How to generate fluid uv mesh with 19.5 Sop flip fluiids?

[Keanu]

I start to learn the new sop fluid workflow, but there is no uv data on final output mesh. How can i keep uv from the beginer model to final fluid mesh?

 

 

fluid-test.hip

[Atom]

On fliptank1, add a custom attribute of a type vector named uv. Promote the vertex UVs to point before supplying them to the flipboundary1 node. This should cause them to flow through the flipsolver1. On fluidcompress1, make sure to add uv to the list of attributes to save inside the compressed volumes. Also, on the particlefluidsurface1 node, include uv in the transfer attributes list to recover it when the surface is built.

It seems like a lot of steps, but you're basically creating a volume named uv to hold the attribute during the simulation.

I have seen other approaches that use an attribute transfer at the end to recover the lost UVs using an attribute transfer. Creating extra volumes at the particle separation rate does increase memory usage.

Edited October 9, 2022 by Atom

[Keanu]

14 hours ago, Atom said:

On fliptank1, add a custom attribute of a type vector named uv. Promote the vertex UVs to point before supplying them to the flipboundary1 node. This should cause them to flow through the flipsolver1. On fluidcompress1, make sure to add uv to the list of attributes to save inside the compressed volumes. Also, on the particlefluidsurface1 node, include uv in the transfer attributes list to recover it when the surface is built.

It seems like a lot of steps, but you're basically creating a volume named uv to hold the attribute during the simulation.

I have seen other approaches that use an attribute transfer at the end to recover the lost UVs using an attribute transfer. Creating extra volumes at the particle separation rate does increase memory usage.

Thanks a lot! that really help me!

But i wonder why there are lots of stream when i render with redshift?

[Atom]

Those smeary edges are the result of the attribute promote. Toggle it on and off before the simulation, and you'll see that is the problem. That's why other methods are used sometimes. The rubber toy is a big offender because it's multiple parts with seams. A geometry with a single UV map on a surface might work better in this case. There are tutorials out there on how to recover UVs after a sim, rather than pass them through the sim.

[ryew]

Not sure how well it still applies for the SOP-level system, but Tim van Helsdingen had a technique that used vertex split to help alleviate some of the seam issues in classic FLIP sims that might still be helpful