Animation idea for bookshelf interactions

[Noisycricket]

I'd really like to do some AAA quality stuff with bookshelves. I've come up with an idea that will require some coding support me thinks, but i think it will really be worth it. So here an idea.

 

I could make 3 animations for characters taking books from three heights on the bookshelf. Low, middle and high. Before those animations play, there could be a general animation for looking at the books, where the character slowly panned around looking at various books. The key is to make the character look in the direction of the subsequent [take book] animation. So, in other words, the "browse bookshelf" animation would play, the character would eventually look downward near frame 280 and at 298 the "grab lowest book" would play and the character would bend down at take the book from the lowest shelf.

 

So the "browse books" code would randomly choose between three when it played as well as how long the character would browse before it took a book. Hopefully the mapper could set the "time to pickup", if need be for scripted sequences.

 

Now after that, the character could take the book with him, or opens the book there and read, turning a page once per cycle. There is already a generic animation for the right hand reaching out and taking something, so the reading book animation wouldn't have to take place at a shelf.

 

The book would have to have a 3d model and skeleton, something i could easily do in Max, but not Maya, so that would have to be done by someone else. I'd animate the character to open the book.

 

 

Thoughts? Anyone want to commit to anything? I could post in the public forums too about helping to make the book.

[greebo]

Can you whip up a rough/quick animation to demonstrate how something like that could look like? Does the book actually need to be a rigged model, do you want to make it openable?

 

Sounds interesting in any case.

[Noisycricket]

Yeah, the book model would be openable, and if it had 2 bones I could control, i could synchronize the hand to the book movements nicely. The model itself should be of average size of course. If the character sees the player, a standard model book could drop down unless its fine for the book to coexist within the world. I'll get some basic animations going.

[Springheel]

I can't think of an existing case where an animated object is bound to another animated object, so you'd be heading into new territory there. The reaching out to grab a book part would be pretty straightforward--the existing attach/drop code would suffice for that, once it's working reliably. But syncing up the animation of the book and the AI would definitely require some code support.

[Noisycricket]

How about having the ai grab a regular non-animated book model and attaching the animated book model to it? Keeping the regular book invisible until needed, or deleted if the ai puts the book back.

[Noisycricket]

Ok, this is taking awhile to figure out, i've also got some RL activities going on, but heres some thoughts thus far. I don't think we need any code to fire off animations at specific frames. As long as the take-book animations can start at the end/beginning of the browse-shelf cycle animation, it will be fine. Which would also make each of the take-book animations look good when played alone. I can just incorporate a glance toward the intended direction within the animation. Anyway, almost done with a rough example. Rough being the keyword.

[Noisycricket]

Heres some basic key poses to give some idea. The beginning is the browse-shelf cycle, it plays just once. Then a grab book animation plays, but it is only the main poses that are keyed in and the tangents are currently set to step, so it jumps from pose to pose. But id like to make it as good as mocap, at least eventually. Anyway, programmers, let me know if you'll commit to implementing this, or if you have any ideas or questions. I'd honestly rather animate some smaller things like hammering and some smithing activities, so feel free to take your time thinking about it.

 

I should mention that this was done quickly and without any reference footage (one should never do that btw), its very basic, the browse_shelf animation has only the head an pelvis animated, everything would have a little animation in the final.

 

 

[Springheel]

I'm not a coder, but a limited version of that behaviour could already be implemented without any special coding.

 

The entire browse, grab book, read and put away would be one animation. The mapper would use an animation pathnode set next to the bookshelf to activate it on the AI. Existing attachment code could spawn a (static) book in the AI's hand, and sound effects could be used to accompany the turning of pages. The downsides would be that the book would teleport in and out of the hand, and the pages wouldn't actually turn...additionally the AI would read for the same duration each time.

 

Ideally what we need for this behaviour is something that I've already put in a request for, and that is something like "sitting" behaviour--code that allows an AI to enter a new animation state, stay in it for a set amount of time playing some random animations, and then exit it.

(don't think I ever specifically added it to the tracker before, but it's there now--http://bugs.angua.at/view.php?id=2466)

[Noisycricket]

Ideally what we need for this behavior is something that I've already put in a request for, and that is something like "sitting" behavior--code that allows an AI to enter a new animation state, stay in it for a set amount of time playing some random animations, and then exit it.

(don't think I ever specifically added it to the tracker before, but it's there now--http://bugs.angua.at/view.php?id=2466)

 

I was thinking this exactly. A sort of activity state manager, which might work well with some animations i was going to ask about doing, which involved workers moving crates and passing crates from one to another as well as covertly passing items, like money or a payoff... I was also willing to create some variation of the book shelf interaction animations so they didn't all look the same, but that would require a different branch of animations?

 

In summary, it would nice if the Ai could browse the book shelf [and maybe not take anything], take a book and either read it there standing and then take another book, or take the book back to a table and read it there. (A few minutes of programming at the very very most)