Based on this post from earlier today,
viewtopic.php?f=197&t=11025#p47996
we now have a documentation page (still Work In Progress) which explains how the various parameters affect the final result:
thinkboxsoftware.com/kc4d-repopulation
very good … thank you … it would be help 
thank you !
very awesome! thanks! I’ve been on holiday and haven’t been able to play much, but I just saw Percy Jackson Sea of Monsters… and Im itching to play with ksrc4d
FYI, I added another illustration of the Density Falloff Start value to the online documentation to visualize the effect and what the value means…
I’m rendering an X-Particles stream with particle repopulation and am getting odd results. Rather than all the particles moving in concert, as you’d expect, the result I’m getting looks like the ‘glow’ is moving through a field of static particles; you can see the exact same grain pattern from one frame the next and the next and so on.
The attached images were taken two frames apart - if you add them all to Photoshop in order and switch the layers on and off you can see the same bright pixels in exactly the same places. So either I’m doing something wrong, or Krakatoa is! (Happy if it’s me, just tell me how to put it right!)
First rule of Krakatoa: “If you are seeing individual particles, your per-particle Density value is too high, or your distribution density is too low!” 
It is doing exactly what it is supposed to do. The Repopulation creates a voxel grid and seeds new particles in it. These particles are deterministic. That means that the random distribution of the particles inside a given voxel is always the same, controlled by the random seed parameter. So if you have a low density distribution where you can see the new particles, the same voxel will have the same particle distribution. The Repopulation does NOT create particles that are moving with the original particles, but the original particles reveal a pattern of random particles in space which get the same values. Thus, if you had motion blur on, the particles would draw as if they were moving, but on the next frame they would NOT be where the velocities of the previous frame were pointing at (because this would require History-Dependent simulation, and that’s what X-Particles etc. are for to start with).
So the Repopulation feature was meant for the cases where you have loose particles and you want to created a volume where you do NOT see any individual particles, but a solid volumetric cloud with millions of particles merging into a continuous mass without gaps. At that point, the random pattern would not be obvious.
In other words, the Repopulation does not work the way you thought it would, and is used for cases where the resulting particles cannot be seen individually.
The Repopulation approach was based on the following idea where Frost and PRT Volume were used to produce the new distribution (see esp. the second part of the tutorial: thinkboxsoftware.com/krakato … and-frost/)
Of course, you could animate the random seed over time to produce a changing noise pattern, but it will look like random noise and will not keep any consistency from one frame to another. I have used this for foam for example where you would expect constant change in the pattern. See the video in the end of the above example.
For the case where a few particles have to move actual particles around in space, we have a dedicated product called Stoke, but it is for 3ds Max only at the moment. You might be able to do that with X-Particles though (I don’t know the product so I could be wrong). You can watch a short video about advecting particles with particles here:
youtube.com/watch?feature=pl … 2r3SIuqBk0
or see all Stoke webinar videos here:
thinkboxsoftware.com/stoke-m … ar-videos/
Hope this helps.
Ah right, understood. Thanks, that clears it up, thank you!