I think a way of avoiding higher numbers
of particles would be a changable particle
size. Will this implemented or will it stay
a Point-only Renderer?
A billboard based material system would
be nice too … as you don’t rely on the
slow transparency rendering of Scanline, fake volumes could easy to achive …
But before I forget, thanks a lot Mark and Bob for this new tool!
Rob 
In most of our own uses of Krakatoa, we’ve wanted to achieve a more detailed volumetric look that moves genuinely as a fine detail particle system instead of have the feel of a coarse particle system with billboards. This is why there are features focused on cranking up the particle counts.
We do also intend to add larger particle sizes and billboard-style rendering, just probably not for the 1.0 release. All suggestions are welcome, and we’re prioritizing them for future releases!
-Mark
A Z-based particle size would also add to
the 3d-dimensionality of far stretched
particle systems …
Rob
Definitely, in particular for smaller particle counts. For high particle counts, all the particles blend together and you get a volumetric effect, but for smaller particle counts where you see individual particles, that’s an important factor.
-Mark
Another nice option would be vector colors
for RSMB … this would reduce the need
for the internal motion blur …
Best, Rob
I can’t imagine vector blur working. These particles are sub-pixel in size. Of course, if you wanted, you could just set the particle color to be based on the velocity.
If you render a higher resolution this might
work, but I wouldn’t know how to output the
velocity itself… Rob
What I mean is, if particle 1, in the foreground is moving left, and it 10% opaque, and particle 567345 is directly behind it in camera space, and is moving right, what vector do you put into the velocity channel?
Hmmm… Mark, would it be possible to save the “passes” used in the painters algorithm to individual files? The back to front ones. So for each frame we would get 6 or whatever files written out? Might be useless, hard to tell yet.
Oh, I see … never thought of that, sorry
for my naivety! BTW I’m looking forward to
particle works from Anatomical Travelogue
Best, Rob
That same issue exists with a z depth render idea. The ‘Blended Z Depth’ render mode is what we came up with as something which kind of works, and we could likely apply it to pixel velocities as well. Basically, blended z depth ends up being an averaged depth slightly pushed into the volume. In that mode (and in a potential ‘Blended Velocities’ mode), the post-divide by alpha flag has to be on in order to remove the alpha pre-multiplication that’s normally in the render.
-Mark
Could it be possible to render out the back to front passes separately? To aid compositing?