simulation finished (WIP Yapping)

Video should release October!

Wanted to see how far I could push fluid simulations with this project. This was honestly more of a stress test to see what exactly was possible.

I think the result came out pretty good!

The first result was... Ok

Nut 2 & 1 comparison

They both look pretty similar, but you may notice there's way more particles in the 2.0 version. It is very surprising how consistant this software is...

Simulations honestly look alot better as particles. Meshing them properly seems to be the actual final boss...

The 1.0 version looks really good as particles, but meshing it and making it look good turned out to be a nightmare. Too little particles and too much Viscosity makes it clump up in an ugly looking way when it lands on the male (me btw). This was as high as I could get with the Viscosity on the 2.0 version without making all of these bumps everywhere. I still think I could have put it higher, but making the values perfect for every section was really challenging.

The first bake was too high, and the second I do feel like it is just a little too low. But I think I may be looking too narrow at the problem and didn't look more into other settings...

I still think it is behaving too much like water. I'll look more in future projects on how I can make it more sticky.

Getting the right settings for a normal animation for seems way easier than trying to make a perfect system in a 4000+ frame animation like I tried to lmao

Custom Normals

Also, I have no idea when this happened but exporting custom normals as an Alembic file and then importing them into Blender is no longer broken!

Nut Custom Normals

Custom Normals my beloved...

You can make bumpy geometry appear smooth by editing normals, without touching the geometry at all!

making things less lumpy with lower visocity may have been some wasted time, and we can definately go a little higher next time

I really learned alot about how impactful some settings are, and how far I can go with optimization. The first simulation took ~8 hours to bake.

I didn't like the results from the first one, so I tried again with some settings tweaks, and raising the quality, but after letting it sit for a while, I realized I was going to spend over a week simulating :)

Use Oversampling!

The real MVP was realizing how OP reseeding is. You can get so much more quality by raising the Surface Oversampling, with only like 1/8 the processing power required.

fine tuning the Death Threshold to be as low as you can get it also saved so much time for this project specifically. I might try turning it off for more reasonable simulations that aren't over 4000 frames long

The 2.0 result looks way more high quality, and it took almost the same exact amount of time to bake as the 1.0 version with these settings! :)

I think I could make it better, but I also think almost reaching 1TB of fluid simulations alone is probably a good place to stop :)

I'm going to turn into The Joker

It took way too long to be comfortable in Houdini to get to this point, but I think that's just the nature of software with only bad beginner tutorials available. No matter what I watched nobody was actually a capable teacher and instead all I learned was how to copy set ups and never why I was using these nodes at all. Half the documentation only repeats what the hover over text states in this mfing software where every slider you adjust takes literal minutes to figure out what it actually does every time :)

What i'm trying to say is I think it's not difficult to learn Houdini. You're just going to struggle if you treat it like Blender where you can fully master everything you'd ever want from a quick google search and be presented with plenty of results results no matter how obscure your question is. There's alot of things I would love to just watch a quick video to see if it's possible and worth my time investment. But... No results...

Revisit cloth simulations?

With the nut simulation being pretty close to where I want it, I would like to look into doing cloth simulations in Houdini. Not because I necessarily want to. I really like Marvelous Designer, but they keep adding AI slop updates in their software, and it's genuinely too expensive for me to justify with how infrequently I'll open the software. I'd rather not give them any money if this is where their development focus for paid software goes to. :)