PDA

View Full Version : Dynamics in Game-engine beats LW



Ztreem
03-10-2005, 04:35 PM
I just looked at some videos of the new Unreal game engine, and WOW the dynamics looks better and more stable than LW's, but in realtime.

Look at video 3 at this site!
http://www.1up.com/do/download?cId=3138759

NewTek PLEASE do upgrade the dynamics so it's faster and more stable.
When I say stable I don't mean that LW crashes I mean the objects that scatter around when using dynamics.

Vincenzo
03-10-2005, 07:13 PM
Dynamics in games is alot more limited. They allow certain dynamics with certain items, and since they are dealing with a specific situation, they can optimize just for that situation.

Vincenzo

Karmacop
03-10-2005, 07:24 PM
Game engines cheat a lot and don't do per poly collision.

WilliamVaughan
03-10-2005, 08:37 PM
Vincenzo and Karmacop have valid points. The Game Dynamics I have used work great for low poly game content but would choke on a characters cape on one of my characters. This is not to say LW's dynamics can't be improved....just be sure you are comparing apples to apples.

wacom
03-10-2005, 08:37 PM
Zstream does have a point though. It would be nice if we had the option to get either more accurate Dynamics, or go with something more like a game engine version. Sometimes you need a ner-simulation, and sometimes you just need a 300 foot monster destroying Tokyo in real time.

Havok is kind of like this...great for manythings...but for simulation? Still it would be fine for a lot of stuff.

Captain Obvious
03-10-2005, 09:34 PM
Something along the lines of Zspheres along with dynamics could work really really well. At least it'd be a heck of a lot faster than per-poly dynamics on complex models.

mav3rick
03-11-2005, 01:45 AM
proton great answer ! nice to see always optimistic answers like "This is not to say LW's dynamics can't be improved....just be sure you are comparing apples to apples."

Ztreem
03-11-2005, 02:06 AM
:D I was certain that I would get answer like this and you are all right.

Maybe game engines cheat and is not sutiable for real -life simulations, but then Lw isn't either.

Have you all looked at the video??
If I could get some lowpoly objects to react an behave like that in LW I would be more happy with that dynamicengine than the one we have today. (I talking about HardFX when I say dynamics)

Mylenium
03-11-2005, 02:49 AM
:D I was certain that I would get answer like this and you are all right.

Maybe game engines cheat and is not sutiable for real -life simulations, but then Lw isn't either.

Have you all looked at the video??
If I could get some lowpoly objects to react an behave like that in LW I would be more happy with that dynamicengine than the one we have today. (I talking about HardFX when I say dynamics)

I certainly agree. Not so much that I'm a gamer, but in comparison to other programs (or even Impact!) LW's dynamics are slow, clunky and awkward. It's strange to find that even something simple as dropping a view gemstones on a simple floor/ plane with proper rebound (yeah, sounds cheesy, but I had to do it) seems nearly impossible in LW. Did it in Maya and could adjust everything while playing back my scene in realtime... Try that in LW and feel the pain. And yeah, once more: There was/ is absolutely no need to implement hard bodies as a displacement effect/ custom object which makes it impossible to properly bake motion and export it easily to other tools.

Mylenium

Ztreem
03-11-2005, 03:21 AM
There was/ is absolutely no need to implement hard bodies as a displacement effect/ custom object which makes it impossible to properly bake motion and export it easily to other tools.

I agree! Imagine having every object move and rotate as normal opjects instead of the displacement thing, then we could alter and change the simulation in graph editor instead of using EditFX.
But I guess that this is the kind of software you get when you integrate 3rd party workarounds into the software. Please don't flame me for this. :cool:

PS. I'm not a gamer either, I just looked at the video's for fun and was amazed of the dynamics.

CB_3D
03-11-2005, 03:31 AM
impressive demo,no question about it.

Red_Oddity
03-11-2005, 04:19 AM
They use this dynamic engine : Download the Rocket demo and be amazed : http://www.ageia.com/novodex.html

And it really IS an amazing engine...and this thing isn't just doable in games...this dynamic engine is so incredibly powerfull (especially combined with the special CPU that is build for it) it makes me wonder why not ONE dynamic engine in professional software can't even come close to the calculation speeds shown here...

With the custom build cpu (for game video cards) this engine can push 30.000 rigid/hard body dynamic objects in real-time...

Para
03-11-2005, 04:45 AM
There's a lot of valid points in this thread already so I'm not going to say anything about that but I'll tell ya what I think of the evolution of dynamics.

I think that dynamics were done in wrong order: First the super-accurate systems were made (those that take anything from days to weeks to calculate) and only then someone kinda came up with the idea that this stuff could be done in realtime to some extent by adding certain kind of optimizations and approximations to the algorithms. Those two keywords were remembered only after people already had the product which wasn't a good thing since usually those two keywords play a big role in the early stages of finishing the product. Basically what for example Havok, meqon, ODE and Ageia now are is what the whole dynamics system should have been from the start.

Ztreem
03-11-2005, 04:54 AM
They use this dynamic engine : Download the Rocket demo and be amazed : http://www.ageia.com/novodex.html

I've tested it and WOW! Now it's not game-engine beats LW now it's game-engine beats the hell out of LW! :eek:

Lightwolf
03-11-2005, 05:29 AM
I think that dynamics were done in wrong order: First the super-accurate systems were made (those that take anything from days to weeks to calculate) and only then someone kinda came up with the idea that this stuff could be done in realtime to some extent by adding certain kind of optimizations and approximations to the algorithms.
Not really, if you consider where most of the research and algorithms come from. "dynamics" have been used in engineering for over a decade, from the simulation of buildings and materials under stress to virtuall crash tests (which still take a cluster to computer, and are nowhere close to realtime).
It is usual quite easy to go back when someone revolutionized the field and say: Oh, they should have done it like that from the beginning ;)

Cheers,
Mike

P.S. YOWZA !!!! Post #2000, time to open a bottle ;)

Ztreem
03-11-2005, 05:48 AM
from the simulation of buildings and materials under stress to virtuall crash tests (which still take a cluuster to computer, and are nowhere close to realtime).

The question is, what do you want LW to do?
Super accurate take a week to get some bricks to fall on the ground or Looking as it's accurate and takes a second? :D
It's like radiosity, do you want the accurate and slow variant or the looking good and fast one?

PS. Or do you want what we have now, looking bad and not accurate and takes forever to calaculate? :p

Lightwolf
03-11-2005, 06:04 AM
PS. Or do you want what we have now, looking bad and not accurate and takes forever to calaculate? :p
Oh, no question about it, I was just replying to Para's musing about the origins of dynamics system. And since my woman is an engineer who worked with finite elements solvers in the late eighties/early nineties, going so far as to prove the realiabilty of a solver by checking the solution by hand (i.e. calculating it by hand), I do know where that kind of software comes from. This is yet another area where two disciplines cross, and only now seem to start communicating.
Then again, PFX was a major milestone when it appeared on LW due to the speed.

Cheers,
Mike

mattclary
03-11-2005, 06:41 AM
On a related note, make sure to read this:

http://vbulletin.newtek.com/showthread.php?t=34445

Ztreem
03-11-2005, 06:54 AM
On a related note, make sure to read this:

http://vbulletin.newtek.com/showthread.php?t=34445


I've read it and I think it's very interesting and cool if LW supported more hardware stuff, like harware rendring for one example. But it's nice if we don't have to rely on hardware too make nice and fast things in LW. I know one thing don't have to exlude an other. :cool:

hrgiger
03-11-2005, 06:57 AM
P.S. YOWZA !!!! Post #2000, time to open a bottle ;)

Ah yes, the good ol days....

Those dynamics are impressive, game engine or not. I hope if Newtek is still hiring engineers, they consider a dynamics person.

Lightwolf
03-11-2005, 06:59 AM
Ah yes, the good ol days....
...ahhh, you just shuddup, go away and give a kid a bazooka, will'ya
:D :p :cool:
Anyhow, you registered a couple of days before me I think ;)
Cheers,
Mike

hrgiger
03-11-2005, 07:05 AM
I just played with that rocket demo and it is absolutely incredible. I mean, just amazing. Stunning. Aside from the amazing physical collisions and interactions, did you notice the interface speed? Wowza.

Chris S. (Fez)
03-11-2005, 07:40 AM
Wow! Yeah. That is impressive (****, especially those rag doll demos). Well, I am sure Newtek knows that they are outclassed in this case. Just gotta keep their head down and go at it.

Havok 3 is being released soon and will supposedly offer a "revolutionary" boost in real time dynamics. Rumor is Max folks will be getting these new dynamics directly in the viewport.

It would be nice to have something so interactive in Lightwave.

Ztreem
03-11-2005, 11:58 AM
Look at this demo of Cinema4D it's old but still... realtime dynamics.

http://maxon.net/pages/products/c4d/modules/dynamics/movie_capture_bowl.html

Please NewTek do something!!
If you could do a realtime dynamic engine for lightwave and proper collisions that don't go through each other, I know everybody would feel a little happier.

PS. I mean come on, in other softwares people can interact while using dynamics in LW we can if we don't move or touch anything while calculating hopfully watch was going on, if we're lucky.

Mylenium
03-11-2005, 12:31 PM
Look at this demo of Cinema4D it's old but still... realtime dynamics.

http://maxon.net/pages/products/c4d/modules/dynamics/movie_capture_bowl.html

Please NewTek do something!!
If you could do a realtime dynamic engine for lightwave and proper collisions that don't go through each other, I know everybody would feel a little happier.

PS. I mean come on, in other softwares people can interact while using dynamics in LW we can if we don't move or touch anything while calculating hopfully watch was going on, if we're lucky.

Mmh, bad example. Did you ever work with C4Ds dynamics? Don't let this particular movie fool you. Setting things up is quite tedious and simulation speed is not at all so great once you use more than bounding sphere/ bounding box collisions.... There were (back then)/ are? also serious problems if your hierarchy was anything more than just plain geometry and contained to many sub-groups.

Mylenium

Ztreem
03-11-2005, 01:15 PM
Mmh, bad example. Did you ever work with C4Ds dynamics? Don't let this particular movie fool you. Setting things up is quite tedious and simulation speed is not at all so great once you use more than bounding sphere/ bounding box collisions.... There were (back then)/ are? also serious problems if your hierarchy was anything more than just plain geometry and contained to many sub-groups.

OK, Maybe so. I havn't used the software, but was amazed when I saw this the first time, back then LW didn't have any hard body dynamics.
But still, LW don't do realtime calculations even with simple objects.
I hope will see a dynamic solver that works like the particles, you can alter and see the changes in realtime then maybe for the more advanced things you have to press the calculate button.

hrgiger
03-11-2005, 01:30 PM
Messiah has also had some real time dynamics since way back at version 1.5.

Dodgy
03-11-2005, 02:28 PM
Yup, I've been trying all the major packages dynamics to see how they compare, and while I don't know them as well as I know LW's, I'm sufficiently able as to make an impartial judgement. In fact I've gone in expecting some of them to be much better, but been disappointed and found all of them wanting in some aspect. I've tried C4d and seen that it can be enormously twitchy, even with simple scenes.
This has also been the case with maya. No, I haven't been playing with it's cloth (or paid for it!) just the soft body stuff which comes with the similarly priced complete.

Most of these packages suffer when trying to collide a hard body with a soft one (note, this is not the same as a soft one colliding with a hard body!) and only collide with the hard shape of a soft body, or aren't stopped at all. This is with any number of cunning uses of wrap deforms and other work arounds.

http://vbulletin.newtek.com/showthread.php?t=34489&highlight=dynamics

http://vbulletin.newtek.com/showthread.php?t=34128&highlight=dynamics
have a couple of fairly simple examples I've tried to replicate with other packages, thinking it would be straightforward (given their rep), if not downright easy.

As I say, I expected some other packages to be superior, having been built on more recent architecture, but that fact that I've failed to achieve results as good, or even get decent with 'quicker/easier' workflow says something I think.

I have seen some impressive demos of cutting edge dynamics though, so I don't think LW's dynamics should stand still, as I have no doubt other packages will be looking at those new methods.

But I have been won over to thinking that LW does compete with other packages, if not beat them in certain areas.

Ztreem
03-11-2005, 04:33 PM
But I have been won over to thinking that LW does compete with other packages, if not beat them in certain areas.

Then we must hope that all the 3D apps gets better dynamics. Because what we have today isn't good enough. I still talking about Hardbody dynamics, try doing a brickwall and crash it with dynamics and make it look good without any brick falling through an other brick. Even with loads of workarounds it's almoust impossible.
Maybe this game engine dynamics isn't that bad after all, even if it's only cheating as someone said earlier in this thread. Because of what I have seen when testing it, I think it does a far better job than LW does.

Dodgy
03-11-2005, 05:22 PM
I absolutely agree. We had a guy working here who was really into this stuff , and he knew all the latest techniques, and had tens if not hundreds of thousands of polys in a cloth simulation running in virtually real time, as well as an example of hundreds of complex skulls piling up in a shallow box. Hopefully the LW bods are keeping abreast of all these developements and working up the next generation of dynamics as we speak... :)

Verlon
03-11-2005, 05:31 PM
Maybe I missed something somewhere.....

does Mylenium=Cim?

The guys at Newtek did release LW8 a year ago and an update a few weeks ago. Lets not get too negative about the stuff they haven't added yet.

They are hiring people (see press releases), and its not like Newtek is paying its programmers to sit around and answer tech support calls. They ARE working on something.

Dynamics now are certainly better than they were in 7, right? So at least they are being improved.

Simulating the real world is difficult, and a LOT more difficult than a game engine. Did that rocket have hair/fur? What about a cloth cape? And of COURSE they don't demo the weaknesses of their engine. You're comparing LW dynamics on its worst day to that game's best day doing what it does best.

Now if you had to work with that environment as your renderer and physics engine every day, I bet you could find all kinds of faults with it. I bet it doesn't have edge tools (not even fake ones). Metaballs? What are the texture limits? Think it would be ok for cinema? Me either.

I bet LW dynamics would be a lot better if we took out some of that 'fluff' that game engines aren't doing.

Mylenium
03-12-2005, 05:25 AM
Maybe I missed something somewhere.....

does Mylenium=Cim?



Nope, definitely not and had you taken the time for some more research before making such a self-embarassing statement, I might even have had some respect for you, but now that seems a bit more difficult from my side of the street. Indeed I have days where I'm nice and helping people (CIM never had/has those) and unlike CIM I'm also not making any claims as to Maya being the holy grail of CG. It's just that I have a very critical view on anything out there in this world and I tend to be a bit cynical. But rest assured, it's not just LW, I find enough bad things in other progs too. My personal feeling only is that LW kinda betrays it's userbase. I've been using it for 10 years and always defended it, but now I feel that what once was a nice and cute baby has grown into an uncontrollable, untamable and rude adolescent. But hey, that at least leaves room for some hope once LW becomes a grown-up.

Mylenium

P.S.: Not specifically related to dynamics, but other great (former) LW-artists such as Christophe Desse, Policarpo, Taron etc. have made similar stements about LW's development, so that proves I'm not the only one and there may be even some truth in my supposedly negative views. so give me some credit here.

Ztreem
03-12-2005, 05:41 AM
Hopefully the LW bods are keeping abreast of all these developements and working up the next generation of dynamics as we speak...

Yeah, I really hope so! :D


Did that rocket have hair/fur? What about a cloth cape? And of COURSE they don't demo the weaknesses of their engine.

I don't know how many times I have to say that I refer to Hard Body Dynamics NOT fur/hair or Cloth, thats an other story.


You're comparing LW dynamics on its worst day to that game's best day doing what it does best.

Ok! So what are you doing when you comparing no Hard body dynamics in LW 7 to HardFX in LW 8???


Dynamics now are certainly better than they were in 7, right? So at least they are being improved.

Just so you now ClothFX in 8 is in the whole the same thing as motiondesigner in 7.
WOW! Great improvement!
(the only thing changed is the GUI and the workflow and it's great, but the calculations is still not that good. Look at all threads about how hard or impossible it is to simulate cloth on a character.)

mrunion
03-12-2005, 06:49 AM
My $0.02 (and maybe a few cents more?)

The game engine has to approximate things. For example, in Half-Life 2 you can't apply full ragdoll physics to everything. There is a limit to what the engine can handle ans still perform. Now, in a 3D app, there is no limits except time -- the physics don't have to be real-time.

Another thing to remember -- if it looks real, it ID real to the viewer. I dare say that not too many of us has ever seen LA blow up for real have we?!? How many have actually SEEN a nuclear explosion except on film?

My point? No one will probably EVER simulate ALL the forces in a REAL physical simulation. We have to approximate virtually everything (if not EVERYTHING). Making a good game physics engine is KNOWING WHERE to approximate. LightWave and all the other apps just need to keep practicing -- see what works, see what doesn't. And always remember that time is the tradeoff.

Ztreem
03-12-2005, 07:01 AM
-- the physics don't have to be real-time.

I agree with that, but maybe a little faster than today. :D


-- if it looks real, it ID real to the viewer.

I Agree with that too. But when things fly through each other when in the real world they wouldn't, that's when the illusion of realism go away.

I don't say that LW must do dynamics in realtime ( but it would be great), I only want the dynamics to look real. Things should come to a rest and don't scatter around when things fall to the ground and collide and not go through each other.

hrgiger
03-12-2005, 07:05 AM
does Mylenium=Cim?



Mylenium and I have butted heads a few times but he is definately no Cim. I had no respect for Cim, he never had any basis for any of his statements and in general was just a complete whiner.

And as far as the dynamics go in LW8 vs LW7, Ztreem pretty much said it. The only thing that changed is it was integrated into the properties panel and we now have a few new features (such as editFX and cut and sew, etc....) but the calculations are about the same and collisions are as poor as ever. It's like MB 2.5 if you will.... Not to say that there aren't some nice things that you can't do with it, but I wouldn't say there were big advancements in the general area of physics and dynamics in Lightwave 8.

Ztreem
03-12-2005, 07:09 AM
The game engine has to approximate things. For example, in Half-Life 2 you can't apply full ragdoll physics to everything. There is a limit to what the engine can handle ans still perform. Now, in a 3D app, there is no limits except time -- the physics don't have to be real-time.

That's the whole thing about this thread, Game engine's just have to cheat and approximate things. Still they make a better illusion of realism than LW.

Lightwave don't have to do it in real time and have all time in the world to do a believable simulation still it doesn't do it. :mad:

hrgiger
03-12-2005, 07:20 AM
My point? No one will probably EVER simulate ALL the forces in a REAL physical simulation. We have to approximate virtually everything (if not EVERYTHING). Making a good game physics engine is KNOWING WHERE to approximate. LightWave and all the other apps just need to keep practicing -- see what works, see what doesn't. And always remember that time is the tradeoff.

In general, I would have to disagree. Of course, we have to approximate, but the math going into physics simulations are based on real world properties. Just like we can predict when a comet will return by calcuating it's speed and trajectory. Or a more real world example, how we can build an aquarium to be strong enough to hold a certain amount of water based on the pressure that a liquid exerts both static and in motion. Objects have measurable properties that will all behave the same way under the same conditions. We are really just limited by our technology and the programming that defines these real world objets. Eventually, simulations in 3D programs are going to be pretty real and almost indistinguishable from reality. Maybe it will be 20 years down the road, maybe only a couple. But it will happen.

Do you mean a nuclear explosion from Hollywood film or a real one? Because only only has to see the film on Hiroshima or Nagasaki (or various test sites) to see a real nuclear explosion in action.

Emmanuel
03-12-2005, 07:41 AM
Mylenium and I have butted heads a few times but he is definately no Cim. I had no respect for Cim, he never had any basis for any of his statements and in general was just a complete whiner.


Oh, and he did write some useful plugins, too. :cool:

Mylenium
03-12-2005, 11:28 AM
Oh, and he did write some useful plugins, too. :cool:

Yeah, strange case of split personality, I guess. But let's not gossip about friends long gone. We all have our personality quirks and have to live with them (as well as the people around us).

Mylenium

Ernest
03-12-2005, 05:00 PM
My main problem with LW's dynamics is that the jump_dress scene, from the content, is exactly what I need, but it only comes with an mdd and I can't for the life of me, reproduce it. Lynx3d asked since about six months ago and I've been checking that thread every day ever since to see if anyone posts the settings but I still haven't found it.

Dodgy
03-12-2005, 05:38 PM
Yeah, it is very annoying that most of the dynamics scenes came recorded, without the original settings :P

Proton went some way to rectifying that with some new dynamics scenes, but it would have been much handier to have the original pre-baked scenes.

Celshader
03-12-2005, 05:46 PM
My main problem with LW's dynamics is that the jump_dress scene, from the content, is exactly what I need, but it only comes with an mdd and I can't for the life of me, reproduce it. Lynx3d asked since about six months ago and I've been checking that thread every day ever since to see if anyone posts the settings but I still haven't found it.

Out of curiousity, I opened up the dress scene in Wordpad. It has the same problem as all the dynamics stuff I donated (http://www.celshader.com/gallery/md/) to the LW[8] content directory -- the dress was actually calculated in Motion Designer, but some soul with well-meaning intentions played back the MDD file with the newer ClothFX. All of the original MD settings are still present in the scene file. However, if Motion Designer senses that ClothFX is somehow connected to an object, Motion Designer backs off and won't display the object in its interface.

You can see the original MD settings for the dress in the "jump_dress.lws" scene file if you look at the scene file in WordPad. It'll require a little guesswork, but it can give you a start:


LoadObjectLayer 1 Objects/Dynamics/skirt5.lwo
// MDS0:name="skirt" weight=90 spring=1000 viscosity=4 res
// MDS1:name="fix" weight=60 fixed smooth=2 sside skint=0.
// MDT:name="skirt" target cobj="mocap_chic_collision" md
// MDS1:03 friction=4
// MDS1:
// MDS0:istance=5 subsp=1000 holdst=100 smooth=3 non_stl=0
// MDS0:.14 cstress=0.14 shrink=0.7 self-c sside skint=0.0
// MDS0:3 friction=0.5

You could also remove/deactivate ClothFX from skirt5, open up the old Motion Designer interface, activate skirt5 in MD's Objects tab and then look at that object's original settings through the Motion Designer interface, especially in MD's "Surface" tab. Write those settings down and try using them as a starting point in MD's successor, ClothFX. Use the same technique to look at the other example LW[8] scene files with default ClothFX settings that play back MDD files.

Having used ClothFX, I think that it is a vast improvement from the original Motion Designer, and whoever swapped out MD for ClothFX when prepping the content CD must have had the best possible intentions. However, this makes looking at the original settings a lot trickier.

EDIT -- I forgot to mention that the old Motion Designer associated its global settings with the Camera. Here's the global settings for that "jump_dress.lws" scene, including the Environment settings:


// MDF:t_step=1 res=10
// MDE:grav=0,-8.9,0

Celshader
03-12-2005, 06:19 PM
Things should come to a rest and don't scatter around when things fall to the ground...

Try playing with "Fix by Event" in ClothFX/SoftFX, or "Stop by Event" in HardFX. I just started using those settings myself, but they give interesting results.

Dodgy
03-12-2005, 06:59 PM
Aha! Very cunning... :)

Though I tend to get the arm brushing at the skirt more in Cloth FX than in MD... Does this mean cloth FX's collision is better I wonder?

brade_andersen
03-12-2005, 07:29 PM
has anyone been following the dev of ageia's ppu's? sorry if this has already been discussed. there was a thread started over at nvnews.net (im there alot) about it. sounds very very exciting if you ask me.

it does for real-time physics what gpu's did fro real-time rendering.

anyways, here's a few links

Gamers Depot (http://www.gamers-depot.com/interviews/agiea/001.htm)

The Inquirer (http://www.theinquirer.net/?article=21648)

hrgiger
03-12-2005, 08:08 PM
I'm very interested in following the ppu. I sure would like to see it implemented in City of Heros (my game of choice at the moment).

Celshader
03-12-2005, 10:52 PM
Aha! Very cunning... :)

Though I tend to get the arm brushing at the skirt more in Cloth FX than in MD... Does this mean cloth FX's collision is better I wonder?

ClothFX is a lot smarter than MD about a lot of things. Chop the arms off the collision object if you don't want ClothFX to see them. That's what I'd do. :D

I took a quick stab at translating the MD settings into ClothFX settings. I used the MD settings as a starting point for the ClothFX settings, chopped the upper body off of the collision object, created a few weight maps and bandsawed the skirt in the front to add more points to the front of the skirt. I then fooled around with a few settings. The raw calculation looks like this:
http://www.celshader.com/gallery/md/jump_dress-Jen-sv3.mov

The girl's wearing the ClothFX-controlled skirt; a copy of the skirt that plays back the original MD calculation is running next to her.

The left knee intersects the skirt at the end, but I could fix that easily with EditFX.

If anyone's interested, my quick-and-dirty changes to jump_dress.lws are included in the zip file attached to this post. You will need the LightWave[8] Content CD to load the objects not included in this zip file ("skirt5" and Proton's "mocap_chic" character model).

Dodgy
03-13-2005, 03:03 AM
Funnily enough that was the first thing I did when I tried to get it working :) It worked, but I felt that was kind of cheating given that the scene was set up as it was and looked like it should have worked without modification when some else had got it working. Didn't realise they'd got it working with MD though :)

I'm just refreshed to see someone else getting half decent results with the LW dynamics and not thinking it's dodgy :)

Tesselator
03-13-2005, 04:12 AM
I'm just refreshed to see someone else getting half decent results with the LW dynamics and not thinking it's dodgy :)


I'm digging ClothFX too... I never felt it was dodgy at all.
It's a huge improvement over what we had before in terms
of speed, accuracy of calculation, ease of use, and tool
functionality (options!).

So far it's always done what I wanted it to do and done
it well. The most elaborite example being a fight scene
between 3 mocapped characters all wearring matrix
style trench coats.

I like anyone actually using it, do of course see room for
improvement though. It ain't perfect.. but it's miles
from where we were!

:cool:

hrgiger
03-13-2005, 05:57 AM
[indent]


So far it's always done what I wanted it to do and done
it well. The most elaborite example being a fight scene
between 3 mocapped characters all wearring matrix
style trench coats.




Yeah. I'll believe that one when I see it.

Ztreem
03-13-2005, 06:37 AM
Try playing with "Fix by Event" in ClothFX/SoftFX, or "Stop by Event" in HardFX. I just started using those settings myself, but they give interesting results.

Yeah, I know. I have played quite alot with the dynamics and I know you can get decent results if you play around a bit.
But I see it as a workarounds, I don't think it's good when you have to use a thing like 'stop by event' and animate a event object to make objects to stop.
Maybe if you have one or two objects it's works fine, but if you have 100 or 1000 or even 10000 objects, how do you think it works then???

Dodgy
03-13-2005, 09:58 AM
Yeah. I'll believe that one when I see it.


I seem to remember the Matrix didn't exactly have the best cloth dynamics in the world. Looked a bit rubbish to me.... And what was their budget?

While I would like LW's dynamics to constantly improve, I don't believe that tone is helpful.

Dodgy
03-13-2005, 10:02 AM
Yeah, I know. I have played quite alot with the dynamics and I know you can get decent results if you play around a bit.
But I see it as a workarounds, I don't think it's good when you have to use a thing like 'stop by event' and animate a event object to make objects to stop.
Maybe if you have one or two objects it's works fine, but if you have 100 or 1000 or even 10000 objects, how do you think it works then???

Use stop by stabilizer? This is the way other packages do it, even if they don't make it as explicit. I would like to see a separate control value for this function, as at the moment it's dependent upon the resolution, which also determines the collision values. So if you try to get more accurate collisions, you get more vibrations, so it's a trade off.

hrgiger
03-13-2005, 10:17 AM
While I would like LW's dynamics to constantly improve, I don't believe that tone is helpful.

It's not a matter of being helpful or not. It's a matter of stating the current implementation of Lightwave's dynamics.

As I've stated a few hundred times now, I was very glad to see improvements in LW8 in this area. I like how it was integrated into properties and I like some of the new features. You can achieve quite a lot of nice effects with it. But what Tess is describing, as I said, I'll see it when I believe it. Now if he is talking about controlling the upper body and sleeves with bones and just the hanging part of the coat with dynamics, I could buy it. But clothing that is adhereing to a characters limbs and torso during a fight scene, not so much. The collsions are just not that dependable. I tried for weeks to get a trenchcoat to work on a character and could not achieve desirable results. That doesn't mean that someone else couldn't do it but again, I'll believe it when I see it. And believe me, I would love to think it were possible.

Dodgy
03-13-2005, 10:25 AM
I apologise, I maybe read a bit much into the way you phrased it.

I would agree I would never expect LW's cloth to handle a close fitting coat, but then they only package that I've seen do that kind of close body dynamics (and only in demo videos, which is a different thing to doing it yourself) is sysflex, and then you're paying an extra $2200 for that ability. As I said, the matrix did it badly, and they had the resources to lay out for that. While I would love to have syflex's power in LW, I won't hold issue with having to work around and not pay that extra.

Incredibles also had the nicest cloth stuff, but then they cheated like baskets.. sorry I meant 'Worked around' and I see nothing wrong with that.

If I was going to do a trench coat, I would have done exactly as you described by the way :)

Ztreem
03-13-2005, 11:04 AM
So if you try to get more accurate collisions, you get more vibrations, so it's a trade off.

OK, but I want both. No vibrations and accurate collisions and that seems impossible for now. I'm not against workarounds, I use them all the time, but it would be nice if it for once could work without them. I feel like LW is becoming one big workaround and that's not what I want.

Celshader
03-13-2005, 11:48 AM
But clothing that is adhereing to a characters limbs and torso during a fight scene, not so much. The collsions are just not that dependable. I tried for weeks to get a trenchcoat to work on a character and could not achieve desirable results. That doesn't mean that someone else couldn't do it but again, I'll believe it when I see it. And believe me, I would love to think it were possible.

Samus came up with a killer technique (http://cohen-plugs.tripod.com/outdated.htm#cloth) for pinning down closely-fitting clothing with ClothFX. Yet...I wonder sometimes if I'm doing everything wrong with close-fitting clothing when I read this ancient inscription on the DStorm website:


From http://www.dstorm.co.jp/FXMnD/e/
FX_DISTORTION -- FX Distortion is a plug-in that can set Soft Body Simulation settings easier than FX MotionDrive. With a special function called "bump," you can let the polygons expansion and contraction to make wrinkles. This is useful to create wrinkles with the movement of the clothes or pants. [emphasis mine -- Jen.] Like FX MotionDrive, you can use EditFX to refer to the real time changes of the parameters.

"FX_Distortion" and "FX_MotionDrive" are SoftFX and ClothFX, respectively. Note that their orginal names do not limit their functions, while "ClothFX" makes the tool sound like it's the only one that can handle cloth.

One of the oldest "FX_Distortion" tutorials tackles pants. Wrinkles appear on the area marked out by a weight map:
http://www.dstorm.co.jp/FXMnD/fxdstdocs/index.htm

---

Again, sometimes I wonder if I do things in LightWave all wrong, after seeing the programmer's intentions for a given tool. Glancing through these ancient docs, I never before realized that ClothFX could receive collisions from a HyperVoxel emitter. Yet, look at this:
http://www.dstorm.co.jp/FXMnD/fxmddocs/index.htm
http://www.dstorm.co.jp/FXMnD/fxmddocs/index.htm

What other stuff am I missing?!

Celshader
03-13-2005, 12:06 PM
OK, but I want both. No vibrations and accurate collisions and that seems impossible for now. I'm not against workarounds, I use them all the time, but it would be nice if it for once could work without them. I feel like LW is becoming one big workaround and that's not what I want.

"Stop by Stabilizer" might deliver both, according to the DStorm website. I can barely read these docs, though:


From http://www.dstorm.co.jp/FXBREAK/e/11up.html

stopByStabilizer
This function was made to reduce 1/10 of resolution for there velocity, and you can stop the position when you set consecutive collison. You can restart the motion with startByCollision or RestartByCollision

When you make a scene of piled objects, you can slightly fix their move with this option.

sample scene >>> Scene/FXBreak/11/fxbreak11stopby0.lws
sample scene >>> Scene/FXBreak/11/fxbreak11stopby1.lws


I wish I knew where to get these sample scenes...but the above description makes it sound like "stopByStabilizer" reduces calculation-resolution based on...something. Velocity? What?

Red_Oddity
03-13-2005, 04:07 PM
I apologise, I maybe read a bit much into the way you phrased it.

I would agree I would never expect LW's cloth to handle a close fitting coat, but then they only package that I've seen do that kind of close body dynamics (and only in demo videos, which is a different thing to doing it yourself) is sysflex, and then you're paying an extra $2200 for that ability. As I said, the matrix did it badly, and they had the resources to lay out for that.

My gripe with the Burley Brawl wasn't so much the cloth dynamics but more the overal shading...somehow you could SEE when you where looking at the real coat and the fake one, which really is odd when you think about it as they used some really expensive measured BRDF techniques and equipment.

Dodgy
03-13-2005, 04:50 PM
It would be better in close body simulation to just use SoftFX to add secondary motion to morphs I think. It's kind of what they did with the incredibles. They did all they simulations on one scene with the character bending and moving various bits, then used the data generated to drive the animation in the actual animated scenes. That way the animators got near realtime feedback and they could CLEAN UP the simulations before even using it. So even their system isn't perfect.

I was disappointed by the matrix because they got it so right in the first, using actors and faking the rest. When Spidey came out with it's badly animated character, I thought to myself now the matrix two is going to show them how to do real people doing superhuman things, and then the first trailer immediately showed me they were making the same mistake. Oh well.

Karmacop
03-13-2005, 07:36 PM
Have you heard that bit on the spiderman dvd? Someone said that they showed a clip of the cg spider-man walkign up the wall and told the producers it was Toby in the suit. Then when the producers didn't say it was computer they knew it looked good enough. I laughed at that because that spider-man looked very fake.

The matrix was disappointing too. That Burley Brawl looked like a computer game :( Although the fight in the rain was awsome. The bit with the slow mo punch and you see the fist breaking the rain drops and then leaving an impression in the agent's face. That was very well done.

Verlon
03-13-2005, 11:26 PM
My apologies Mylenium. If there are others that will vouch for you, clearly you are not CIM.

I only recall maybe a dozen or so posts by you and they were all negative, so I jumped to a conclusion rather than looking up the rest of them. Of course, who has time to research every negative poster on the internet? You'll just have to forgive me that one.

My point is still that you are looking at this thing on its best day, not the time that it crashed windows, or how it is incompatible with WinXP SP2, or that for some reason red objects can't have more mass than green ones or whatever bugs ail that piece of software.

To be fair, you have to apply the same level of criticism to both pieces of software, or your expectations become unreasonable.

X-MAN
03-14-2005, 05:00 PM
I'm following the development of the AGEIA Novodex Physics SDK since several months now. I had the pleasure to meet some core developers of the AGEIA team.
When I saw it first it was grate fun playing around with the demo’s included in the SDK. The speed and robustness just blew me away.

Most people here think the speed of Novodex is paid by the expense of accuracy or robustness.
But after some tests I did with LW and Novodex, I can say the Novodex dynamics are at least as accurate and robust as LW's are. And also very complex meshes are no problem.

e.g. If u try to smash down a wall consisting of 10 x 10 boxes within LW u first need a long (I mean really long) calculation time (in my test it took me more than 8 minutes to calculate a single frame) and if u where patient enough u get an unusable result where boxes shoot around. So u have to go back, adjust some settings and wait an other day to see if the result fits ur expectations (I bet it wont).

So even if LW's dynamics are more accurate (I'm really in doubt about that) it's useless because the workflow is the exact opposite of interactive.

Newtek should buy a license by AGEIA and implement it into LW. Why spend years to invent the wheel twice? And as soon as gamer graphics cards with an PPU reaches the market, the days where we had to wait for the physics have gone.

Celshader
03-14-2005, 05:13 PM
e.g. If u try to smash down a wall consisting of 10 x 10 boxes within LW u first need a long (I mean really long) calculation time (in my test it took me more than 8 minutes to calculate a single frame) and if u where patient enough u get an unusable result where boxes shoot around. So u have to go back, adjust some settings and wait an other day to see if the result fits ur expectations (I bet it wont).


:eek: HardFX has never been that slow for me. Do you have an example scene or screenshot of these boxes? Are they separate objects or one object with self-collision?! How many points-per-box?

Ztreem
03-15-2005, 01:42 AM
HardFX has never been that slow for me. Do you have an example scene or screenshot of these boxes? Are they separate objects or one object with self-collision?! How many points-per-box?

I know what X-man is talking about. Even if the objects only is boxes with one ploy/side and self collision is set to box it's slow like that. ( maybe not 8min. per frame but at least 1 min) But the time doesn't matter here, it's the result. In LW you can wait hours for calculating the dynamics and it looks like crap and all the bricks fall unrealistic and through each other.

Cellshader: have you played with the AGEIA Novodex Physics engine???
If not I recommend you to download the demo and test it, then say that LW is capable of nice realistic hard body dynamics when dealing with more then 5 objects .

Celshader
03-15-2005, 04:27 AM
I know what X-man is talking about. Even if the objects only is boxes with one ploy/side and self collision is set to box it's slow like that. ( maybe not 8min. per frame but at least 1 min) But the time doesn't matter here, it's the result. In LW you can wait hours for calculating the dynamics and it looks like crap and all the bricks fall unrealistic and through each other.

Cellshader: have you played with the AGEIA Novodex Physics engine???
If not I recommend you to download the demo and test it, then say that LW is capable of nice realistic hard body dynamics when dealing with more then 5 objects .

Not yet.

For what it's worth, the following took less than three minutes to calculate on an Opteron 246:
http://www.celshader.com/gallery/md/boxPuddle2.mov

One thousand boxes collapse into a puddle over 300 frames. I set this up in five minutes, so it looks terrible. I guess if I made an actual effort, it could look nice. But, it can be done.

I am not a patient person, and you were right. HardFX was too slow to apply to 1000 boxes directly, so I cheated this one. But it's all LightWave. HardFX, SoftFX, and HardLink. Yup.

EDIT -- I spent another five minutes on it and made it slightly less terrible. The link above leads to the new calculation -- again, less then three minutes for 300 frames. Not real-time, but enough to tide me over until HardFX gets faster.

Para
03-15-2005, 05:07 AM
You really should check NovodeX Rocket demo linked earlier, it calculates that kind of stuff on my computer around 90-300fps withouth clearly visible clipping and the pieces even stabilize correctly.

Ztreem
03-15-2005, 05:22 AM
Not yet.

For what it's worth, the following took less than three minutes to calculate on an Opteron 246:
http://www.celshader.com/gallery/md/boxPuddle2.mov

One thousand boxes collapse into a puddle over 300 frames. I set this up in five minutes, so it looks terrible. I guess if I made an actual effort, it could look nice. But, it can be done.

I am not a patient person, and you were right. HardFX was too slow to apply to 1000 boxes directly, so I cheated this one. But it's all LightWave. HardFX, SoftFX, and HardLink. Yup.

EDIT -- I spent another five minutes on it and made it slightly less terrible. The link above leads to the new calculation -- again, less then three minutes for 300 frames. Not real-time, but enough to tide me over until HardFX gets faster.

OK! Maybe you could get the calculation times down a bit with workarounds. But when I look a the result you made it's worthless. The objects don't behave realistic and the go through each other.

Celshader
03-15-2005, 05:23 AM
You really should check NovodeX Rocket demo linked earlier, it calculates that kind of stuff on my computer around 90-300fps withouth clearly visible clipping and the pieces even stabilize correctly.

Dammit, it's making me upgrade to DirectX 9c.

This had better be worth the hassle...

Celshader
03-15-2005, 05:26 AM
OK! Maybe you could get the calculation times down a bit with workarounds. But when I look a the result you made it's worthless. The objects don't behave realistic and the go through each other.

Hey, I warned you it was terrible. What do you expect for five minutes' worth of actual work? :D But it's not worthless. I could still use this in the future for a distant shot. With motion blur on and particles spraying everywhere, who's to know?

EDIT -- OK, I tried putting a little effort into my HardFX calculation this time. This took three minutes to calculate, too:

http://www.celshader.com/gallery/md/cocoPuffs.mov

Yup, grey Co-Co Puffs. They show up better in OpenGL that way.

I think I finally got a clue to HardFX's "stopByStabilizer." Anything moving gets calculated at 1/10th of the normal calculation resolution for better collision detection. Anything not moving so fast gets calculated at the normal, higher resolution for more stability. The FXBreak docs almost explain (http://www.dstorm.co.jp/FXBREAK/e/11up.html) this, but I didn't "get" it until I tried using it myself. Go figure.

Now I just have to figure out at what point a calculation resolution gets reduced by 1/10th. It could be a specific velocity, or it could ramp it up and ramp it down as the object speeds up and slows down. Gotta experiment...

Ztreem
03-15-2005, 05:27 AM
Dammit, it's making me upgrade to DirectX 9c.

This had better be worth the hassle...

I promise, it's worth it. If you like playing around with dynamics like I do you will really enjoy this demo. :D

When you start the program choose demos in the menu and choose one of them, interact with the items with right mousebutton.

Celshader
03-15-2005, 05:32 AM
I promise, it's worth it. If you like playing around with dynamics like I do you will really enjoy this demo. :D

When you start the program choose demos in the menu and choose one of them, interact with the items with right mousebutton.

ARGH! Now DirectX wants me to restart my machine!!! I've got stuff open, dammit!!! :eek:

This had REALLY better be worth it!!!

EDIT -- Hm. Interesting. So far it's fun to play with. Definitely worth the hassle of installing DirectX 9c. None of it has been real-time so far, and it jitters just like HardFX on my system (I'm looking at the shimmering, wavering, toppled dominos in the "domino spiral" scene right now).

I wonder how long it took to set up each and every one of these scenes. Fooling around with it's a breeze, but someone had to set all this stuff up, first, and assign physical properties to everything.

That said, it's fun to play with. Any 3D software package have this built-in right now? Maya, 3DS, Cinema4D, anybody?

Para
03-15-2005, 07:03 AM
NovodeX isn't in any major 3D app. at the moment (as far as I know of) but for example 3dsmax uses Havok (http://www.havok.com/) and XSI uses ODE (http://www.ode.org/) for dynamics. Both Havok and ODE are meant primarly for real-time physics.

Emmanuel
03-15-2005, 07:08 AM
It would be cool if the dynamics would be updated by 8.5.
Has anyone expereince with C4Ds system/shortcomings ?
Then there is the ODE project.
There is enough choice for NT by now to make any "erm, we work on it"
a pretty weak argument :/
That aside, I just *love* to set up dynamics by hand :)
I mean, people watch movies because reality is boring, if any car crash
in a action movie was *realistic*, a lot of people would be disappointed.
Whenever I need something looking "hard body dyanmics" I watch movies and just
fake it so it looks interesting enough without beeing realistic.

Edit: Hey, i think the CocoPuffs are nice, maybe they should move a little longe after
touching the ground, but hey: if people consider THIS bad, I really wonder what they exspect of a 1200 dollars all-purpose-package :/

mrunion
03-15-2005, 07:22 AM
Reading this thread does bring to my mind what seems to be a recurring theme:

Collision Detection.

When simulating "physics" (or "dynamics") in 3D apps, game engines -- whatever -- there are really two problems needing solved: detecting collisions and recating to those collisions.

Reacting to collisions is "easy" compared to the remaining problem. Formulas have been around for quite some time to handle that. We still approximate -- we ALWAYS approximate (for example, area under a curve, that's an approximation!). But we approximated based on "real" circumstances -- like using limits in calculations, small time-change values for better approximations of instantaneous rates of change, etc.

On the other hand, collision detection is NOT a physics problem. It is definitely a "problem", but not physics. In the real world, collisions "happen". A master simulator doesn't detect them, then react. To that effect, there are many ways to handle collision detection. Collision detection is "hard".

In my opinion, Lightwave needs better collision detection. Collision detection takes time -- I'd argue that it averages more time than the actual "physics" part of a simulation. For example, 1000 boxes in the simlation presented can be treated as 1000 point mass objects by the physics engine. *After* a collision occurs it is a simple matter of a bit of vector math, and some force application (and torque calculation) to find the new position of the centerpoint of the cube. From that a transformation matrix moves all the points.

But to see *if* two cubes collided, that isn't simulated by looking at two point mass objects as if they were TWO objects. You actually have to compare six planes of EACH cube. Without any "shortcutting", you have to do 36 comparrisons. I'd say LW is looking at a little less than than, but when it still has to compare each cube to the other cubes, THAT's the slow part.

Unfortunately it also seems to be the inaccurate part.

Ztreem
03-15-2005, 08:30 AM
Collision Detection.

Maybe it's not the psysics calculation that is the problem. Maybe it's just the collision detection that need some changes, I don't know.

The only thing I care about is that NewTek fixes it so it looks great and make the use of dynamics in LW a joy. :D

Mike Pauza
03-16-2005, 09:29 AM
I just looked at some videos of the new Unreal game engine, and WOW the dynamics looks better and more stable than LW's, but in realtime.



Ztreem, Unreal's new game engine has AGEIA's awesome physics software in it.

The cool thing for us though is that AGEIA wants their upcomming physics accelerator chips to serve not just the games industry, but the animation industry as well. They sound impressive to me...have contracts with "key animation software vendors" already in place, have raised 38Million in funding, and even have a working prototype:
http://www.theinquirer.net/?article=21681

These new physics chips should provide much better technology for everyone....much like GPU's have done.



-Mike

Ztreem
03-16-2005, 12:01 PM
Ztreem, Unreal's new game engine has AGEIA's awesome physics software in it.

The cool thing for us though is that AGEIA wants their upcomming physics accelerator chips to serve not just the games industry, but the animation industry as well. They sound impressive to me...have contracts with "key animation software vendors" already in place, have raised 38Million in funding, and even have a working prototype:
http://www.theinquirer.net/?article=21681

These new physics chips should provide much better technology for everyone....much like GPU's have done.



-Mike

I know, I've played a bit with the demo and I'm impressed.

I would like to see that engine in LW, seems better than what we got today.
The PPU thing sound very interesting. :D

hrgiger
03-16-2005, 12:51 PM
The cool thing for us though is that AGEIA wants their upcomming physics accelerator chips to serve not just the games industry, but the animation industry as well. They sound impressive to me...have contracts with "key animation software vendors" already in place, have raised 38Million in funding, and even have a working prototype:
http://www.theinquirer.net/?article=21681

-Mike

Man, that would be good news.

Para
03-16-2005, 02:07 PM
I know, I've played a bit with the demo and I'm impressed.

By the way, am I the only one who has great fun with the running horse demo? I just love the way it tumbles over :D

hrgiger
03-16-2005, 02:19 PM
That's a good one, although, I'm kind of partial to the tall Janga one.

Ztreem
03-16-2005, 03:45 PM
By the way, am I the only one who has great fun with the running horse demo? I just love the way it tumbles over

Yeah! That one is quite funny, It falls so nice on all the hard stones. :D

qwertykd
03-16-2005, 04:00 PM
yeah, the tall jenga one is my favorite, i love knocking all of them over and seeing how everything reacts.....this engine is definitely very very nice....i dunno what kind of technical differences it might have from being useable in something like lightwave, but i remember that ODE which was being ported to lw being pretty fast (though not realtime)

Celshader
03-16-2005, 04:03 PM
I had to install DirectX 9c before it would work on my machine. Does the Novodex Rocket demo rely heavily on the graphics card for its calculations?

Ztreem
03-16-2005, 04:08 PM
I had to install DirectX 9c before it would work on my machine. Does the Novodex Rocket demo rely heavily on the graphics card for its calculations?

I don't think a graphic card is capable of doing such a thing, but the new PPU card will. The directX 9c is more for the graphics in the demo( I think).

Didn't you like the demo?

hrgiger
03-16-2005, 04:12 PM
I had to install DirectX 9c before it would work on my machine. Does the Novodex Rocket demo rely heavily on the graphics card for its calculations?

Not sure. Most of the demos responded pretty well for me and I only have a 64MB GeForceMMX card which is pretty low end as far as cards go nowadays. The only few demo that I remember having a hard time with was the exploding building. Everything else, I was very happy with the simulations.

Mike Pauza
03-17-2005, 07:37 AM
I had to install DirectX 9c before it would work on my machine. Does the Novodex Rocket demo rely heavily on the graphics card for its calculations?

Jen, I'm 99% sure the demo is software only right now, but it's certainly possible for a newer graphics card to accelerate the math. There's a company (much much smaller than AGEIA) that claims to be able to do all sorts of things with GPU's:
http://www.gputech.com/Cms/index.php?option=com_frontpage&Itemid=1



-Mike

Para
03-17-2005, 07:47 AM
The demo uses DX9c simply because it's compiled with it (those shadows in the demo are most likely just an extension of D3D). There's no magic behind the software in that sense which would make it strictly DX9c.

Tesselator
03-17-2005, 09:02 AM
CelShader,

You might find the scenes you were looking for here:

http://www.dstorm.co.jp/FXBREAK/e/

Ernest
03-19-2005, 10:12 AM
Sorry, I know its late, but I just had to say thank you to Celshader.

Albertdup
02-16-2006, 02:12 AM
I know this is an old thread but hey . Softimage XSi have the new AGEIA PhysX engine that drives its dynamics. Here is a video demo. http://www.softimage.com/downloadsrv/process.asp?file=/Videos/NFT50/12_Rigid_Body_Physics.wmv
http://www.softimage.com/education/Xsi/SelfPacedLearning/Tutorials/webTutorials/XSI_5_0/physx_DT/physx_DT.mov

This clearly show what is posable. NT can do it also.

Red_Oddity
02-16-2006, 03:15 AM
And Maya can use the PhysX Engine aswell by using a plugin called Nima (http://www.feelingsoftware.com/index.php?option=com_content&task=view&id=45&Itemid=67)

Come on you Lighwave!

Ztreem
02-16-2006, 04:22 AM
I know this is an old thread but hey . Softimage XSi have the new AGEIA PhysX engine that drives its dynamics. Here is a video demo. http://www.softimage.com/downloadsrv/process.asp?file=/Videos/NFT50/12_Rigid_Body_Physics.wmv
http://www.softimage.com/education/Xsi/SelfPacedLearning/Tutorials/webTutorials/XSI_5_0/physx_DT/physx_DT.mov

This clearly show what is posable. NT can do it also.

That's how Hardbody dynamics should work, I think I must try a demo of XSI it seems to be so much faster than LW.

Red_Oddity
02-16-2006, 10:46 AM
That's how Hardbody dynamics should work, I think I must try a demo of XSI it seems to be so much faster than LW.
Most of the stuff is practically realtime with the PhysX engine, i have had no problems running a simulation with about 10000 bricks in XSI....LW chokes at about 10 bricks, sorry to say.

hrgiger
02-16-2006, 11:23 AM
Well, they say they are working on all aspects of the program. I'm hoping that dynamics is an area they are going to put some extensive work into. The LW8 dynamics implementation hits the tree, but misses the target.

Ztreem
02-16-2006, 12:50 PM
The LW8 dynamics implementation hits the tree, but misses the target.

I Agree.


Most of the stuff is practically realtime with the PhysX engine, i have had no problems running a simulation with about 10000 bricks in XSI....LW chokes at about 10 bricks, sorry to say.

Not only does LW choke with 10 ten bricks, the bricks pentrate each other too. :thumbsdow

Stooch
02-16-2006, 04:46 PM
messiah has incredible fast dynamics and UI in general. add to the fact that you can connect anything from messiah to LW, you got yourself an excellent dynamics solution.

and incredibly fast renderer as well! i just did a bunch of tests on the messiah forums on CG talk.