Bullet Observations, Questions

[DrStrik9]

I agree it's simple to use, but the motions are slow and I can't seem to get any bounce. In Core the controls were easier and the effects more noticeable. I've been playing around with the controls but don't see a great deal of variation.

Settings seem to be interrelated. For instance, changes in Glue Factor won't show much of a difference if the object weight is too low.

My big question is on Envelopes. They should work, but they don't seem to. I'd like confirmation on this.

[Dreamer]

In the content download there are some bullet scenes that use envelopes. I've looked at them, but haven't delved too deep.

[DrStrik9]

In the content download there are some bullet scenes that use envelopes. I've looked at them, but haven't delved too deep.

I need to look closely at those.

The other thing I'm discovering (the hard way, as usual) is that size DOES matter. I tried to do some breaking cocktail glasses at actual size (glasses about 8" tall.) But it never really looked right. I changed the glass size to half a meter tall, and the sim looks great. Live and learn.

"It's buggy until I understand it." But it does make me wonder if it would be possible to adjust Bullet to accommodate a wider range of object sizes. (I have no idea what I'm asking here, from a developer perspective.) There are a lot of situations where modeling everything at say 24 times normal size (like my cocktail glasses) would be problematic in other ways than to accommodate Bullet's size limitations. Just wondering ...

[Dreamer]

This is what is especially confusing. I left all the controls default and just changed the collision margin to 1mm for the ground plane, 0 for the little pill. I then changed the ovoid to 3mm, 8mm and 9mm. The results are baffling to me. Why does changing the ovoid so little cause such big differences? And why does the the bounce change so greatly between 8mm and 9mm? The objects are all within the size range indicated in the manual.

[lwanmtr]

Yeah, I've had to play around alot and sometimes dont get the results im expecting

[Dreamer]

I'd like to see some better documentation on the controls. Having said that, there is no doubt Bullet is way cool.

[Dreamer]

Now that Newtek posted some very helpful tutorials I am making better sense of how Bullet works.

[jeric_synergy]

Now that Newtek posted some very helpful tutorials I am making better sense of how Bullet works.

Links?

[Dreamer]

http://www.youtube.com/watch?v=FjeTQ...bcFAAAAAAAAKAA

[3djock]

I agree it's simple to use, but the motions are slow and I can't seem to get any bounce. In Core the controls were easier and the effects more noticeable. I've been playing around with the controls but don't see a great deal of variation.

Don't use the the play button it only runs slower, use preview> make preview, it makes things alot easier to see and faster to test. Drstrike9 you are right size does matter because it calculates physics of actual size.

[DrStrik9]

Don't use the the play button it only runs slower, use preview> make preview, it makes things alot easier to see and faster to test. Drstrike9 you are right size does matter because it calculates physics of actual size.

Ironically, after thinking the same thing at first, I'm back to using the play button. You can stop it at any time and do a preview up to that point with no penalty. Overall it's a faster workflow to just set it up and hit play ... then wait. Then do a quick preview.

Yes, I now model everything at size, then if it's going to Bullet, it gets enlarged to between .4m and 10m. I wasted SO much time finding out the docs' size recommendation is trustworthy.

[marcus staffel]

Yes it's embarrassing how easy you can manipulate scenes that force you to recalculate, but you know that they don't need to be recalculated.

Example:
I have some domino-pieces that are strictly sequencial in falling. I use the dominos to form textcharacters/words. Character after character. As my camera moves around, I need the not fallen characters as background. They do not need bulletdynamics until they are going to fall. The same belongs to characters fallen 3 seconds before. They don't move any more.

I would like to freeze already fallen characters with keyframe to have one background and the not fallen other characters as another background depending on cameraangle. In total I wan't to see fallen and not fallen characters in one picture.

But just the moment I switch off dynamics or switch on other characters the hold scene has to be calculated again. But I know, that this situation does not need recalculation, because only a timespace of about 5 secounds is needed to calculate the spot of action. This would downsize the number of bullets/dominos to be calculated drastically. Now I have a 2 minute flight over 2000 falling dominos, that should calculate in a fraction of the now needed time.

As I want to integrate small sequences of a domino flying to the next character (like a rocket) I have to know the timepoint when dominoes reach the end of an character to inject this little sequence. So I have to start many times over to see when bullets/dominoes are at the point of leaving one character to the next one.

As I want to follow the falling pieces with my camera I have to recalculate many times to follow the position of action in time.

I need a freeze function to freeze everything but the action spot. This frozen all object-keyframes should be repositionable on timeline without calculating again. So if you know what you do (what fractals of the timeline do not change) you should be able to reposition them without new calculation.

Many times there are no changes in an area at a span of time. This could be extension of videocompression not only 2 dimensions but 3 dimensions (x/y/z) plus time. Todays harddiskarrays should be able to handle this data in lossless compressed manner.