Hi Guys,
I am trying to upres a PRT sequence using a PRT cloner set to repopulation and I am getting these blurry renders and I can’t understand why as the particle count is quite high 35mil . I have attached a screenshot with the issue.
thanks
second image.
The blurry renders are caused by the way Repopulation works. Basically, it creates a voxel grid around the original particles, dumps the data of the particles onto the grid, and then seeds new particles on the grid, assigning them the values taken from the source particles, and distributing the Density appropriately to produce a similar result.
The result is very similar (and was originally inspired by) the creation of a Frost blob mesh around the original source particles, then filling the mesh volume with a PRT Volume. The larger the radius, the larger the resulting mesh, and the fuzzier the result. Of course, repopulation does not use a mesh, but the principle is very similar. Here is the example using Frost and PRT Volume what inspired the feature:
thinkboxsoftware.com/news/20 … katoa.html
The Repopulation mode works best when used with fluid simulation source particles (Naiad, RealFlow etc.), because SPH simulations keep the particles at well-defined distances, and Repopulation would nicely fill the void between them.
Here is the KMY Repopulation tutorial (the feature was first added to the Maya version of Krakatoa, so we have more info there):
thinkboxsoftware.com/kmy-rep … particles/
And here is a similar discussion from the KC4D documentation that explains some of the details:
thinkboxsoftware.com/kc4d-repopulation/
So it is not just about how many particles you have, but also where these particles are. 35 million particles created with a relatively large radius will be blurry because they will be scattered loosely around the positions of the original particles. This is not a magic bullet solution to produce more particles out of few. The way it works defines the way it looks, and it is not a substitution for good old Partitioning which runs the same simulation rules again and again, producing the equivalent of simulating the actual final number at once minus the memory problems involved.
I hope this helps.
Thanks Bobo for that explanation. I usually go for the partitioning as it is fast. In this case I needed 7 million particles so I did 7 partitions of a 1 million but the result was not as good as doing 7 million straight from Pflow. That was why i was looking into the PRT cloner. I was getting clumping with the partition version and I think that was because there were no values to seed except position. I am only using wind and deflectors for this. What is a good workflow if you have such a pflow?