Jump to content
Double Fine Action Forums
Sign in to follow this  
DF Oliver

Programming Update #4: Animating the Jack

Recommended Posts

Sword and Sworcery uses pixelated characters and to keep that consistent they have to draw every frame, if the characters were attached to a vector rig/skeleton the pixels would from time to time turn and be angled, kind of like you see in those 8 bit game spoofs on Youtube. Even with these pixelated simple characters you could see what was going on at all times. Of course this style asks for more detailed characters, but that means more work for the animators because mistakes get more visible.

I was referring to the "still looks great in HD" bit. I know about pixels already but thanks anyways =)

Share this post


Link to post
Share on other sites

Any idea on how character movement was achieved in Rayman Origins? That game is absolutely stunning -one of the best looking games ever made without a doubt, IMO-, maybe something similar could be done here? I don't really know about the technical limitations and stuff, so I'm not sure if this is something viable.

Also, great post Oliver. I love the amount of insight you give us with each update :D

Share this post


Link to post
Share on other sites
Any idea on how character movement was achieved in Rayman Origins? That game is absolutely stunning -one of the best looking games ever made without a doubt, IMO-, maybe something similar could be done here? I don't really know the technical limitations and all the technical stuff, so I'm not sure if this is something viable. It's just a thought.

It looks like the UbiArt engine uses some form of skeletal animation, but the engine itself does a lot of the work automatically.

I found a Youtube video of the engine being used by artists (doesn't explain much, but it's the only video I could find that shows off a bit of how it works):

Share this post


Link to post
Share on other sites
Any idea on how character movement was achieved in Rayman Origins? That game is absolutely stunning -one of the best looking games ever made without a doubt, IMO-, maybe something similar could be done here? I don't really know the technical limitations and all the technical stuff, so I'm not sure if this is something viable. It's just a thought.

It looks like the UbiArt engine uses some form of skeletal animation, but the engine itself does a lot of the work automatically.

I found a Youtube video of the engine being used by artists (doesn't explain much, but it's the only video I could find that shows off a bit of how it works):

That's so cool, thanks for sharing!

Share this post


Link to post
Share on other sites
Any idea on how character movement was achieved in Rayman Origins? That game is absolutely stunning -one of the best looking games ever made without a doubt, IMO-, maybe something similar could be done here? I don't really know the technical limitations and all the technical stuff, so I'm not sure if this is something viable. It's just a thought.

It looks like the UbiArt engine uses some form of skeletal animation, but the engine itself does a lot of the work automatically.

I found a Youtube video of the engine being used by artists (doesn't explain much, but it's the only video I could find that shows off a bit of how it works):

I think read somewhere that it uses silhouettes to control the rigs... can't find it at the moment but will add it in if I find it. In any case yeah, it does look very similar to what DF are doing here.

Share this post


Link to post
Share on other sites
Any idea on how character movement was achieved in Rayman Origins? That game is absolutely stunning -one of the best looking games ever made without a doubt, IMO-, maybe something similar could be done here? I don't really know about the technical limitations and stuff, so I'm not sure if this is something viable.

Also, great post Oliver. I love the amount of insight you give us with each update :D

I assume it's a mix of this style and some hand drawn alternative poses.

Most important factor here is the Snappiness of the moves, that basically aren't smooth, but fast so they feel right. They also have very simple characters. and that also makes custom in-betweens a whole lot easier, because they don't have to trace the arcs on every square on a shirt or what ever and that will be easier for the animators to animate fast as well.

For Idle/walk/run animations they seem like they are using the rig system more, but for action moves they seem to use a lot of hand drawn poses, or at least strongly supervised.

Edit:

Also note that by keeping moves snappy with added smear frames etc. you can get away with something like 3 custom dawings per move and it will also be funnier to watch. Of course he walk cycle needs to be more natural as constantly watching the same snappy move over and over will become boring as well. But contrast is important and Rayman's style really looks great, only problem is that in order to achieve that the character's must look simpler so custom frames can be produced more quickly for it to be efficient.

I strongly believe that anyone can draw, but the detail of the final drawing will govern the time it takes to do it.

Share this post


Link to post
Share on other sites
I think read somewhere that it uses silhouettes to control the rigs... can't find it at the moment but will add it in if I find it. In any case yeah, it does look very similar to what DF are doing here.

I heard that as well, but I couldn't find any details about how that silhouette animation system worked.

I assume the silhouette system is probably used for squash and stretch to exaggerate the animations on top of the skeletal/bone animation system shown in the video.

Share this post


Link to post
Share on other sites
I think read somewhere that it uses silhouettes to control the rigs... can't find it at the moment but will add it in if I find it. In any case yeah, it does look very similar to what DF are doing here.

I heard that as well, but I couldn't find any details about how that silhouette animation system worked.

I assume the silhouette system is probably used for squash and stretch to exaggerate the animations on top of the skeletal/bone animation system shown in the video.

Yeah, that's what I assumed from the video too. It looks like the bounding boxes that they were drawing are used both to build the rig and as a lattice-like deformer. Thanks for sharing!

Share this post


Link to post
Share on other sites

Once again, a great post Oliver (I wish I had've had time to read it sooner). Thanks for sharing :)

Share this post


Link to post
Share on other sites

Yeah, that's what I assumed from the video too. It looks like the bounding boxes that they were drawing are used both to build the rig and as a lattice-like deformer. Thanks for sharing!

Yeah, I think you are on to something there.

And it works great for anything that is basically looking in the same direction all the time,

turning it, (unless you do it with a back flip and have it upside down,) needs some redrawing;

Sure we can flip the character and have the same animation, but the transition needs some custom drawings and the amount depends a bit on the style of animation.

For realistic smooth animation you need several drawings or it will feel very poppy.

If you have a more snappy style you might do with, 1 anticipation drawing, 1 breakdown and 1 overshoot drawing. and you can go back to the cycle flipped, giving you snappy movement, (that works great with interactivity,) and also makes it look good and simple.

There is a problem to this though and that is if you want this snappy transition and in general have everything moving smoothly, because then it will feel like a mistake rather than a natural move this character could make.

This isn't to say that this move has to be done in 5 frames, (in real time counting the end and start frame of the cycle,) but it can be timed out so maybe the anticipation holds for 5/24 of a second, the breakdown 2/24 and the overshoot 3/24 of a second, (in the case of less than a half a second turn.)

The game can still run in a gazillion frames per second on your computer/device, but it wouldn't make a difference on what you actually see.

If you have ever played a well animated game with low fps (something like 12) you can still play it fairly well, because you can read based on the pictures where the targets are going to be and as long as you can aim freely it's controllable. (Not the case in games where Multiplayer or Non-scripted AI comes into play.) But in an adventure these two elements don't really happen that often.

Rayman is also a great example because the characters in that game always act the way you expect them to, more or less. So there they really get to tailor the movement, but what actually kept me from from enjoying Rayman was the pacing of the gameplay; too stressful for me, I couldn't just watch the game and enjoy its great aesthetics the game required me to react. Now if Double Fine gets the mechanics driving the aesthetic part covered I really think they are making one of my favorite games of all time; pacing has already been discussed in a video and it seems like they have that covered, humor will be awesome, as with all DF games and story has never been disappointing with theses guys.

That said instead of getting a really long game I'd rather have a short and perfect game, that I could play over and over every once in a while, than a stretched out game that never really amazes me in any way, but I play it because I waited for so long for it to come out.

In the latter case I might never pick up the game again, but if it's a 2-3 hour game I would probably love it to bits and play it several times a year, because it's not bad and wont take me forever to finish. (this time is as if you knew the answer to every puzzle.)

Game can still be filled with a lot of side characters and unimportant stuff to interact with that would lengthen the experience if you wanted it.

If the game is really well made it might be easier to attract people to it as well, if it looks generic new people wont bother picking it up at all. I really want a game I can easily present to my non adventure playing friends and have them immersed in that style of gameplay.

Coming back to old-school to me it means to get the best interactive audio-visual experience available at the current time, back in the early 90s graphic adventures looked so much better than the general action games, because they didn't have to pump out high fps and could focus on storytelling instead of guts and gore. It's a bigger challenge now as you have epic action adventure games around every corner. I really like when they in some of the adventures implemented a double click at the exit to have your character just pop to the next room. That adds to re-playablility as well because there's not going to be anybody coming out guns basing and having to watch your character do the same walk cycle form one edge of the screen to the next for the nth time really isn't why I play these games.

I know that the Cave is going to solve that issue, by not having a backtracking ability, but I've always been fine with that as long as I don't have to wait for it to happen. If anybody is familiar with the Inspector Hector series I like the way they made the last one snappier by implementing a map where you basically could jump from one spot to the next very fast without having to walk through all the rooms if you wanted to go from one edge to the other.

Share this post


Link to post
Share on other sites

Nice update!

Have you guys given any thought to bulging the geo out a bit in Maya instead of having it all completely flat? I'm not talking about a full 3D character, just pulling the middle vertices forward a bit to round it out some. This might give a bit more depth illusion, adding a bit of parallax distortion as the character moves, and it could also help with dynamic lighting.

Also, are you guys planning to do any blending between animations, or to overlay animations (blending between skeletal animations I mean obviously, not between flip-book sprites)? Are you having the animators control the animation state-machine, or are you planning on hard coding it?

Share this post


Link to post
Share on other sites
Nice update!

Have you guys given any thought to bulging the geo out a bit in Maya instead of having it all completely flat? I'm not talking about a full 3D character, just pulling the middle vertices forward a bit to round it out some. This might give a bit more depth illusion, adding a bit of parallax distortion as the character moves, and it could also help with dynamic lighting.

Also, are you guys planning to do any blending between animations, or to overlay animations (blending between skeletal animations I mean obviously, not between flip-book sprites)? Are you having the animators control the animation state-machine, or are you planning on hard coding it?

Bulging out to make parallax...

It sounds like a great idea in theory; you would get something closer to a desired light blending and then you could add/remove/work with it after, but it might not be the best choice for a game; will make it necessarily to make the game in a 3D camera/perspective for it to work properly, as opposed to now, they can basically just scale/and move the character in a orthographic camera. they would have to supervise more curves; control the Z dept curves, as well as Rotation on the X axis, multiply that with every limb and controllers that would make it pointless to have the character in 2D non the less, because it would be faster to draw it manually. They would in that case be better off actually modeling the entire character.

It will also make it more complicated in terms of making custom frames; not only do they have to trace a new silhouette they would also have to model up the custom frames as well.

All their planes of set artwork would have to be painted for the perspective expected, and Bagel would have to pain in 3D instead of whatever technique he is used to, because the environment would feel very flat in a perspective camera.

I think the idea is to treat this whole thing as sells as if it was from actual 2D animation.

If they move the character in Z space as apposed to scale him, that won't really happen, because then it would ultimately be necessary to calculate the distance between each plane and tilted to fit the perspective and so on. Yeah sure the character in most 2D adventures don't feel like he/she steps perfectly on the ground, (for that you need to animate every frame manually or do it in real 3D,) but it will look a whole lot more like a paper cutout figure in a theater by this technique.

As far as I understand it the final product should be as close as a moving/interactive Bagel painting as possible.

I think in theory your idea is great, and it might be possible to pull some of it out to get affected by a rim light orthographically, but the risk for making it overly complicated is very big, because as long as the char is flat, for rigging as well will be soooo much easier, just think of how it would look if the rig exploded in Z space. that would be avoided instantly by just not having Z space involved with the rig at all.

Of course this is my opinion, great idea though, keep em coming :D

Share this post


Link to post
Share on other sites

Great update Oliver!

I liked how the animation was done and I enjoyed the maths at the end.

Share this post


Link to post
Share on other sites
Anyway, SpaceVenture will have nice high-res graphics, but it will be very old school in every other respect, including having verbs and even a text parser, as well as the ability to die. I expect it to appeal less to the mass audience than it does to hardcore fans of late 80s/early 90s adventure gaming.

Is there a confirmation about text parser? I thought it was just an idea. Tying together verbs (five action icons in fact), text parser and touch interface sounds like an extremely difficult task to me. Also death is just for fun, it is possible to continue playing from the last action. As far as I see now SpaceVenture is slightly more old school than DFA and much more modern than Quest for Infamy.

Share this post


Link to post
Share on other sites

If hand-drawn is so hard, what is the system that Daedalic uses for their incredibly pretty adventure games? Do they use skeletal animation + "flipbook" animation hybrid? What does Rayman Origins or Vanillaware use? Like, is Ubi-Art Framework also skeletal animation?

ibsPgOcOKhwDfK.gif

tww_screenshot_013.jpg

Share this post


Link to post
Share on other sites
If hand-drawn is so hard, what is the system that Daedalic uses for their incredibly pretty adventure games? Do they use skeletal animation + "flipbook" animation hybrid? What does Rayman Origins or Vanillaware use? Like, is Ubi-Art Framework also skeletal animation?

Ubi-Art Framework uses skeletal animation in addition with real-time Bezier tessellation (for squish & stretch) as far as I know.

Share this post


Link to post
Share on other sites
If hand-drawn is so hard, what is the system that Daedalic uses for their incredibly pretty adventure games? Do they use skeletal animation + "flipbook" animation hybrid? What does Rayman Origins or Vanillaware use? Like, is Ubi-Art Framework also skeletal animation?

Ubi-Art Framework uses skeletal animation in addition with real-time Bezier tessellation (for squish & stretch) as far as I know.

lol "squish and stretch" tech. Thanks for the info!

Share this post


Link to post
Share on other sites

Great post! Lots of cool stuff. Now for a few (lol) questions:

So the silhouette of the characters will be defined by vertices? And here I thought 3D artists wouldn't be useful on a 2D game project!

Does this mean an anti-aliasing solution will be implemented? What about the backgrounds? Are the layers defined by meshes as well, or do they simply use an alpha channel?

Also, is the character geometry completely planar? Would he disappear like Paper Mario if we saw him from a side view? How does the overlap of the arms and legs work? Are some parts, like his arms, separated from the main body? Or does the system have some other way of knowing what parts should overlap the others?

Share this post


Link to post
Share on other sites

I just saw this post thanks to the backer email sent out recently. I love love love love this style, and (as an animator, too) see it as a totally awesome and valid aesthetic choice and not some sort of "boobie prize" we get because "traditional animation"* is too awesome and expensive for us.

I'm all for a game with an animation style that evokes the likes of Yuri Norstein, arguably the greatest animator in history, rather than one that apes the same cel-animated style used in a million generic TV cartoons and insipid Disney movies. This demo footage already looks a million times better than the Whispered World, which was an object lesson in how cel animation doesn't magically keep a game from looking bland and soulless and being populated with irritating and ugly characters with awkward, choppy movements.

I am so excited to see this style further refined, and think it's an awesome direction to go in. Thanks so much for a window into this process. (And, incidentally, a look at Yuri Norstein's character-based stuff like Hedgehog in the Fog or Tale of Tales could provide the DF animators with visual solutions for making a "cut out" figure appear to move and rotate in space - though he obviously didn't have rigs to wrangle).

*"traditional animation" is kind of a daft misnomer for cel animation, which was in fact predated by various forms of stop-motion, puppetry, and cut-out animation in trick films and so on, which are arguably more "traditional." Better terms than "traditional" for describing cel animation could include "commercial," "ubiquitous," or "generic-looking."

Share this post


Link to post
Share on other sites
If hand-drawn is so hard, what is the system that Daedalic uses for their incredibly pretty adventure games? Do they use skeletal animation + "flipbook" animation hybrid? What does Rayman Origins or Vanillaware use? Like, is Ubi-Art Framework also skeletal animation?

Ubi-Art Framework uses skeletal animation in addition with real-time Bezier tessellation (for squish & stretch) as far as I know.

lol "squish and stretch" tech. Thanks for the info!

Thanks for pointing out that it is called Squash and Stretch animation.

Share this post


Link to post
Share on other sites
Great post! Lots of cool stuff. Now for a few (lol) questions:

So the silhouette of the characters will be defined by vertices? And here I thought 3D artists wouldn't be useful on a 2D game project!

Does this mean an anti-aliasing solution will be implemented? What about the backgrounds? Are the layers defined by meshes as well, or do they simply use an alpha channel?

Also, is the character geometry completely planar? Would he disappear like Paper Mario if we saw him from a side view? How does the overlap of the arms and legs work? Are some parts, like his arms, separated from the main body? Or does the system have some other way of knowing what parts should overlap the others?

That is a great question!

Yes we'll most likely need some kind of anti aliasing solution.

As far as the environment goes the polygons will not be very tight, we really just want to minimize the overdraw. The textures will still contain transparency to blend nicely with the layers behind it.

We are still investigating whether or not the character meshes will be completely flat or slightly curved. There are benefits for both techniques. I'm sure we'll talk a bit more about that later on. :-)

Share this post


Link to post
Share on other sites
cool. :)

just curious how the animation system would handle transitions between different versions of the character (e.g. when the character is drawn from a different perspective to walk toward or away from the screen). Are the "puppets" just swapped out between two frames or would you use a number of non-skeletal animation frames for the transition or some kind of blending magic?

ps

tu-dresden shoutout

(couldnt resist :))

That is a very good question. Right now we are simply replacing the rig, so there is a pop (blending wouldn't really work in this case). The animators will have to figure out whether or not that looks good enough. We could try to connect transitions like these with a special animation, but that might be a lot of work.

I'm curious why blending would not work in such case. There is interesting library which you could probably use - cal3D

Share this post


Link to post
Share on other sites
If hand-drawn is so hard, what is the system that Daedalic uses for their incredibly pretty adventure games? Do they use skeletal animation + "flipbook" animation hybrid? What does Rayman Origins or Vanillaware use? Like, is Ubi-Art Framework also skeletal animation?

On a technical level, Daedelic games use flipbook animation, but as for how that art is CREATED, it varies from game to game. Deponia uses flash-style segmented animation, and it generally looks fine in the game (but bad in the cut scenes, which suggests to me that part of the problem is, in fact, just abuse of tweening to get too much fluidity more than the entire technique).

I'm all for a game with an animation style that evokes the likes of Yuri Norstein, arguably the greatest animator in history, rather than one that apes the same cel-animated style used in a million generic TV cartoons and insipid Disney movies. This demo footage already looks a million times better than the Whispered World, which was an object lesson in how cel animation doesn't magically keep a game from looking bland and soulless and being populated with irritating and ugly characters with awkward, choppy movements.

I am so excited to see this style further refined, and think it's an awesome direction to go in. Thanks so much for a window into this process. (And, incidentally, a look at Yuri Norstein's character-based stuff like Hedgehog in the Fog or Tale of Tales could provide the DF animators with visual solutions for making a "cut out" figure appear to move and rotate in space - though he obviously didn't have rigs to wrangle).

Norstein places every frame by hand, though. Which, whether they use skeletal animation or not, is what they should be doing here. I'm not necessarily saying lock this at 15 fps or anything, but rather animate the frames by hand at a rate that's appropriate for each animation and lay off the super smooth tweening because it doesn't look good.

If you're going to commit to using skeletal or "cut out" animation. Norstein is certainly the best reference you could use. I'm glad I'm not the only fan of his around her.

Share this post


Link to post
Share on other sites

On the subject of dropping frames to make the animation look hand-drawn I'd like to present Galactic Mail as proof that it works wonderfully:

Share this post


Link to post
Share on other sites
On the subject of dropping frames to make the animation look hand-drawn I'd like to present Galactic Mail as proof that it works wonderfully:

Wow that looks really cool! I think Pocoyo does a great job of this too.

A lot of the animation in Pocoyo is at 12fps and this makes it feel like stop-motion and hand-drawn animation

Share this post


Link to post
Share on other sites

I LOVE pocoyo! but I think that that traditional animation feel has more to do with strong poses and timing than framedropping across the board. of course, having too minute movement between frames, especially over a duration of time, is a dead give-away for being interpolation and generally looks bad. having really snappy timing and strong holds is one good way of doing it!

it would be cool if DFA could have some of that snap for sure! and it makes sense with the parts/hands/expression swapping. but oliver did say that they dont expect to do too much painting of unique frames....but heres to hoping that they can find time to. just having some basic different angles of head and feet and hands to pop into walks and other generic animation would go a long way I think.

I hadnt seen galactic mail before and though the animation looks good I dont see how its strengthened by low framerate generally. it still looks like cg animation. it could have been full frame just as well...?

Share this post


Link to post
Share on other sites

One of the biggest problems in CGI is the skinning, not only the way the character is moved generically and smoothly, in CGI we tend to rig in a T-Pose, and the character is designed to look at it's best while modeled in this T-Pose. The problem with this is that once we start deforming the character; lowering the shoulders, bending knees and so on. all the crisp angles that were in the original design has a tendency to be very rounded out. Look out for Disney's new short, Paperman, (Paired with Wreck it Ralph,) where they are trying to solve this problem. The basic idea of it is that they try to get all the performance out of (3D)CGI animators as they possibly can by having them animate in CGI first, then The 2D animators are set to draw new edge-lines around the characters, that will track the underlying 3D. From what I've seen it's the closest to traditional anybody has come. And I can't wait to see the whole thing.

Galactic Mail was pretty cool, but didn't look cooler because they reduced the framerate. They generally had some good animation in there. The only thing you solve by reducing framerate is to prevent yourself from having to clean up all the frames of a shot, I personally think that should be something you decide on a shot to shot basis. Like the shot in Galactic Mail, where the camera is zoomed far out and the guy is diving through space... The only thing going on in my head was that this looks really choppy. Even at Disney back in the 30s they would still record that in 24 individual frames/sec, they would probably not change the drawing on every single frame, but they would at least try to space out the position of the character to feel less choppy. The only place I find cutting down the amount of frames really can help is when the character isn't moving around that much in the first place. You do different things generally in different situations; Pocoyo doesn't look traditional, but it's really well timed out and really feels great to watch. All their poses are really well planned out so you don't have to have X amount of inbetweens, because all the frames show where things are heading.

Frogacuda, I couldn't agree with more! it is fine to animate on 3s 6s 2s 1s what ever as long as the animator knows what the audience is going to watch, I'm completely game.

As long as it's not about free frames, but making the animation tools more compatible with the current staff, I totally agree with the decision. Back in the Lucas Arts days almost no animator did 3D, now most animators don't do anything but 3D, and to waste the budget on searching for the perfect 2D artists to able to also replicate Bagels visual style would be so hard. How ever, I am hoping that they ad some extra angles in there for the character to work with, because the character's attitude will become very bland over time of the only angle that changes on the character throughout the entire game is the Z rotation.

I'm also asking for as fast travel function; in case the walkcycle turns out to be very uninteresting to watch, at least it would be possible to jump between room very quickly. There's nothing like having to wait for a character walk in a generic slow pace from one end of the screen to the next when you just realized what to do in a puzzle on the other en of the series of rooms.

Share this post


Link to post
Share on other sites

very interesting thanks.

I was interested in the Moai engine and bumped into this video:

also, it turns out to be an open source project, that's cool :)

Share this post


Link to post
Share on other sites
Sign in to follow this  

×
×
  • Create New...