PDA

View Full Version : Some Bullet Dynamics Fun



Spinland
09-04-2015, 11:16 AM
Working on an explainer video for a construction company, and decided this sequence would be a good fit for Bullet. For me the most time-consuming aspect was tweaking the densities of the chains and the slab so there was a sense of weight without breaking the chains. The simulation seems very sensitive to a careful balance.


http://www.youtube.com/watch?v=8bcVo4ORPdk

Greenlaw
09-04-2015, 11:38 AM
Nice practical use of Bullet!

G.

Spaceland
09-04-2015, 11:56 AM
interesting animation. Nice work.

Snosrap
09-05-2015, 03:38 PM
Simple but Very impressive!

Spinland
09-05-2015, 03:42 PM
Thanks! Yeah, it's pretty simple, this is just a proof of concept to make sure I have the Bullet techniques down before I apply them to the actual animation sequence. I have already switched to a simpler scene where the excavator is replaced by some simple proxies and then baked out the final dynamics motions for the chains and the slab into a couple of MDDs. It came out beautifully in the actual animated scene. Good stuff!

Snosrap
09-05-2015, 03:50 PM
Using 2015 or 11.6?

Spinland
09-05-2015, 03:52 PM
2015.2

Greenlaw
09-05-2015, 11:23 PM
'Simple' but only if you're using Bullet in LightWave 2015. Try animating those chains using Bullet in 11.6. :)

Bullet in 2015 has seen such great improvements with features like Bones, Motor and other useful Constraints. Even Start Time Offset, which seems like a small thing, saves me a lot of time because I can now use actual frames ranges when running my sims passes. In fact, 2015 got me though a number of tasks earlier this year that would have been impossible or at least very difficult to do with earlier versions of LightWave.

This is why I'm baffled when some users still insist that 2015 isn't a worthy upgrade. These days, I only launch 11.6 to run one third party plugin that's not yet 2015 compatible. Otherwise it's 2015 all the way for me.

G.

hdace
09-07-2015, 09:03 PM
Agreed about 2015 generally and also especially about the Start Time Offset. Boy, that was suuuuuch a critical update!

Spinland
09-08-2015, 07:10 AM
Yeah, definitely a great add! Honestly I hadn't needed much Bullet stuff until this recent gig and that offset is golden. Having to time the proxy dynamics to the actual scene without that feature would have been a total pain.

jwiede
09-08-2015, 01:24 PM
'Simple' but only if you're using Bullet in LightWave 2015. Try animating those chains using Bullet in 11.6. :)

That's kind of the problem: SBDs in V11 weren't particularly usable (read as "reliably functional"), even for simple models like simulating point-to-point chains. Then again, HBDs in V11 weren't all that stable or functional either.


This is why I'm baffled when some users still insist that 2015 isn't a worthy upgrade.

At least in my case, the problem was that Bullet Dynamics features delivered in V11 weren't fixed to a point of stable usability within that version. The fixes that made basic SBD sims stable and reliable should have been delivered in V11, not 2015. Alternately, had they focused on delivering HBD and constraints, bringing those to stable, usable functionality in V11, and holding back SBDs altogether until 2015, that would have been fine as well. Instead, in V11 LW3DG delivered seriously under-functional HBDs and unstable/unusable SBDs (neither really reaching usability within that version), then in 2015 (at additional cost) delivered the stable, fully-functional releases of Bullet Dynamics features.

Their approach, IMO, was wholly unacceptable treatment of V11 customers, essentially charging us for fixes. That treatment was a huge factor in why I did not purchase the 2015 upgrade. As a user, I felt LW3DG hadn't yet fulfilled their feature deliverables for V11, then turned around and demanded payment to receive the functional/usable versions of those features (in 2015).

This was not a case where LW3DG had no viable options outside their path: There was an easily-visible way to staged the Bullet releases such that both V11 and 2015 customers would receive stable, functionally-useful dynamics features within their respective releases (and thereby fair value), by delivering fixed HBD+constraints in V11, and fixed SBD(+constraints, implied) in 2015. In fact, many users were suggesting that LW3DG deliver Bullet Dynamics in just such a manner, starting well before V11's release and continuing throughout the V11 development cycle.

IOW, it was that too much of what makes 2015 a "worthy" release came at the expense of V11 customers.

Greenlaw
09-08-2015, 02:11 PM
In my case anyway, I certainly found Bullet incredibly useful during the 11 cycle. Was it a perfect tool? No. Was it appreciated? Most definitely, having used it successfully for several commercials, short films, video game cinematics and features. Between 2012 and the beginning of 2015, I felt the 11 cycle paid for itself many times over, and I could say that just from using Bullet in 11 alone. When 2015 rolled around, upgrading was a no-brainer for me because it made the program even more productive. But of course, your mileage may vary.

IMO, if a tool doesn't meet the requirements of the job, it's time to look at other options. Otherwise, it's kind of silly to keep using it and grumbling about it.

I think if I had any problems with 2015, it was that LW3DG should have announced it sooner. I think many users (myself included) were expecting an '11.7' release and they should have squashed that rumor long before they released 2015. But once I started using 2015 for work, I got over that annoyance pretty quickly. :)

G.

Greenlaw
09-08-2015, 02:24 PM
IOW, it was that too much of what makes 2015 a "worthy" release came at the expense of V11 customers.
Not if they didn't upgrade. :p

Spaceland
09-08-2015, 03:31 PM
Not if they didn't upgrade. :p

Even if they didn't upgrade, something broken and fixed with a new version is still at the expense of the version before.

jwiede
09-08-2015, 06:08 PM
Even if they didn't upgrade, something broken and fixed with a new version is still at the expense of the version before.

Exactly. V11 customers wound up with unstable/unreliable SBDs, and comparably-minimal HBD capabilities. LW3DG could have addressed that by releasing final 11.7 bug-fix update for V11, but chose not to do. That those fixes were released in 2015 in no way improves the (final) state of V11 software for customers.

hdace
09-08-2015, 07:42 PM
Now that they've changed to an Autodesk-style naming/upgrade convention all of this arguing is academic. Software companies often cheat when it comes to the difference between an update and a paid upgrade, saying there are new features which are nearly useless when really they're just fixing bugs (and introducing new ones!). The new system is more honest.

At the end of the day you've got a bunch of programmers doing their job and they have to be paid. It makes sense that we should pay them on a regular basis. Once a year, or even once a month. We're using their software and we always want the most recent stuff 'cause if one isn't using it, someone else is and their stuff might look better than one's own. If you don't think that's fair then take the risk and keep using the old stuff. But your reasoning about how they're cheating us doesn't really work anymore.

Spinland
09-08-2015, 07:49 PM
I'm not going to take sides in any "done us wrong" type of discussion. For me, LW is a major tool in the kit whereby I earn my living. To my way of thinking keeping current with the latest version is just a professional expense like renewing my business insurance every year, and just as deductable come tax return time.

I'm trained and experienced in 3DS Max and Maya, and I chose LW as my focus. Ponying up regularly for the periodic updates is, as I see it, my part in the relationship whereby I get to use my favorite tool to do my job.

Greenlaw
09-09-2015, 02:25 AM
hdace and Spinland,

I agree with both of you and I couldn't have said it any better.

G.

Spinland
09-09-2015, 04:23 AM
Getting back to the topic, I've not been happy with the amount of sag in the chains. Reducing that is, as I see it, a function of making the chains lighter and the slab heavier, but before I reach the point where the sag straightens out the chains start breaking.

From what I understand based on reading the docs and searching here the solution appears to be having a large enough value defined in Dynamics Framerate. The issue with finding that value has been that, in this case, even 5000 turned out to be insufficient and once you start getting into big numbers the sequence takes forever to run and so an iterative approach to finding the sweet spot becomes impractical.

What I just did was set the weight values where I've found the tension pulls just taut enough, then start the simulation with a crazy value for framerate: 20,000. I decided screw incremental testing. It's been running overnight and is still only 75% done but it's already past the chain tightening phase with no breaks so all that's left is to wait for the slab to be put down again and then bake this sucker off into MDDs and declare victory.

hdace
09-09-2015, 05:02 PM
Hope that works. Sounds good.

I sometimes fix problems by writing expressions for the gravity envelope curves, using a special "velocity" null (again, requires writing an expression). I parent a velocity null to something that's affecting the item that isn't moving correctly, then in the gravity curve expression I adjust a negative or positive multiplier until I get exactly the right result from the sim. Sometimes x, y, or z gravity, appropriately applied, can help push something that isn't moving quite enough, or help slow something that's moving too much.