More info here
This tutorial consists of a series of video tutorials that show how to get started using Blender fluids in Lightwave. The process starts with a scene that is exported from Layout as a Collada format file. A custom script is run on the Collada file to make it compatible with Blender. The Collada file is imported to Blender and the fluid simulation is setup and run. meshes are setup as fluid domains, obstacles, and particles. A custom Modeler script is used to import the fuid mesh and particle files as Lightwave object files. Next the original scene is loaded into layout and the mesh and particle sequence lwo files are loaded into the scene. The object sequencer is applied to the meshes, morphs are applied for motion blur and another custom script is used to assign a surface material to the fluid meshes. The same surface is manually recreated in the Hypervoxels assigned to the particle mesh.
right click the links below to download the videos
Video 1, 14.6MB, 5:34
Walk through the scene in Lightwave. Export Collada file.
Video 2, 39.5MB, 13:00
Import Collada file to Blender. Set up the fluids simulation.
Another option for particles is to have them generated directly into the fluids mesh. Under domain > fluid boundary set subdivision to at least 2. It's convenient but less flexible than generating the particles as purely particles. In one tutorial I saw they recommended a final domain resolution between 80 and 200. To get the original domain box back after a simulation just change the viewport display from preview to geometry. The fluid domain needs to be just big enough to contain the fluid and no bigger. Here the domain was little bit too small and the splash can be seen against the invisible walls of the domain.
Video 3, 78.8MB, 13:56
Fluid domain identity transform.; Look at the simulation result in Blender. Import fluid files to Modeler. Load sequence files into Layout.
Video 4, 63.1MB, 12:59
Hypervoxels. Create water shader and apply to the sequenced mesh and hypervoxels.
Video 5, 22.7MB, 9:02
Fix hypervoxels so invisible until needed. Motion blur. Render.
MentalFish Motion Baker - the motion baker that works
I suspect that the fluid simulator does not work well with sparse, thin fluids. The image from fram 91 shows the pixelated fluid just outside the frame of the render. The even rows of "pixelation" are even more noticeable when it's animated. It is hidden a little bit by the transparency of the fluid in this case, but the individual drops have a strange rolling motion. Higher resolution in the fluid sim might help. The best solution in this case may be to have a layer of fluid that covers the floor to give some depth and avoid the problem.
Moving fluid obstacles are treated as No Slip surfaces so the fluid is sticking to the ball like goo. One solution would be to scale the ball down after it enters the fluid to hide it.
The fluid is sticking to the ramp and has some bad render effects on it. I would fix it by making a render without the fluid mesh or hypervoxels and masking out the trouble spot using compositing techniques.
I had trouble getting the cuastics to render. They wouldn't work until I changed the light to a spotlight.
The fluid domain needs to be just big enough the encompass the fluid simulation. The domain was a little bit small here and it's invisible walls show in the render because of the splashing fluid.
The transparent shadows gave me some troubles. If an object is too transparent it's shadow does not show and when caustics are added the shadow has no transparency at all. Since i had caustics on I just faked it by setting the shadows to be a lighter color. The best solution would likely be some kind of render passes to combine the light from the caustic effect with a shadow pass.
Blender settings used in the fluids simulation:
The simulation took 4 hours 15 minutes to run on an Intel Core i7 1.6Ghz machine. 964 fluid files files were generated taking 883MB of disc space. 1201 lwo files were generated from the fluids files and took up 3.54GB of disc space. Not all of the files generated were used in the final render.
It took about 6 hours 15 minutes to render 240 frames. The particles were rendered seperately because I didn't know how to handle them at the time of the render, though the total render time would have been about the same.