|
|
Rube Goldberg scenario
About Truespace Archives
These pages are a copy of the official truespace forums prior to their removal somewhere around 2011.
They are retained here for archive purposes only.
Rube Goldberg scenario // The Garage
Post by frootee // Jul 2, 2007, 6:58am
|
frootee
Total Posts: 2667
|
Hi 3dv.
Regarding the wood fixture:
From my experience with physics in truespace, using a perfectly sized hole as a fixture does not work very well - objects stick and do not move. Make the hole a little bigger, then the object wobbles ever so slightly due to momentum. Of course, that may be due to lack of trying different things, like setting friction to 0, but what usually happens is the spinner in our case, starts to move just barely, then vibrate back and forth a tiny bit; as opposed to doing what we want, which is to spin freely. That is one of the purposes of the fixation points: they act as pivot points. Put two of these on each end of our lateral cylinder, and temporarily move the wooden fixtures-with-holes out of the way, and run the simulation and generate keyframes. This allows the spinner to spin freely. Then put the wooden fixture back into place.
One thing I have not tried (which I read this morning) is to set objects to not participate in the physics simulation in the LE. I was wondering how to do that in 7.5. If we can do this, then there would be no need to move the wooden fixture. But we would still need the fixation points I believe, in that case, to allow the spinner to spin along the axis defined by the fixation points.
One question: are the portions of the spinner that contain the 'spoons' equal length? If they are not (one arm is, say, 4 units, and the other is 5), that may cause an imbalance. I did not check that.
Regarding the axes, I did have some difficulty in troubleshooting because I did not align the entire object with the world axes. That would make my life easier. The object itself by default appeared to be rotated about 45 degrees. As for the local object axis system, maybe I could align that to the world coordinate sytem; just to simplify things.
Perhaps another issue to investigate is the center of the local coordinate system: is it at the same location as the center of gravity? Does it matter if they are not colocated?
I noticed that if the frame rate in physics is set too low, the ball literally sticks to the spoon that is struck, then rides on the spoon vertically. I can see why this would cause an imbalance, and hence wobbling. So, possibly some energy is being transferred from the ball UPWARDS on the spinner, which may be causing this vibration. But I was under the impression that the fixation points serve as rigid supports, meaning, even if you drop something onto this spinner (which I will try), the dropped object would fall off, and the spinner would not budge. Perhaps not.
Aaaanyway. I will try some stuff tonight and post the results if I have time.
Thanks!
Frootee
I doubt this would have anything to do with the off-axis movement of the axleshaft but I suppose it might... when I created that particular shaft the hole that it goes through on each end (through the wooden-textured vertical blocks) was very slightly larger in diameter than the actual shaft. If... the initial generation of keyframes from physics which you made somehow "needed" that shaft to be exactly the same diameter as the shaft retaining holes I suppose a slight wobble would get worse as more keyframes were introduced. Just a thought, but again I doubt it.
In the manual videos from chapter 10 / Physics (which Norm uploaded here to the forums after the manual was ready) there is a scene where a sphere is rotated in place along various axes. I would assume that in the case of that particular sphere either world coordinates for spin axis determination were used or the sphere was spun in place acording to it's object axes. The reason I mention this, is that if the axleshaft were lined up perfectly so that the dead center of the axleshaft (along the length of the shaft) were at a precise known Z and Y axis point identical at both ends, with only an X axis point distinguishable between each end point - before any spin simulation or keyframing is done and both before and after a ball is introduced - we should be able to determine then the exact cause of any variation in altitude as it occurs. Makes sense?
- 3dvisuals dude |
Post by 3dvisuals dude // Jul 2, 2007, 12:43pm
|
3dvisuals dude
Total Posts: 1703
|
Hahahaha!!!
Who would'a thought a stupidly simple object like this would become such a PITA! Hahahaha...
Anyhow, I just did an extremely close examination of the object in question here and discovered several ~potential~ factors which ~may~ be part of this problem, I then eliminated them and I'm posting a zipped rsscn file now which will help you diagnose whether they in fact had anything whatsoever to do with the odd wobble we're seeing here. Either way, it will eliminate potential contributing factors in the search for a solution.;)
EDIT: I also carefully checked the positioning of the original paddles and they were all perfectly equidistant from the precise center of the axle axis. The shortening done to one side of the original shaft to make it perfectly equidistant to the paddle's central hub was minor and most likely did not contribute to the problem at all. I hope not also because not all objects can be symmetrical in order to simply have physics applied of course.
It just seems to me that the coordinates are screwy on this somehow as it is, and it may well end up being that the center of mass is being calculated incorrectly due to the non-primitive nature of the overall paddle assembly shape rather than the center axis of the axleshaft... we'll see.
Here's an image with the explanations followed by a zipped rsscn file with two new paddle assembly versions to examine.
I'll keep trying to figure out the problem on this end as well but have unusual family distractions here which will be rather constant over the next three days, so I'm not sure how much I'll be able to concentrate on this. I will try though.
Good luck!
- 3dvisuals dude |
Post by 3dvisuals dude // Jul 2, 2007, 12:56pm
|
3dvisuals dude
Total Posts: 1703
|
Frootee,
There's one other thing I never did mention about this original design. When I first created it I was looking at the paddles with the thought that they should be almost constantly spinning after collision with the marble in order to provide a decent amount of gear turn for whatever energy conversion functions are later applied to the gears.
It occured to me at that time that one good way to gradually slow down the paddle spin to an eventual halt (before another marble collision) would be to have the paddles magnetized as well as the two paddle-opening-shaped fins on each side of the paddles which are attached to the upright axle supports. That's why I made them in the first place.
This way the friction for the assembly could be zero, as it wouldn't be needed to slow down the spinning at all. Needless to say though, the magnetic force applied to both the paddles and the fins would have to be very weak in order not to effect the path of the marble itself to any significant degree.
EDIT: Additionally, friction could be applied to the gear blade interactions to gradually slow the spin to a halt, but the original fin magnetism idea in contrast to this method would ensure that the paddles were always correctly positioned after the spin cycle ends. A variant of this same "magnetism positioning control" concept was also employed in that "food for thought / truePlace trueRoller" model I posted here earlier, in that case the cab would be controlled by mild magnetic force applied to the overhead circle arm, serving as a sensor to activate a script to stop cab spin beneath it completely until a "go" signal is activated afterward. This is why it exists there.
- 3dvisuals dude |
Post by frootee // Jul 2, 2007, 2:17pm
|
frootee
Total Posts: 2667
|
#1 WORKS FINE.
Here is the file. |
Post by 3dvisuals dude // Jul 2, 2007, 2:21pm
|
3dvisuals dude
Total Posts: 1703
|
#1 WORKS FINE.
Here is the file.
Awesome!!!!
:jumpy:
Way to go Frootee!
What was the issue by the way? The axle length/symmetry, the non-booleaned assembly version, or something else?:confused:
Man that was FAST! Thanks!:banana:
Need anything else? Just say the word!
- 3dvisuals dude |
Post by frootee // Jul 2, 2007, 2:24pm
|
frootee
Total Posts: 2667
|
And to illustrate the point here:
Yes, for proper rotation we do need symmetry, OR, move the COG (Center Of Gravity) for proper balancing, I booleaned a big cube to the end of the lateral cylinder. This changes the mass and of course the COG, but by default the COG is centered. Recall how one end was a little longer than the other? The cube is an exaggerated example of that.
Here's the file.
Lesson learned here. I do think the issue is COG placement. |
Post by 3dvisuals dude // Jul 2, 2007, 2:43pm
|
3dvisuals dude
Total Posts: 1703
|
And to illustrate the point here:
Yes, for proper rotation we do need symmetry, OR, move the COG (Center Of Gravity) for proper balancing, I booleaned a big cube to the end of the lateral cylinder. This changes the mass and of course the COG, but by default the COG is centered. Recall how one end was a little longer than the other? The cube is an exaggerated example of that.
Here's the file.
Lesson learned here. I do think the issue is COG placement.
Well done Frootee...
Sure enough, just tested both your uploads here and that's the issue for sure. I guess we'll have to keep all future spinning objects symmetrical in relation to COG then.... that will definitely have an impact on design ideas.
So since the one you got to work was the booleaned assembly (just as the original assembly was booleaned) and the only changes were (1) to remove symmetrical but unnecessary additional booleaned parts inside each paddle and (2) the shortening of one side of the axle to make it perfectly latteraly symmetrical, it had to be the COG calculated automatically by the physics engine.
Whew.... one down.... next?!?!
What do you want to do next with this project in the way of "fun?" !:D
- 3dvisuals dude |
Post by frootee // Jul 2, 2007, 2:48pm
|
frootee
Total Posts: 2667
|
the unnecessary booleaned parts inside each paddle are fine. Just the extra length on one end was the problem.
The other one, the one with a non-booleaned cylinder, there is a lot of friction there. So the paddle-thing won't move.
Nonsymmetrical parts may be an issue; we just need to ensure we know where to PUT our chosen COG. In the physics buttons, you can make the COG visible and move it.
Note that if we only rotate the nonsymmetrical object, we have no problem. Only when we HIT that object with something.
Not sure where to go next...
gotta eat first!
Frootee |
Post by 3dvisuals dude // Jul 2, 2007, 2:53pm
|
3dvisuals dude
Total Posts: 1703
|
the unnecessary booleaned parts inside each paddle are fine. Just the extra length on one end was the problem.
Nonsymmetrical parts may be an issue; we just need to ensure we know where to PUT our chosen COG. In the physics buttons, you can make the COG visible and move it.
Note that if we only rotate the nonsymmetrical object, we have no problem. Only when we HIT that object with something.
Not sure where to go next...
gotta eat first!
Frootee
(1) First off, eating sounds great... I'm off to the kitchen myself now!
(2) I didn't know that about the visibility/moveability of the COG. Where is that?
(3) Back to number 1
Thanks... let me know where you want to go next with this after some food!;)
- 3dvisuals dude |
Post by frootee // Jul 2, 2007, 4:50pm
|
frootee
Total Posts: 2667
|
In the Physics buttons, select the COG button, shown in a copy here, of the Workspace Toolbar, from the QuickGuide. When you do, the COG will be visible in the workspace. You can then drag it around as you wish.
The good thing about a moveable COG is you can simulate nonhomogeneous weight distribution. For example, your paddle object could have mercury on one side of the cylinder, but not on the other. You can simulate that weight distribution by shifting the COG. You can also get nonuniform objects to rotate in a nice way, with a little effort. :p
One thing I think would be neat to have is one of those drinking birds: they have a vial of fluid. When they heat up, the fluid rises to the top, and the bird tilts down into a glass of water. The water cools the bird, and the fluid goes back down, thus shifting the weight. So the bird raises its head back up.
Frootee
EDIT: IT appears that moving the COG makes no difference if you use 2 fixation points. If you have a wobbling object it will still wobble if you adjust the COG. The mass has to be even |
Post by 3dvisuals dude // Jul 2, 2007, 5:23pm
|
3dvisuals dude
Total Posts: 1703
|
Thanks for the COG info Frootee.
Was it Edgar Allen Poe who said the best place to hide something is right in plain sight?
with a little effort. :p
Just "a little?" ;)
Good to know, thanks for that too.
One thing I think would be neat to have is one of those drinking birds: they have a vial of fluid. When they heat up, the fluid rises to the top, and the bird tilts down into a glass of water. The water cools the bird, and the fluid goes back down, thus shifting the weight. So the bird raises its head back up.
Frootee
Well, if you'll settle for a Shot glass I think that can be arranged.;)
One Feathered Lush coming right up then!:D
Stay tuned!
- 3dvisuals dude
EDIT: not sure how you want to simulate the liquids involved, although some use gasses, here's a related page to think about while I serve you up a model: http://www.backstreet.demon.co.uk/oddstuff/drinkingbirds/drinkingbirds.htm Note that there are also detailed physics help files related to problematic bird function there. Hahahahahaha.... you never know when "problematic bird function" help may just come in very handy!:D |
Post by 3dvisuals dude // Jul 2, 2007, 8:45pm
|
3dvisuals dude
Total Posts: 1703
|
Here we go folks...
The TS7.51 Rsscn file is zipped for download after the image.
Enjoy! :D
- 3dvisuals dude |
Post by 3dvisuals dude // Jul 2, 2007, 9:56pm
|
3dvisuals dude
Total Posts: 1703
|
EDIT: IT appears that moving the COG makes no difference if you use 2 fixation points. If you have a wobbling object it will still wobble if you adjust the COG. The mass has to be even
Hmmm...
What happens if you delete the existing fixation points, move the COG, then create two new fixation points in relation to the new COG? :confused:
Oh, the bird was fun to make by the way... never did that before. What else do we need next now that is non-bird-related for our physics objects inventory here? (Spice of Life and all) :)
I was kind of expecting someone else here to have put those marble track parts together into a nice interesting marble course by now, maybe I'll wait a few more days... hopefully somebody will. :confused:
- 3dvisuals dude |
Post by frootee // Jul 3, 2007, 5:26am
|
frootee
Total Posts: 2667
|
You don't have to delete the existing fixation points.
You can move them as you wish.
I will make a couple copies of your working sample object in a single scene file. Each will have a different location for the COG and fixation points. Holding the fix points constant and moving the COG will cause the object to rotate in different ways. I expect something similar will occur if we hold the COG constant and adjust the fix points.
As for the bird, I would practice with physics by putting the COG up by the head and adding fix points to the pivot bar in the legs, to make sure it will work. It would be good to have a working physics model ready for users to download.
I was working with your gear model last night. I booleaned the two gears on the sides. One is bigger than the other of course. No wobbling occurred, which is a good thing. I suspect this is because the COG was not adjusted but rather, was kept in the center of the paddle wheel.
The other gears which interact with those on the paddle wheel would need to be scripted for rotation I suspect. That's where scripting with physics comes into play. I tried plain old physics, but when the gears ground against each other, the smallest sort of got slammed into the ground.
Instead, when the paddle wheel starts to move, we could use the watchdog timer event to start rotating the other gears.
I tried working with your marble tracks, but got ... side tracked... :cool: on your paddlewheel/gear model.
The cube with the tunnel does not work very well. The ball gets stuck. This is where testing would be a good thing. :p
But these would be good to add to your paddlewheel object.
Frootee
I was kind of expecting someone else here to have put those marble track parts together into a nice interesting marble course by now, maybe I'll wait a few more days... hopefully somebody will. :confused:
- 3dvisuals dude[/QUOTE] |
Post by 3dvisuals dude // Jul 3, 2007, 5:52am
|
3dvisuals dude
Total Posts: 1703
|
Thanks Frootee,
I'll test the cube with ball and see what's up there. Scaling should fix it but if not I'll just remodel the inner spiral helix larger in relation to the cube or fix any related diameter inconsistencies, although I did look for those when I first made it and didn't spot any on a quick glance. No sweat either way.
I've been looking into old Rube on the web and have a few related new ideas to develop for us here too. This inventory can be pretty far-fetched and still be applicable for sure! What a crazy imagination this guy had!
Stay tuned!;)
- 3dvisuals dude |
Post by frootee // Jul 3, 2007, 6:02am
|
frootee
Total Posts: 2667
|
No problem.
The ball should roll more smoothly if a smaller portion of surface area is interacting with the cube-tube surface. If several points opposite each other on the sphere are touching the cube tube, it is harder for the ball to roll; i.e., both the left and right sides of the sphere are trying to rotate in opposing directions. Unless we set the friction to zero on both the cube and sphere (I wonder what would happen if we make the sphere with 'greaseball' physics properties? :D
Frootee |
Post by 3dvisuals dude // Jul 3, 2007, 6:22am
|
3dvisuals dude
Total Posts: 1703
|
No problem.
The ball should roll more smoothly if a smaller portion of surface area is interacting with the cube-tube surface. If several points opposite each other on the sphere are touching the cube tube, it is harder for the ball to roll; i.e., both the left and right sides of the sphere are trying to rotate in opposing directions. Unless we set the friction to zero on both the cube and sphere (I wonder what would happen if we make the sphere with 'greaseball' physics properties? :D
Frootee
Excellent, thanks for the clear explanation... makes similar design work a snap in the future too. I'll ream out the spiral helix a bit more then. No sweat.
"Greaseball" properties? LOL... :D
I was just looking over "Rube Goldberg" on a Google Image Search for fun this morning and got a few very funny ideas from that stuff... you aught to do the same when you have a few minutes sometime... that guy was brilliant but seriously "whack" too!
I'm thinking of making (among many things now) a pole which looks like a tetherball pole, but instead of a rope tether it would have a very thin long pole extending from the top of the larger vertical pole downward at a 45 degree angle and have at it's lower end a magnet... it would spin upon collision between the iron marble and the magnet, lifting the marble off the track and dropping it somehwere else upon impact with some other object.
Just one idea... I'm getting hundreds after that Google Image expedition!
Stay tuned!:D
- 3dvisuals dude |
Post by Norm // Jul 3, 2007, 6:22am
|
Norm
Total Posts: 862
|
It is good to see activity here on the Rube Goldberg scenario.
Good to see you guys are getting into physics and some of the settings. Only way this will work is if we learn the settings. Take electrostatic settings for instance :) |
Post by Norm // Jul 3, 2007, 6:25am
|
Norm
Total Posts: 862
|
The objects in the collection are too difficult to use. The only way it will work is if you were to break down all the components and name them in the Link Editor ... and of course they have to "work" with physics. Would problably be good idea to only include working components. I really liked the components but sort of lost interest when I read the contents of the link editor. I especially liked the roller bearings embedded in the track. Could set them up for speed of ball so it is a great idea to work with. Opens the door to many possibilities! |
Post by frootee // Jul 3, 2007, 6:32am
|
frootee
Total Posts: 2667
|
Good to see you back Norm.
Since you mentioned electrostatic forces, that would be a good way to use the drinking bird:
the bird is positively charged, and we have two discs: one at the back of the bird, and one in the drinking glass. Both of those are positively charged.
May need to script the charge value, so, to get the bird drinking, increase the charge of the back-bird disc, while decreasing the charge of the drinking glass.
Then when the bird takes the drink, reverse the process.
I am typing this rather than doing it because I cannot install TS where I am (at work on lunch!)
BTW Norm... what would you like to see in the link editor for physics objects? I am not sure what it is you did not like...
Frootee |
Post by 3dvisuals dude // Jul 3, 2007, 7:21am
|
3dvisuals dude
Total Posts: 1703
|
Good to see you Norm!
You typed that as I was making this so here's one more that's extremely simple. Just a pivoting swing arm to magnetize in order to lift the ball and swing it away from the tracks to drop somewhere else.
Sorry about the complexity of the other stuff.
- 3dvisuals dude |
Post by frootee // Jul 3, 2007, 7:26am
|
frootee
Total Posts: 2667
|
Hey 3dv. Looks cool!
I can't download this yet so I'll just ask:
are there physics attributes already added to this? Can you run it with physics sim yet?
If not here's a good learning exercise:
1. add a single fixation point at the top of the pole, for the swing arm
2. make the wrench head at the bottom magnetic (positive)
3. make the sphere magnetic (negative)
4. Give the swing arm a 'push': add a velocity component
5. observe whether the ball gets picked up
6. If not make adjustments and repeat
cool stuff!
Frootee
Frootee |
Post by frootee // Jul 3, 2007, 8:52am
|
frootee
Total Posts: 2667
|
It's ... RUBE TUBE!
Check out this video of a Rube-style mousetrap game, created in blender:
http://www.youtube.com/watch?v=E5rv48hBgeI
Frootee |
Post by frootee // Jul 4, 2007, 5:39pm
|
frootee
Total Posts: 2667
|
Here we are... Poor Crazy Bob got totally toasted on Independence Day and decided to ride the mechanical chicken. Unfortunately he passed out just before the ride. Here's the result.
There are 2 versions of the bird. The legs get attached only after physics simulation is run and keyframes generated. The legs (i.e. hinges) and other attachable items for other objects cause interference with the phys sims.
The rar contains a scene file and an avi of crazy bob busting his a** on the mechanical chicken.
Thanks to 3dv for the bird parts!
Study the physics items and experiment with it.
Have Fun!
Frootee |
Post by ProfessorKhaos // Jul 5, 2007, 7:14am
|
ProfessorKhaos
Total Posts: 622
|
Am going to try and re-engage in this project. Looks very interesting so far. Time has been limited a bit as of late but we'll see what I can do too. :)
P.K. |
Post by frootee // Jul 5, 2007, 7:20am
|
frootee
Total Posts: 2667
|
Excellent PK! Do you plan to port your roller coaster script? That would be so totally cool!
I have not meddled with scripting yet but have some thoughts. For example, the electrostatic button, when checked, causes discharge to occur. It would be interesting to recharge the charge value on disippation, using the Sin / Cos / Tan math functions or other methods. That would allow for a pseudo-virtual machine, say, a pendulum, or grandfather clock.
Frootee |
Post by ProfessorKhaos // Jul 5, 2007, 7:40am
|
ProfessorKhaos
Total Posts: 622
|
I hope so. Will take me a bit of time to review just to remember what I've written. Also, modelling a mesh directly via VBScript or Jscript isn't quite as easy as it was with mesher which hid some of the details at the triangle level. I'll see what I can do with it. :) |
Post by frootee // Jul 5, 2007, 7:44am
|
frootee
Total Posts: 2667
|
mesher was a plugin right? The SDK is not released yet but perhaps the SDK would make it easier (For the devs: HINT HINT HINT)
Frootee |
Post by ProfessorKhaos // Jul 5, 2007, 7:52am
|
ProfessorKhaos
Total Posts: 622
|
No problems with the SDK. Mesher was Lars' baby and it was an awesome addition to Python. Not really a SDK per se. More of a plugin to Python that interfaced with the old trueSpace SDK and added a few extra commands to python itself.
Similar capabilities are already present in the current scripting language but at a slightly lower level. I have to build up polygons from triangles myself. Not terribly difficult when the original polys were quads but I will have to review the code.
Have done this before when creating the cable object in the Base objects directory. If you look within there's a rope mesh generator brick who's job is to create the mesh from various inputs. This would be a similar level of effort. |
Post by frootee // Jul 5, 2007, 7:58am
|
frootee
Total Posts: 2667
|
Oh that's right. I forgot that the mesher was a python item.
I'll look at the cable example tonight. Sounds good!
Frootee |
|
