Creating a Pile of Sugar


#23

Well Realflow is a three trick pony, Houdini is an effects suite and more. Honestly all the money I have spent on plugins, maintenance and 3rd party apps I could have bought an maintained a license of Houdini FX/Master, and at least in my mind I would be pretty proficient at it by now :smiley: BUT Houdini indie gets opened every once in a while, just to remind me of what I could have spent the last 20 years learning/using.

For all its many faults I still love using max (for the most part).


#24

I noticed Ephere Lucid got mentioned in this thread. I bought it a few years ago for a project that needed a light lava simulation, but at that time the plugin didn’t feel production ready. It crashed A LOT, and caching simulations to disk seemed impossible. On the other hand it was quite fast: my 1000 frame simulation with 1.3M particles took about 15 minutes to simulate, so I decided to skip caching altogether and ran the sim when needed. Haven’t followed the development, perhaps the plugin is more stable these days.


#25

I’m gonna derail the thread a bit, because the conversation has devolved into talking about how Max is lacking in a lot of simulation areas…

Obviously Houdini is still the goto for a complete and independent VFX package, but for those who plan to stick with Max, I’ve actually been working on a replacement for Particle Flow, massfx, etc, that should help bring Max back into the 21st century. It’s a complete rewrite of Particle Flow from scratch. Some things already implemented:

  • Every operator is fully multithreaded
  • Tons of new operators, featuring things like a granular solver, cloth, rope, proper path follow, DLA growth, constraints, etc
  • Spline operator and a super fast spline meshing modifier for converting particle trajectories/neighbors/constraints/etc into spline meshes
  • Up-to-date PhysX support
  • Voronoi fracture, brick fracture, boolean fracture, convex decomosition, etc, built-in
  • All of PFlow’s terrible O(n^2) operations that make it totally unusable for productions have been properly accelerated (mesh collisions, nearest neighbor searches, surface lock/bond, etc)
  • No more hidden PFlow operator/event nodes in the scene to manage, each flow is fully self-contained
  • Tons of extra control has been added to all the base operators, and each operator has timing activation controls
  • A C# script operator that executes nearly as fast as native c++ and that supports multithreaded scripts
  • VRay instance rendering support, GPU instancing in the viewport (easily display 100s of millions of polys in the view no problem)
  • Super fast auto-caching, similar to how Houdini caches things on the fly
  • PhoenixFD grid support, PRT import/export
  • Simulation retiming
  • A custom file caching system that is often hundreds/thousands times smaller in file size than alternative systems like xMesh

It’s basically what PFlow could have been, if development didn’t stop like 10 years ago.

I’ve been posting development tests occasionally here, and plan to release a public beta version once it gets a little more feature complete. Still working towards adding things like crowd simulation support, ragdolls, etc.

Some examples of cool stuff made with it here, here, here, and here.


#26

This is just amazing! :scream:


#27

Wow! probably the best derailing…in the world!


#28

Wow, freaking amazing mate!

When will this be available for public?


#29

Not sure the release date yet, still some features to add and bugs to fix…hopefully soon though! :slight_smile:


#30

We want it with BUGS! :wink:

Seriously man, I’ve been following your work more then 10 years (since that spiderfly animation), and I was kind of sad you switched to programming.
But I never expected that you will be creating something this big and cool.
Tests are incredible! Autodesk should hire you as their main guy for these stuff.

If anything deserves separate thread, this deserve it. I hope you will open one soon.
Looking forward for updates.

Cheers to all.


#31

Purely for selfish reasons, no to Autodesk hiring him because they’d make it exclusive for their subscription-customers.
If it is really as good as the Instagram-videos would lead us to believe, he can make more money and have more freedom not working for Autodesk.


#32

This looks amazing Tyson! Can’t wait to see this thing. I’ve seen your posts on instagram.

m


#33

I was more referring, how it would be in an ideal world. :wink:

i don’t care in what form this plugin (or better: bunch of plugins - heavy extension of 3DS Max), will came to us, I am only afraid that Tyson wont abandon it (for any reason).
For example, this guy:
https://vimeo.com/155278137
I really like his progress, but for a long time, he hasn’t posted any update.

Or this plugin, for example:
http://forums.cgsociety.org/showthread.php?t=1467403

That’s why I said, we want it with bugs. :slight_smile:


#34

Both of those have direct competition though if they haven’t already partnered or released them.

//youtu.be/EmNTrOX7kJg

and for the fluids there is Phoenix FD for commercial use and Blender for hobbyists. Also 3ds max now has fluids if you’re a subscriber.
Until released it makes no money and so making the commitment and finding the time to work on it after the day-job is difficult.
Also, I’d imagine that even for an experienced-programmer, this stuff isn’t easy and so just because you haven’t heard anything does not mean that work has stopped. Perceived progress is not necessarily linear with respect to the work put in.


#35

It devolves for a reason :smiley: and this derail… TAKE MY MONEY!


#36

ok…so next time some dev tells me…most PF ops simply CANNOT be multi-threaded because of their ‘nature’…I’ll just ignore him/her ?


#37

If I’m not mistaken, I recall orbaz once mentioning on his forum that he couldn’t multithread OPs in pflow due to the way his random number generator and/or particle seeding is calculated (most likely particles are all seeded from a single, shared PRNG) . I’m not sure how much work it would take to change that part of pflow specifically, but it would explain why all OPs – even simple ones – cannot be multithreaded. tyFlow was created with multithreading in mind from the start, so I could avoid accidentally building any pitfalls like that into its core.

Thanks for the feedback everyone! I’m happy to see people see potential value in tyFlow :slight_smile:


#38

A whole particle system crippled because all the emphasis is on frame 0!?


#39

Looks amazing, Ty! Really does. This will make particles so ridiculously much more useful in max…


#40

Tyson’s work does look amazing and look forward to it’s release. Just to give some options :

https://effectivetds.com/resources/fx-tools/storm/


#41

I see you uploaded a several new tests. That’'s looking amazing, especially that last one with character spawning the cloth lines.
Maybe it’s not best time and place to ask (you don’t need to answer anyway): but is there rough estimation of release time, and price?


#42

No ETA yet…but every day the todo list gets shorter :slight_smile: