My instinct was to wait until I had a lot more of these done, but, I figured that it would be silly if there was a fundamental error in the more basic stuff which ruined everything more complex that followed. So, I'll post the first round of really basic ball bounces that I've been messing with. The goal is to work my way through basic exercises in an attempt to improve. Constructive critiques are encouraged (I am trying to get better)...thanks in advance for any pointers.

This set of four starts with a ball drop without any squash and stretch, then a small amount of lateral movement is added, then squash and stretch is added and finally some roll is added.

Once I'm pretty sure everything in these is working, I'll kick out another set.

The ZIP file includes the model, Actions, Choreographies and shaded wireframe renders (everything is animated at 24 frames per second)...my thinking is that it will make it easier to find problems with what I'm doing.

Hi David,

Here are some notes...


Click to view attachment

Very helpful, Robert! Thanks for taking the time to critique this. Animating is a weak spot for me and this illustrates how much work I need to do. I'll re-tweak these and update the post tonight.

For the Bouncing ball with squetch exercises, I'd also suggest something on the scale of a soccer ball being dropped from about head height (~6 feet). Then the ball is big enough for us to think about squetch on a nuanced level.

Is there a way to zoom while maintaining a 1:1 aspect ratio in the Timeline? I can get close using the grid as a guide, but I'm thinking there is probably an easier and more accurate way.

I ran out of time this evening, so I'll have to post the update for this tomorrow night. Sorry for the delay.

Not that I know of. There really is no 1:1 in the timeline since the axis units are different.

No problem. I'll probably be en vacances by then, so I may not get to looking at it until next week.

Probably for the best, I'm thinking I should spend a few days on this anyway. Hopefully, it will mean a better finished exercise...we'll see.

Have a great Thanksgiving, Robert!

David,
After staring at various Bouncing Ball exercises for the past few years I've come to something of a conclusion regarding them. Namely, that in order to get a good result one must start with a good beginning. What I mean by that is that usually it's not enough to tell someone to Bounce a Ball in animation... you'll get as many variations on that Bouncing Ball as there are people bouncing balls at you. It will always help to state up front what kind of ball it is that is supposed to be bouncing.

- A bowling ball
- A tennis ball
- A ping pong ball
- An intelligent ball
- An angry ball
- A ball that thinks he's a rabbit.
Etc. etc.

Without this motivating factor (implied weight, force, intent or intellect) we are left with pure physics without all the required inputs to be measured (if we fail to give them) and (forgive me here) if that is all that we are bouncing balls for then we might as well use Expressions or Newton Physics and let the computer bounce them for us.

This isn't to say bouncing balls isn't fun. It's a heck of a lot of fun and it's always educational. But I don't think subtle movement and reciting physics isn't going to get us to the goal in this (unless you are in fact teaching Physics with this). I never get tired of seeing variations on these bouncing balls (so I'm really enjoying your Project) but the best of the best and the most useful always have a clear purpose behind them.

I played with several of your Actions and thought I had fairly good success. Then I played with another (The no squetch distance one) and the darned ball started bouncing against A:M's Action window!: Too much fun this is!

Those differences will be in subsequent exercises, Rodney. However, the first exercise has already demonstrated that I need to get a handle on the most basic stuff...general physics as it would appear in the Timeline window F-curves. The next exercise will use a soccer ball, but part of it will use a more extreme squetch than what would be typical so that I can get a better handle on the F-curves of squetching an object...it won't be a real-world soccer ball in every part of the exercise.

I'll call my shots on the subsequent exercises so that it is clear what I'm trying to accomplish.

Sounds great to me David.
I appreciate your effort to do this as sooner or later every animator runs into the Bouncing Ball.

What I did (and still do) is that every time I see an interesting take on the Bouncing Ball exercise on the internet I grab it and save it to my harddrive. If I was more organized I'd have hundreds upon hundreds of examples to share with you.

Looking at it from another perspective... every animator has at least one Bouncing Ball exercise in them... those who are apt to teach animation usually will have a collection of fun examples to share with students. To some extent these examples provide the artist's/animator's signature while demonstrating their approach to the Principles of Animation.

Of course, the most important aspect of the Bouncing Ball exercises is TO DO them!
Something that looks like it will work on paper might not work quite as well as expected in execution.

Here's the most important thing regarding the Bouncing Ball exercises:
In my experience there is one common problem aspiring animators run into with the Bouncing Ball. They get bogged down in the details of it too early. When it is a lesson that they should be... will be... frequently returning to whether they want to or not.

On a completely different note. Something to consider when putting together exercises for CG is the number of keyframes used. Why? So that the exercises/setups are easily readable and editable. See example below:

(All this is... for what it is worth)

Ouch. I just realized I left off perhaps the most important example in the above arcs... the one where there is no Breakdown but the entire motion is dictated by the Vector(s) of the Arc! To my way of thinking this one represents the essence of Path animation.

Update/Correction attached:

This'll seem silly...

I just used the green jumping ball emoticon [ ] in another post and a question immediately popped into my head, "Why is this bouncing ball animation so successful?"



It's got Clarity. (it conveys the theme of 'excitement', hence it's filename 'yay.gif')
It's got Squash (on contact).
It's got Stretch (both to and from contact).
It's got Eyes (so we recognize it as a sentient Character).
It's got Secondary Action, Overlapping Action and Follow Through (in both the Eyes and Hands).
It's got Symmetry (which works to augment the character's movement physically and emotionally. Note: Usually our goal is Asymmetry).
It makes use of Negative and Positive Space/Changing Shapes (Negative space is further enhanced by the layering of transparency).
It's upward Progression works as an Anticipation and conveyor of a sense of propelling Energy
It does this all in only eight (8) frames (because they are cycling frames we can easily focus until the message is received without a chance of missing anything).
Note: There is not enough time for (effectively) blinking.

All of this proof positive that it takes so very little to entertain me.

Nice analysis, Rodney!

YES - "It's got silly" & "It's got entertaining" tooo! That's more reasons why works so well.

I agree: Excellent analysis! Thanks for providing a good checklist for any animation. I'm filing that post to my "animation tips" folder.

Who'd of thought such a great animation example was under our very noses all this time!

Hope I'm not spending too much time in your topic David... if I am let me know and I'll post this stuff over in the "Drawn to CG" WIP.

While considering the Bouncing Ball I was surfing through sites looking for good timing charts when I happened upon a university site with a pretty decent animation curriculum. One little graphic grabbed my attention but I felt it could be expanded on to encompass more animation principles. Every time I thought I was finished with the graph I discovered something else missing... and it still could use a lot of tweaking. I then got distracted for entirely too long adding some uselessly strange weathering effects to the image.

Pending feedback, I've settled on this chart as something of an introductory for setting the stage to the whole idea of ball bouncing with the added element of Squash and Stretch. Looking at it now I really am tempted to change each of the arrows so they are pointing outward on both ends but for now let's settle for that indicating the direction of where the object came from or the source of forces causing a change in shape. I'm also a little bothered by the fact that I didn't get all the terminology I wanted into the graph. For instance, there are many words that should be incorporated along with their opposites; expand/contract, compression/extension, push/pull, etc. Of course trying to incorporate too much will just make a mess.

Please let me know if it makes sense!

There is a site out there that has a whole lot of animation made in the early days of computer games. Some of those little animations are especially cool because they use only a few pixels only four (4) frames! (I believe the site in question included a historical look at the game 'Lemmings' because the artist responsible won a bet that he could animate them successfully in those four frames)

Okay, let's see if I can do better than the first attempt.

I made a modification to the squetch ball to make adjusting the direction of the squetch easier (for me at least)...I think it helped. If it survives a bunch of testing, I'll be adding it to the Squetch Rig.

To "call the shot" I made a quick set of notes that include captures of the timeline and the animations in the "bouncing_ball_exercises_01v2_notes.zip". The squetch ball, Choreographies and another copy of the renders is in the "bouncing_ball_exercises_01v2.zip".

I've got the second exercise partially completed, but I think I'd better wait for the critique of this one before I go farther with it.

Constructive critiques are encouraged. Thanks in advance for any pointers.



--------------------------
EDIT
--------------------------

I realized after posting this that I should have cleaned up curves on the 'X' movement of the Null target. It isn't extremely obvious and I don't have time to correct it at the moment, so I'll leave it as-is unless I need to make another version.

I'm still traveling, but i willcertainly take a look when I get back home.

I'm looking at them and see a frequent problem. I'm trying to figure out how to explain it.

Try this...

How long does it take a ball to fall... x feet? There's probably an exact formula for that which we dont' need to calculate right now, but...

How does that time to fall x feet change if the ball is dropped from x feet or 5x feet or 5000x feet?

(In each case I'm referring to the time it takes the ball to drop the first x feet.)


What is your expectation of what happens in those three situations?

I'm not sure what you mean, Robert. I used the tutorial (from http://www.animationphysics.com/) that is linked in the notes to figure how fast the ball would fall...did I misunderstand the tutorial or is there something wrong with the tutorial? I used the same rate of fall for each bounce because there was no force other than gravity involved...was that inaccurate?

-------------------
EDIT
-------------------

Or maybe I just thought I had made it the same rate of fall....I'll have to check later today. I assumed a couple of things, I'm thinking.


-------------------
EDIT
-------------------

According to the tutorial, it should have been 1/4 of the distance at 1/2 the time...so, if the height was 48 inches (the initial height I used), it would be approximately 36 inches at 6 frames since the ball should take 1/2 second to hit the ground (also according to the tutorial). I adjusted this slightly, so mine was around 37 inches at 6 frames...I explained it in the notes.

This is a trick question.

If I'm understanding the question, it would be the same rate. That rate is on the chart from animationphysics.com that I linked to in the notes.

Where I possibly went wrong (I think) is that I assumed that using the exact same bias settings on each bounce would produce the same results if everything was scaled equally...that might not be the case. I didn't check it close enough is what I'm thinking.

Yes, it will always take the same time to fall x feet, no matter what height it starts at.

Now... how can we use that to inform how long each bounce takes?

I made each bounce last half as long because I thought I made each bounce half as high...unless I made a mistake in there somewhere.

When I get home this evening I'll see if I can figure out where I went wrong...I'm stumped at the moment (I can't take a look at it where I am).

There's the problem. A ball that falls from half as high does not take half as long to reach the ground. It takes a bit longer.

A ball falling from 6 feet takes longer to fall thru the first 3 feet than the second 3 feet, right? Because it was accelerating all the way thru.

Consider a ball that falls from 6 feet, bounces up and falls again from 3 feet. The new fall from 3 feet should take as long as the first fall from 6 to 3 feet.

Consider that a fall always starts falling from a velocity of zero, no matter how high or not high it has bounced.

Hmmm, I'll have to digest that a little.

I assumed that on the second bounce (and after) it would also fall 1/4 the distance at 1/2 the height and would have the same curve to the spline as the first bounce.

-------------------
EDIT
-------------------

I'll let it sink in for a little while and tackle it when I get home.

Okay, I think I get it...this post that Yves made puts it into numbers. I'll fix this tonight.

d=(gt**2)/2

where d = distance traveled
t= time (seconds)
g= gravity constant (or acceleration rate) = 32 feet/second squared

so no matter what, any object (on the earth) will travel if dropped from any height (no forward velocity), after n secs:

0 sec, distance traveled = 0 feet
1 sec, distance = 16 feet
2 sec, distance = 64 feet
3 sec, distance = 144 feet
etc

In the non-cartoon universe:

How far the object bounces after falling that height depends on the material of the object (eg rubber versus brick), the material of the object it impacts, friction, and initial forward velocity

(EDITorial: Use Newton, note the curves it produces if you want strict earthly realism. Otherwise make that bouncy-ball do a dance!)

Your bounces are not wildly in error, but as they get to the smaller bounces the time problem becomes more visible.


Here's some footage shot at 60fps. The stick I'm holding is marked off in 1 foot increments and is six feet long.

Click to view attachment

Watch me drop the ball and the brick then go back and count frames:

How long does it take the ball to fall from top to bottom? I count about 36 frames.

How long does it take to fall the first foot? About 15 frames.

Frame thru it yourself to convince yourself that is the case. I had to. I was surprised it took so long to get started. Almost half the time is taken up getting thru the first foot.

The ball also happens to bounce up one foot. How long does it take to fall back down from the peak of that bounce?

That is 1/6th the distance that it fell originally but it doesn't fall in 1/6th the time, it takes a lot longer.

One sixth the time would be about 6 frames, but the ball falls that foot in about 15 frames, the same as it took to fall one foot originally, starting from the top.


Presuming we have animated the first fall correctly, we can use the expectations it has created to time subsequent bounces.

If the ball bounces back up to one foot, it takes as long to fall back down as the ball took to fall one foot originally.

If the ball bounces back up to four feet it takes as long to fall back down as the ball took to fall four feet originally.

And how long does it take the ball to bounce up to the peak of each bounce?... As long as it takes to fall back down from that peak. They are almost always symmetrical.

Count the ball bouncing up from the ground, peaking and falling down again. The up and down might be different by a frame because the camera isn't quite catching it at exact moments of impact.


What about the brick? I did that just to show it falls almost the same as the ball. Even though it is much heavier, it might be one frame faster. Maybe air resistance slowed the ball down a tiny bit or maybe the camera is just catching them a hair differently.


Should the bounces still make parabolas in the Y channel? Yes, but they are not scaled copies of the first parabola, they are copies of the first parabola with the bottom clipped off.

I've just skimmed thru the "Physics" pdf. I like some of it and some less so. He may be trying to fit too much info too fast into one lesson.

I like the "Odd Rule".

The "Fourth Down at Half-time" is getting a poor result we can see in the second half of his graph.

I'll look at it some more.

Here's my sermon about Bouncing Ball.

It's not really about being able to make great bouncing balls. The occasions for actual bouncing balls in animation scenes are pretty few. I don't think there was a single bouncing ball in all of TWO. It's hard to think of any big studio feature that has any significant bouncing ball activity.

It IS about animating an object so that is appears to be plausibly under the influence of gravity. That happens all the time in character animation. Anytime a character steps, hops, lurches, falls, or just shifts his weight you have to imply there is gravity at work or you get that floaty, weightless look. It's exceedingly rare that we visibly caricature gravity. Mostly we try to get it very right, if we cheat it it still has to be plausible, not obviously wrong.

We use bouncing ball for our first attempts in this gravity thing because it's easy to revise and simplifies the problem down to one mass rather than the 50 or so connected masses of a character. If you can't get it right with the simple ball, you can't get it right with more complicated models.

David knows this already, that's why he's doing these bouncing ball exercises, but I say this for anyone else looking in, wondering why Bouncing Ball is such a big deal.


David, here's some further explanation of my theory of using that first curve to inform your later bounces...

Click to view attachment

That explains it very well, Robert...it clears up some things for me. Thanks for taking the time on this.

I'll do some work on this today. If I'm lucky, I'll finish another version and post it.

For the moment leave out the squetch and rotation and just do one ball bouncing in place and one that travels from left to right across the screen as it bounces and lets look at those.

Here's an interesting visualization. I've taken the dropped ball and moved it sideways which is much like plotting vertical motion versus time, which is what our channels in A:M do.

Theoretically these are near-exact parabolas being traced out. They are a bit sharper at the top and flatter on the sides than I tend to come up with just by eyeballing the shape.

Click to view attachment

I did this with "Time Echo" in After Effects.

That might be the best move, Robert. Here is what I got for those along with notes in HTML, renders, squetch ball and Choreographies.

Now don't those just... feel... better? A lot better. They dont' have the rushed look that the previous takes did.

Picky points:

If this were for a class, I'd leave the stripe off on a non-rotating ball test, just to avoid the rigid look of it being motionless. Unless they demand a stripe on their non-rotating ball.

The moving ball looks fine and slows down appropriately. Maybe it would benefit from an even more gradual slow out at the end.

You have a single continuous curve to slow the ball's horizontal motion. Straight line segments between the impact points might be more accurate (the ground can't slow the ball down while it's in the air) but not significant at this time scale.

On moving ball tests I've found it effective to start on the upswing, maybe about 1/3 from the top, as if it had been tossed up. It gives the viewer a bit of time to catch up with the ball once it starts moving. It's curve of course would be a mirror of the curve from the peak.


Try a squetchy ball next. My basic theory of squetchy balls is to treat the squetch as overlapping motion. A stripe on a squetchy ball is a problem that is difficult to treat well.

I don't really have a great eye for this yet, so I'm still getting used to it. Initially, it felt like the ball hung in the air a little long, but the more I look at ball bounces, the more correct it appears.



I added an on/off Pose for the stripe this time.



I attempted to add everything suggested, we'll see if I was successful.

Here's the next one with included notes, renders, squetch ball and choreographies:

Getting an eye for it is most of the battle. The hang time around the top is something you can play with a bit in animation. My non-pointy approximate parabolas tend to make it a bit longer than normal.

Film history story...Silent films started out being shot at fairly low frame rates in the 12-16 fps range. Theater owners found they could get in more shows by speeding up the projector a bit. Film makers tried to counter this by shooting their films at higher frame rates and theater owners countered back by speeding up the projector more.

All thru the 10's and 20's there was this frame rate arms race going on, and audiences got accustomed to seeing most motion faster than real-life but accepting that as normal. When sound came in and standardized frame rates and everything had to be shot and projected at the same speed audiences commented that it all looked like it was moving in slow motion!

Your squetchy bounce looks great!

Here are some notes:

Click to view attachment

Very cool! I'll have to figure out the next thing to work on. Thanks for taking a look at these.

The rigging on the ball allows the squetch to be independent of the rotation of the ball (although it looks like it contributed to the source of the squetch not being at the contact you mentioned...I'll have to do a little thinking about that). I didn't bother with the ball rotation in this one since the stripe was hidden. Here's a quick video showing the controls.


-------------------
EDIT
-------------------

I think if I animate the pivot point, it might get rid of the problem with the offset squetch source...I'll mess with that tonight.

At AnimationMentor the next assignment would be a ball bouncing off various walls and obstacles. The challenge is to correctly handle bounces and rolls off surfaces that are not horizontal. Also the ball had to jump off the initial surface on its own.

oooo...NICE!

love your squetchy ball model!

It's a modification of a ball model that Robert made several years ago...the direction of squetch being independent of the rotation of the geometry is what I've added.

That looks cool...I'll give something similar a shot.

Just to tie up the last bouncing ball exercise, I attempted to incorporate Robert's suggestions and get the origin of the squetch more accurate. I started the sequence at about 1/3 from the top of the arc on the initial drop, had the squetch start about 60% from the ground, moved the origin of the squetch to the center of the ball except on the contact frames and added some roll. The "aim at" constraint proved to be inaccurate in some situations (due to how the rig in the ball works), so, I just manually aimed the "pivot" bone using "Onion Skin" set to show where the ball would be on the following frame.

Here is a ZIP with the ball, Choreographies, notes and renders.

I've done some thinking about the next one...now I can do some more serious planning for it.

David,
I've been falling behind in watching your exercises.
I'll see what I can do to get back in the game.

Firstly, I thank you for posting A:M files to the forum. This is something I need to do myself.
In many ways this is more effective than seeing a rendered images... much more effective in many ways.

Right now I'm mostly looking at '02_ball_bounce_with_squetch_with_distance_update_with_roll'.
There are some fascinating aspects about this.
Most importantly is that it works. In simple terms... it reads.

There is an aspect of it that does escape me in that I think there is something a bit off on the contact frame but I'm not sure exactly what it is. My gut feel is to say that the ball appears to lose some of it's volume while reorienting itself for no apparent reason.

Here are some images that attempt to capture what is happening. Note that this attempt to analyze doesn't take anything away from the fact that it works. If anything the odd little adjustment on contact adds an interesting flavor/anticipation to the bounce that suggests foreshadows the pliability that is observed later in the animation. It is my feeling that the early squetch does not equate with the latter in terms of consistency. By this I mean to say that at the first bounce the ball seems react more (it is softer) than upon the second bounce. This may be more of an effect of the reorientation and apparent change in volume than anything?

In any case... you've given us some really fun stuff to analyze.

Here is a video that alternates odd and even frames in wireframe and shaded wireframe that when moved through frame by frame should show some of the variation in volume. You can track the rotation and squetch pretty well in this way as well. If you only have a little time to apply to the analysis my suggestion would be to concentrate most of your attention on what is happening on/near the first contact frame.

I think I know what's going on in there, Rodney...I'll have to look at it tonight to sort it out.

The moving of the pivot changes where the source of the squetch comes from, but it also moves the geometry some. I should have looked at the contact frames again to make sure everything was working correctly.

Nice catch, Rodney!