Particle Flow Discussion

[SoLiTuDe]

In fact it had two new features that fit exactly into some needs we had on two separate projects at work and we upgraded to it as soon as it was available. After a week of trying to make the new features work (realize that our facility has long used TP)…we had to give up on it for our projects. The features worked as shown in the feature demonstration videos, but completely lacked the flexibility to achieve the results we needed in production. TP is a very good tool that does some things really, really well, it is NOT however, the glorious, god given, be all end all particle tool that cebas would have us believe.

Mind if I ask which features? I haven’t used tp5 yet (I’ve since moved on to Houdini and love it for most things), but I’m really curious since I have a history of pflow/box3 and TP. …and you’re right, there is no glorious be all end all tool. They all have their pros and cons and I’m pretty sure at least 80% of the time it comes down to what the person is comfortable with or learned on in terms of preference. Never limiting yourself to one choice is always a good idea.

[charleyc]

The two features we were after were the rope physics and the car rig. The artists that was after the car rig had worked at Uncharted for 2012 and I guess they had set up a rig system in an older version, but it had some issues that v5 looked to solve (mainly some collision jitter issues). At first v5 made the whole thing a piece of cake, it seemed to work perfectly, then we realized that it could not be built upon as we needed. Adding the additional functionality required adding a second layer of physics and that reintroduced a whole new set of problems. We ended up just using a scripted car rig. As for the rope, I could not get it to properly create a system where I could animate one object (or particle), hang a rope off it and have the other end attached to a particle that acted as cargo on the end of the rope. Once I got the rope created, it would instantly crash if I set it up to animate one end (whether it be as a particle or scene node reference). For that I ended up using Box2. A 30 min setup gave me exactly what I was after. Even let me dynamically pick up the cargo. I was worried that the box2 approach would jitter, but it worked great.

Several of our fx artists only know TP and they use still use it regularly. But there are many instances where it is not the right tool for the job.

I am also migrating myself to Houdini…mainly because so far, it seems to give the power you need to work around nearly any issue as long as you know the way. I am hoping it is a much less limiting tool than what I have been using in the past. However, I really hope that Oleg takes max particles to a whole new level. Where I work now is completely a max facility and trying to integrate additional software brings its own challenges.

[SoLiTuDe]

The two features we were after were the rope physics and the car rig. The artists that was after the car rig had worked at Uncharted for 2012 and I guess they had set up a rig system in an older version, but it had some issues that v5 looked to solve (mainly some collision jitter issues). At first v5 made the whole thing a piece of cake, it seemed to work perfectly, then we realized that it could not be built upon as we needed. Adding the additional functionality required adding a second layer of physics and that reintroduced a whole new set of problems. We ended up just using a scripted car rig. As for the rope, I could not get it to properly create a system where I could animate one object (or particle), hang a rope off it and have the other end attached to a particle that acted as cargo on the end of the rope. Once I got the rope created, it would instantly crash if I set it up to animate one end (whether it be as a particle or scene node reference). For that I ended up using Box2. A 30 min setup gave me exactly what I was after. Even let me dynamically pick up the cargo. I was worried that the box2 approach would jitter, but it worked great.

Several of our fx artists only know TP and they use still use it regularly. But there are many instances where it is not the right tool for the job.

I am also migrating myself to Houdini…mainly because so far, it seems to give the power you need to work around nearly any issue as long as you know the way. I am hoping it is a much less limiting tool than what I have been using in the past. However, I really hope that Oleg takes max particles to a whole new level. Where I work now is completely a max facility and trying to integrate additional software brings its own challenges.

Thx for the info. I thought those were the ones too. I was also there at Uncharted with the old car rig. The new one doesn’t look very different, but I know it uses bullet instead of sc. It def is limited, but still has it’s use. The whole car destruction pipeline there was very complex, we only used the sc rig for a small portion of it anyway – mostly just to give us an initial animation to work that reacted to the buildings and roads being destroyed. All the other fancy stuff was done separately (as you probably already know) with their own rigs, just basing animation off the sc car (that was exported to geo). As for the ropes, yeah I figured the initial release would have bugs. They used Tp4 for the ropes on Final Destination 5’s bridge sequence. The setup is probably similar fashion to how you (i’m guessing) did it in box2, with lots of joints/constraints that create a chain of particles, also worked a charm and is a quick setup. I was able to make rope in tp4 that held up a chunk of bridge and could hand animate and release/explode the connections to the bridge at any time while doing some voronoi frags as well. It was a sweet setup. I left before I could do any work on the actual show though, so I’m sure the setups got way more involved.

Houdini is great, but IMHO tp is easier to use and is faster for heavy RBD work. Of course Houdini is way more open, and less limiting, but it often requires more set up time and figuring stuff out and such. The best thing is how well integrated all the Houdini stuff is with well… itself. lol There’s no worrying about whether X plugin will work with Y plugin, etc.

[charleyc]

Yeah…rdb. seem to be one of tps stronger points. The broad cross tool interaction is the biggest appeal for houdini to me. I hate getting what seems to be a solid solution and have it totally fall apart in one of the shots and have no easy way to fix it without recreating it all over again. The solution is usually easy to seebut the tool just wont hand it over.

Anyway…what fun would our job be without the challenges.

[JohnnyRandom]

Anyway…what fun would our job be without the challenges.

LOL it wouldn’t be

[Glacierise]

I did cars with manual jointing and particle based deform, and ropes with jointed particles/skinwrap with TP4 in a recent show, it all worked quite well

[J-Bond]

@charleyc
I can also try to find simple problem in pflow that i don’t like and made a decision that pflow is not right tool.
Lat say this: - To write a long expression I need to waste couple hours to build a tree in Box3 and minute to write it in Expression node in TP.

You just to much love box2,3. They not near to Ice and Houdini.It’s too much subjective thinking.

[JohnnyRandom]

@charleyc
I can also try to find simple problem in pflow that i don’t like and made a decision that pflow is not right tool.
Lat say this: - To write a long expression I need to waste couple hours to build a tree in Box3 and minute to write it in Expression node in TP.

I see your point BUT have you actually timed the evaluation? I may be wrong but I suspect that running that expression versus a data op, the expression is going to be slower. If you had time, I would be curious to see the results.

[charleyc]

You just to much love box2,3. They not near to Ice and Houdini.It’s too much subjective thinking.

LOL
That is why I am working on learning an entirely new application…I just love the max tools soo much…

I could write an essay on the problems with PFlow. I know them very well. But, for me, I prefer Box3 over TP in general because the straight forward math of Box3 comes very easy to me. It makes it a good fit for me. Our facility uses both and I have seen the times where TP falls short and where PFlow falls short.

But all in all, I am not happy with where these tools are at in max and I am anxious to see where Oleg and Autodesk take them from here. I am learning Houdini because I need more.

btw…writing expressions or using scripts in max particles is most often NOT a faster solution.

[J-Bond]

@JohnnyRandom
I make a scene with 4 helpers,10k particle per sec., 200 frames, 1 step/frame, near 83k particles at the end - so results on picture. I just run slider from 1 to 200 and take a timestamp.
Also i find that TP emit this count in 0,0,0(without any action) 13 sec compare 3 sec to emit in pflow.
So hard to say where is time loss to do a clean result. But anyway tp is slow in this case.

@charleyc

But all in all, I am not happy with where these tools are at in max and I am anxious to see where Oleg and Autodesk take them from here. I am learning Houdini because I need more.

Ye i’m also more and more sitting in houdini. For max not sure what they can add in future, not like this friendship with physx.

[charleyc]

PhysX might not be bad. As long as it is a unified system that works with itself and is stable…I don’t care what they use.

[JohnnyRandom]

@JohnnyRandom
I make a scene with 4 helpers,10k particle per sec., 200 frames, 1 step/frame, near 83k particles at the end - so results on picture. I just run slider from 1 to 200 and take a timestamp.
Also i find that TP emit this count in 0,0,0(without any action) 13 sec compare 3 sec to emit in pflow.
So hard to say where is time loss to do a clean result. But anyway tp is slow in this case.

Thanks for testing

I wasn’t sure, I suspected that Box#3 would be faster, of course maybe there is a slowdown in the way you formed your expression too, dunno. Have you tried the new math node in TP5 out of curiosity?

[J-Bond]

Thanks for testing

I wasn’t sure, I suspected that Box#3 would be faster, of course maybe there is a slowdown in the way you formed your expression too, dunno. Have you tried the new math node in TP5 out of curiosity?

AFAIK expressions is much faster then maxscript in calculations. It’s fun to check in 2013 where particle redraw is faster.
It would be cool to check tp5 but cebas not hurry up to provide demo.

[charleyc]

It depends on a lot of things, not the least of which is the developers implementation. I have seen Bobo create scripted operators in PFlow that evaluate and scale nearly as fast as Box3, but there are a lot of box3 functions that completely kill the system if done though script. In general, coded, pre-compiled tools are faster than scripts or expressions.

[J-Bond]

It depends on a lot of things, not the least of which is the developers implementation. I have seen Bobo create scripted operators in PFlow that evaluate and scale nearly as fast as Box3, but there are a lot of box3 functions that completely kill the system if done though script. In general, coded, pre-compiled tools are faster than scripts or expressions.

Sure but expressions are much faster then any scripted operator. Here i did the same test with scripted node. Look at the time it take and compare that i posted above.

[PsychoSilence]

“The recompiles of Box#2 and Box#3 are available for Max 2013 now.
Please keep in mind that in order to be able to use the recompiles, you need to install 3ds Max 2013 Product Update #3 (or higher) and the latest Subscription Extension.
Please send a request to Orbaz Support with your serial number(s) to get the links.
Thanks,
Oleg B.”

[coldside-digital]

Wait, so even if you have purchased Box2 and 3, you need to be on 3ds Max subscription to be able to use them in 2013? This won’t go down well…

[PiXeL_MoNKeY]

Wait, so even if you have purchased Box2 and 3, you need to be on 3ds Max subscription to be able to use them in 2013? This won’t go down well…

No, if you own them Oleg is allowed to release recompiles of the existing plugins. However, there will be no update/new features (that stuff will only be exposed to the Autodesk releases).

-Eric

EDIT: After readin Oleg’s Orbaz forum post I am not sure. If it is a requirement I am guessing that the Subscription requirement is needed for MassFX compatibility or read-only option. Why don’t you ask Oleg about the situation?

[OlegB]

Wait, so even if you have purchased Box2 and 3, you need to be on 3ds Max subscription to be able to use them in 2013? This won’t go down well…

Details: For Box#2, you just need 3ds Max 2013 Product Update #3 (or higher). For Box#3, you need the Product Update, and the Subscription Extension installed first before installing Box#3. This is due to a deeper level of integration of the Box#2/Box#3 release with 3ds Max 2013.

Now, the question - how many Box#3 users are not on the subscription?

Thanks,
Oleg B.

[PsychoSilence]

I thought of canceling actually and i know Charley has canceled already…
I use but every 4th release only and with subscription going up 45% or so it is cheaper just buying the release i will use in production as stand alone.