different PWM frequencies (usually higher; the 1829 seems to be spec’d especially high), and some conceivably could have a different method of doing PWM that allows more frequent changes (i.e.: changing the on/off status within a PWM cycle if the register value changes).
I haven’t tested with the 2801 yet; I have some on order. But as a heads up, if you have a very high speed video camera, the PWM speed and method may also need to be taken into account, not just how fast you can load registers.
The interesting thing with the 1829 is not the high pwm speed (though honestly, give me a camera with a 1/8000 shutter and ill show you pwm).
It’s that in addition the 255 levels of pwm it also has 32 levels of constant current brightness. I have a set on order and while I don’t think they will make it here in time for the first release of FastSPI_LED2, it should be pretty close on its heels.
Ws2801s have other problems too, like having issues with > 1mbit data rates. I’ve been able to push lpd8806’s at over 22