[QUOTE=Grayman;45815518]That's a good idea, but I think the volume is needed. Feel free to prove me wrong, preferably with a link to a detailed explanation.
I'm not majoring or minoring in physics (oddly enough) so I wouldn't know.[/QUOTE]
Well when youre talking about points, are you talking about vertices, or the object is split into a certain amount of points per cubic unit of some sort? I have no idea about point clouds.
What i am saying isnt a realistic water simulation whatsoever, its just more realistic than it is at the moment. The only issue is that if it wasnt just an amount of points per cubic unit, but just vertices, if you had an object that was complex on top and simple on the bottom, it wouldnt float right.
[QUOTE=Walrus Viking;45813997]Wait what? [img]http://puu.sh/b9ULT/559a314713.png[/img][/QUOTE]
4Mass divided by the number of points in the object. Then the output of that would be applied perpendicular to the water surface on each point that was touching the water.
The 4 isnt special either, its just a rough guess, if the points were spread out evenly, assuming its PPCU rather than vertexes, it would work fine.
[QUOTE=Phoenixf129;45816283]I doubt anyone at VALVe was either when vphysics was written, so no issues there![/QUOTE]
I believe vphysics was havok, and it was licensed to valve, so thought did go into it :/
ACF doesnt work, its applying the torque to the yaw rather than the pitch or roll.
Awesome work so far, any idea what's causing this though?
[VID]http://a.pomf.se/mfpftu.webm[/VID]
[editline]31st August 2014[/editline]
Damn that is some poor video quality. The airboat doesn't vanish at the end, it just spins off into the distance.
[QUOTE=jackwilsdon;45854707]Awesome work so far, any idea what's causing this though?
[VID]http://a.pomf.se/mfpftu.webm[/VID]
[editline]31st August 2014[/editline]
Damn that is some poor video quality. The airboat doesn't vanish at the end, it just spins off into the distance.[/QUOTE]
Inverse damping?
[QUOTE=ricky23;45854826]Inverse damping?[/QUOTE]
Wheels stuck in the chassis. Suspension gets applied to try and correct it but never works because the wheels never leave the chassis, so the airboat flips the fuck out.
Yeah, the airboat needs to be rewritten anyways. Valve wrote it themselves into IVP, so I'll peek into that code and rewrite the airboat for bullet physics... eventually.
[QUOTE=Grayman;45856271]Wheels stuck in the chassis. Suspension gets applied to try and correct it but never works because the wheels never leave the chassis, so the airboat flips the fuck out.
Yeah, the airboat needs to be rewritten anyways. Valve wrote it themselves into IVP, so I'll peek into that code and rewrite the airboat for bullet physics... eventually.[/QUOTE]
What about the applytorque on the incorrect axis?
Other than that issue, and some issue with ballsockets (theyre created at weird angles when you paste a duplication), this is almost a daily driver for me. I love the work youve done so far
[QUOTE=ricky23;45859139]What about the applytorque on the incorrect axis?
Other than that issue, and some issue with ballsockets (theyre created at weird angles when you paste a duplication), this is almost a daily driver for me. I love the work youve done so far[/QUOTE]
Can you construct a test case for me to use with ACF? Just post a picture labeling how you create it and maybe a video demonstrating what's wrong specifically.
[del]I'll look into ballsockets being weird when duplicated.[/del] Looked at ballsockets, I don't see anything weird. Could you be a little more specific please?
[QUOTE=Grayman;45864874]Can you construct a test case for me to use with ACF? Just post a picture labeling how you create it and maybe a video demonstrating what's wrong specifically.
[del]I'll look into ballsockets being weird when duplicated.[/del] Looked at ballsockets, I don't see anything weird. Could you be a little more specific please?[/QUOTE]
Sure, ill post some pictures in one second
Setup, motor on and at full throttle, not connected to the wheel
[IMG]http://puu.sh/bhoG6/f63260f988.jpg[/IMG]
Attached transmission to wheel
[IMG]http://puu.sh/bhoTQ/32347e8f18.jpg[/IMG]
Whats the crap to make a thumbnail of the image so its not full size
In case you cant tell, it went spinning off like when you flick a coin.
Video of ACF
[URL="https://dl.dropboxusercontent.com/u/96179710/videos/quad%20%2810%29.mp4"]https://dl.dropboxusercontent.com/u/96179710/videos/quad%20%2810%29.mp4[/URL]
Video of bad ballsocket duplication
[URL="https://dl.dropboxusercontent.com/u/96179710/videos/quad%20%2811%29.mp4"]https://dl.dropboxusercontent.com/u/96179710/videos/quad%20%2811%29.mp4[/URL]
Please excuse the quality, im on DSL and cant upload hundreds of megs in a couple minutes like everyone else.
[QUOTE=ricky23;45864907]Sure, ill post some pictures in one second
Setup, motor on and at full throttle, not connected to the wheel
[IMG]http://puu.sh/bhoG6/f63260f988.jpg[/IMG]
Attached transmission to wheel
[IMG]http://puu.sh/bhoTQ/32347e8f18.jpg[/IMG]
Whats the crap to make a thumbnail of the image so its not full size
In case you cant tell, it went spinning off like when you flick a coin.
Video of ACF
[URL="https://dl.dropboxusercontent.com/u/96179710/videos/quad%20%2810%29.mp4"]https://dl.dropboxusercontent.com/u/96179710/videos/quad%20%2810%29.mp4[/URL]
Video of bad ballsocket duplication
[URL="https://dl.dropboxusercontent.com/u/96179710/videos/quad%20%2811%29.mp4"]https://dl.dropboxusercontent.com/u/96179710/videos/quad%20%2811%29.mp4[/URL]
Please excuse the quality, im on DSL and cant upload hundreds of megs in a couple minutes like everyone else.[/QUOTE]
Bam! ACF issue fixed. That video you have on the ballsockets really doesn't help, can I get a minimal case where it happens?
[QUOTE=Grayman;45865471]Bam! ACF issue fixed. That video you have on the ballsockets really doesn't help, can I get a minimal case where it happens?[/QUOTE]
WOW : D youre awesome.
Hmm, let me dink around with ballsockets and ill narrow it down for you. In the situation where it was used in the video, the plates that rotate are frozen with propFreeze and are being rotated with setAngle
The wheels theyre ballsocketed to are makesphereical, so that may be the issue in itself.
EDIT
Heres a video of a very simple setup. The wheel is again ballsocketed to the same plates that are moved with setangle and are frozen in the air with propFreeze.
The wheel is rigid roped in a triangle shape to allow for vertical movement only while still maintaining rotation. Said wheel is then elasticed to a block above it to act as suspension.
[URL="https://dl.dropboxusercontent.com/u/96179710/videos/quad%20%2812%29.mp4"]https://dl.dropboxusercontent.com/u/96179710/videos/quad%20%2812%29.mp4[/URL]
if theres anything you want me to do, please tell me as i would like to assist you in any way possible.
Edit again
Makespherical'd wheels cause an immense amount of lag easily. This is only through MOVEMENT, not through rotation. I can spin a wheel 10k rpm and have no issues, but as soon as it starts moving across the ground or through the air, the game pretty much grinds to a hault.
edit again again.
The issue with makespherical can be solved by turning substeps to 1. Anything higher than one causes insane lag. Same goes for the energy balls that the plasma rifle shoots when you right click.
sooo apparently you have some issues with collision shapes and the number of substeps.
Fixed slowness with sphere shapes. Continuous sweep testing was enabled for convex shapes still (which is really slow, I need to find a quick way to do this).
For that wheel test, I'm assuming the wheel is only supposed to turn, and not move up/down when turning?
[QUOTE=Grayman;45866879]Fixed slowness with sphere shapes. Continuous sweep testing was enabled for convex shapes still (which is really slow, I need to find a quick way to do this).
For that wheel test, I'm assuming the wheel is only supposed to turn, and not move up/down when turning?[/QUOTE]
Yep, weird forces are being applied.
On wheels that are spherical, ballsockets mess up alot, as seen in the first ballsocket video.
This may be very vague, but are there plans or tests for players being able to stand on moving props without being shot back weirdly. Sort of like...boats moving without you being thrown off the back?
[QUOTE=pilot;45868411]This may be very vague, but are there plans or tests for players being able to stand on moving props without being shot back weirdly. Sort of like...boats moving without you being thrown off the back?[/QUOTE]
It isnt as easy as it sounds. Its definitely not like a box sliding on a plate where theres friction between two entities.
Also, grayman. There are some serious issues involving collision shapes and moving on the ground. They get caught up and randomly start trying to flip certain directions. Its very weird, almost like they lag behind when getting moved along and the physics for them gets messed up.
[QUOTE=ricky23;45868552]It isnt as easy as it sounds. Its definitely not like a box sliding on a plate where theres friction between two entities.
Also, grayman. There are some serious issues involving collision shapes and moving on the ground. They get caught up and randomly start trying to flip certain directions. Its very weird, almost like they lag behind when getting moved along and the physics for them gets messed up.[/QUOTE]
Actually... Player collisions are exactly like a box sliding on an object with friction holding them together. Shit only begins to fuck up when input is applied through the player controller.
As of how I understand it now, it looks like source doesn't even factor in the ground's velocity as a base so when you try walking on a fast moving object you'll be shot backwards into space.
Also I hope that second issue you're describing isn't the return of the inverted contact points... I really hate those. If it's objects randomly stopping when crossing internal edges, I know about that issue. For that, basically bullet cannot detect internal edges between two different objects (or in some cases even within the same object).
[QUOTE=Grayman;45870685]Actually... Player collisions are exactly like a box sliding on an object with friction holding them together. Shit only begins to fuck up when input is applied through the player controller.
As of how I understand it now, it looks like source doesn't even factor in the ground's velocity as a base so when you try walking on a fast moving object you'll be shot backwards into space.
Also I hope that second issue you're describing isn't the return of the inverted contact points... I really hate those. If it's objects randomly stopping when crossing internal edges, I know about that issue. For that, basically bullet cannot detect internal edges between two different objects (or in some cases even within the same object).[/QUOTE]
Hmm whats the best way for you to diagnose this?
[QUOTE=ricky23;45874081]Hmm whats the best way for you to diagnose this?[/QUOTE]
Post a video of it happening. Simple test case though.
[QUOTE=Grayman;45875039]Post a video of it happening. Simple test case though.[/QUOTE]
Will do, give me 5 minutes
EDIT
there was a new build today, 9972 that im pretty sure i didnt have before. The issue did not happen.
The ballsockets on the spheres are still weird though.
EDIT
Well grayman, i figured out the ballsockets.
Free movement is broken.
Thats it, thats the only thing thats keeping bullet physics from being used to build cars.
I was wondering where this went. Cool to see it's still in development. If vehicles will be controlled by bullet soon, does that mean GTA style driving physics are within reach?
[QUOTE=Grenadiac;45878176]I was wondering where this went. Cool to see it's still in development. If vehicles will be controlled by bullet soon, does that mean GTA style driving physics are within reach?[/QUOTE]
Are we talking gtaiv (real) or gtav?
Its already possible through building your own vehicles :p
in all seriousness i believe this was talked about a couple pages back.
Also grayman, i tried to send a small donation your way for being so awesome and my computer crashed. Took care of it though
Gonna try to get more people interested in this now that its 99.99 percent functional with any serious building. After you get the ballsocket free movement fixed up (a very popular feature) Im sure that this will spread to more building communities as everyone despises the current vphysics.
Yeah, I'll have to look into ballsockets. There's no specific parameter on the params passed to CreateBallsocket so it looks like GMod does something weird if free movement is enabled.
Yeah, I can change the vehicle driving logic to be anything I'd like. I'm exposing vehicle controllers to lua though so others can screw with making their own cars.
Really? Are you looking at the adv ballsocket tool? Im no lua expert but as far as i can tell its refering to it as
"onlyrotation"
[QUOTE=ricky23;45880907]Really? Are you looking at the adv ballsocket tool? Im no lua expert but as far as i can tell its refering to it as
"onlyrotation"[/QUOTE]
Okay, I see it now. The advanced ballsocket tool creates a ragdoll constraint instead of a ballsocket constraint. So I guess that means ragdoll constraints are still half broken, as usual.
[editline]Edited:[/editline]
So apparently the advanced ballsocket tool isn't in vanilla sandbox. Great.
[QUOTE=Grayman;45881011]Okay, I see it now. The advanced ballsocket tool creates a ragdoll constraint instead of a ballsocket constraint. So I guess that means ragdoll constraints are still half broken, as usual.
[editline]Edited:[/editline]
So apparently the advanced ballsocket tool isn't in vanilla sandbox. Great.[/QUOTE]
I am sorry, I forgot to mention that. You should be able to find it on the workshop easily. It's a very useful tool and everyone Who makes anything usesnit
[editline]Edited:[/editline]
So, what do you think of the situation and is it semi easily fixable?
I saw some files updated on github updated with angularonly and linearonly. Were these just prerequisites to fixing up ragdolls/advanced ballsocket? As far as i can tell, nothing was changed in game
[QUOTE=ricky23;45884774]I saw some files updated on github updated with angularonly and linearonly. Were these just prerequisites to fixing up ragdolls/advanced ballsocket? As far as i can tell, nothing was changed in game[/QUOTE]
Yes. I'll get around to putting it into the interface, give me a moment.
[QUOTE=Grayman;45885510]Yes. I'll get around to putting it into the interface, give me a moment.[/QUOTE]
You have no idea how excited i am
[QUOTE=unrezt;45729015]It is if it doesn't happen with the default physics.[/QUOTE]
[QUOTE=briman0094;45711575]Wheels and welds are MUCH better than with the default vphysics, but if two props collide for too long (more than just a bounce) it drops my client to about 1-2 fps because of all the dust particles and banging sounds. Any chance this could be fixed? I have an nVidia 650M card...lag is not common in any game except for this case where there are tons of particles spawned every tick.[/QUOTE]
Also, sorry you guys, i feel like an ass. I wasnt having any issues until i had about 10 constant collisions going on (things that should have been nocollided) and that kinda screwed things over
Okay. Angular only should be implemented with ragdoll constraints. I may or may not have actually tested this before committing it, so it may or may not work.
[QUOTE=Grayman;45886142]Okay. Angular only should be implemented with ragdoll constraints. I may or may not have actually tested this before committing it, so it may or may not work.[/QUOTE]
Will test
thanks!@?!#
Ho ho holy shit grayman, it works like a charm.
Also hydraulic (spring) damping doesnt work
Also, congrats! you cured manwhore syndrome!
Sorry, i gotta edit again, i cant believe how fucking great this is!
Sorry for the lack of embedded video, just recorded this and threw it up
I am also calling first in the world for building a car on bullet physics in gmod on the source engine. Thanks :D
[URL="https://dl.dropboxusercontent.com/u/96179710/videos/quad%20%2816%29.mp4"]Best fucking physics ever[/URL]
Manwhore syndrome is not fixed. It's an issue that's conveniently with all male ragdolls, their legs have internal collisions that are only solved when open. Looks like I'm only gonna be able to fix this by making ragdoll constraints weaker, yay.
[editline]Edited:[/editline]
By the way, you can downscale the videos then embed them in vid tags. Please do downscale them because FP does not do that.
Sorry, you need to Log In to post a reply to this thread.
