I came up with another solution using pcopen in a solver. It's a bit unpredictable, but I found the result pretty and interesting. There is a hip file if anyone is interested.

@Librarian Thank you!! This is really useful. God bless you!

Hey everyone! How would you create this shape in 3D? https://www.universetoday.com/wp-content/uploads/2020/03/cosmicvoid-1280x720.jpg https://scx1.b-cdn.net/csz/news/800a/2016/blackholesba.png I saw Kate's video series on it, but the result is far from this. Any idea?

According to my understanding only Guided Ocean what is really spectra driven. Wave tank only uses spectra to initialize the waves, but "after the initial frame, the simulation will begin to deviate from underlying ocean spectrum." At least I think that is REALLY it. I mean it's not a trivial thing

To me this made it clear: https://www.sidefx.com/docs/houdini/fluid/oceans.html

Here is a nice tutorial about donut modeling. He managed to achieve a pretty decent result without any simulation. https://odysee.com/@konstantinmagnus:5/houdini-tutorial-procedural-donut:8

Hello Michael! I'm very interested!

Do you want something like this? Also, you can simply paint a `density` attribute and scatter points onto the donut based on `density`, set rotation and pscale attribs, then copy the sprinkles to those points, without any simulation. POP_Sprinkle_Odforce2.hiplc

Ok, I think I finished this. If anyone is interested, the hip file is available below. I hope it could be instructive for beginners. I added leaves, they can have a general motion (breeze), and you can add another wind based on the tree motion (length of averaged velocity). I discovered that Bone Deform is way faster than Point Deform (it was realtime for this tree), but Point Deform is more accurate. My setup includes both, I use Bone Deform for previewing, you can just switch. The wire sim is realtime as well, and cooking everything with leaves and with little branches was around 3 fps on my shy laptop without any pre-caching. (I think it's not that bad.) Actually the leaves part could be better, for example right now leaves don't orient themselfs according to the direction of the wind... so some controlling features like this still would be nice to have. Well, feel free to improve it I would say, theoreatically this setup should work with any tree, bush, plant, etc. created by the Labs Tree tools. The hipfile includes some notes as well, I tried to explain why am I doing what I'm doing. Also, I started an attempt to create a non-simulated tree motion, what I saw in nirfse3D's videos about his Simple Tree Tools. Not really impressive right now... I have no idea how he did it, this approach is getting really slow with more branches Anyway. Best wishes! tree_16.hiplc

It could be... I think it worth a render with just a constant @density value.

Try this: ignore this set_density Attribute Wrangler and set a constant @density value like 0.01 and see what happens. I usually don't use this set_density Attribute Wrangler created by shelf tool because sometimes I experienced jitter from that as well. I don't really have any other idea.

Hi everyone, I'm using the Capture Region SOP. Is there a way to drive the capture region size, with like a pscale attribute? (unfortunately pscale doesn't work) I want it to be smaller as the branch gets smaller towards its end.

Nevermind, I fixed the previous unconnected pieces issue. At last the skeleton preparing and deforming setup is working fine with as many branches as you want. I still use the fuse method, finally i managed to figure it out the way (unfortunately I couldn't make it with constrains ). Now it's time for the leaves! Any idea about the leaves? What's the best way to use the Labs Leaf Generator with a deforming tree? tree_10.hiplc

Well, not long after I post it, I realized that the previous method in tree_5.hiplc is malfunctioning. I improved and simplified it and now it's almost okey! But I have difficulties with constraining the branches to each other. Right now I actually snap the root points of every branches to the closest point on the closest branch, and fuse them instead of constraining but sometimes the snapping flaws. Constrains would be better I guess, but I couldn't figure out how to do it. Way A is unreliable. Way B is almost good but with a high number of branches, it becomes unstable. Anybody could take a look and tell me what would be the bulletproof way? tree_9.hiplc

Recently I had to do some timewarped flip with whitewater sim. I can share with you the workflow I found out, but here the timewarp is happening in simulation, not in post. So it's may not what you want. (I use the custom hda only to get some timewarp related values, nothing fancy) timewarp_flip_ww.hiplc sipi_frame_retime.hda