Turok: Animation Workflow

So as a bit of an extension to the Turok: Creature Reel post, I thought it might be helpful to some to put a bit of info on how I went about creating these animations...

I'm not sure really how to make this a short post, but I will try my best. I feel its important with workflow posts to try and be detailed and not wash over stuff.... animation is hard!

As a kind of "disclaimer"I want to make it clear at this point that I am constantly striving to improve my planning/workflow techniques and I am confident that there are more constructive, creative, and time saving processes for producing this kind of work. My own methods usually produce results I am pleased with, but I admit that I have much to learn about streamlining my approach.
If you are interested in how I created these sequences, then this post will try and outline in as much detail as possible how I went about it, but please don't take my examples as a definitive way of working... what works for some may seem ludicrous to others.

Setting the Scene:

So to summarize a little about the setting/context in which these animations were to be seen, I will outline briefly a few ideas that help give an understanding of what these animations are about:

• The player of the video game is stranded on an unknown planet, inhabited by ferocious creatures, deadly human opponents and a group of allies that are afraid and have little trust for each other. Creatures are deadly and bloodthirsty, and an immense threat without the proper firepower. The player is a hero character of Native Indian desent, fearful of the creatures around him, but willing to use bravery and force to dispose of them in violent ways... often at close range using only a knife.

Starting an Animation:

We are pretty lucky here at Propaganda in that we don't have rigid quota on our animations. Almost all in-game animations are scheduled 'per animation' and not based on any relevance to screen time. The Turok Creature Reel shows a few different kinds of tasks - the less intensive variety ( such as the TRex eating the small dino ) and the more involved type ( like the Lurker attacking the human player ). As a general 'window time' we based an estimate of 2 days for the simpler animations, and 5 days for the more complex ones. We were lucky in that we could usually extend the window by a few days if it needed that time to add the extra polish that we felt was important to our game... we knew that for a compelling dinosaur experience the creatures had to not only move convincingly, but appear menacing and vicious ( even if they were injured or being killed ) ... we were willing to spend the time to do this well.

As an average I would say that I probably spent the allocated 2 days on the simpler animations, but usually the reasonable maximum of 7 or 8 days on the complex sequences.


Thinking High-Level:

Before starting anything, I like to just spend a while thinking very globally about what I should be showing and what I would like to show. As an example I will discuss the animation where the Lurker is attacking and killing the human player for the most part, but I will refer to other animations to make certain points as I go along.



These are the areas and kinds of things I like to think about first ( and usually in this kind of order )

• What will make this sequence entertaining ?
• If I was playing this game for the first time, what would I like to see happen ?
• What are the technical constraints ? ( e.g ground coverage, camera placements etc )
  
• How can I make the Lurker move in a violent, aggressive and dominating way ?
• How can I make the human seem doomed and terrified, yet still seem heroic ?
• How will the two characters actually come into contact ?
  
• How will the Lurker overpower the human ?
• What will be the killing blow or move that finishes the human ?
• Will the human have a fighting chance ?
• What angle do I want to shoot this from ?
• Will the camera be animated ?
• How can I get as much movement as possible in the sequence ?
• Who is the focus of the sequence ? the human or the Lurker ?
• How can I work some decent contrast and shape change into the action ?
• How can I get some cool poses on the Lurker and the human ?
  
• What similar sequences have already been animated, and how will this one be different ?
• What will make someone want to watch this again and again ?
  
• What are the best things to draw out first, and what do I need to shoot for reference ?
• Do I have the tools to do what I need to do ?
  
• Am I good enough to animate this ?

And to give a little background info on the Lurker creature for context:

• Its a fictional cross between a big cat and a lizard
• It moves like a big cat, but can also traverse trees and surfaces like a lizard
• It is vicious and aggressive, like a cornered pit bull, or threatened grizzly bear
• It hates everyone!

Talking High-Level:

I have some great colleagues at work that love to talk about stabbing dinosaurs in the head and wrestling down the odd bigger one. After pondering through the list above, I would usually throw an idea or two at these guys, act them out a bit in the studio, and get some feedback from them. Often I would get new and better ideas, or at least some decent expansion on the ideas I originally had. If they thought it sounded cool, and so did I, then it couldn't be far off.

I am not a good at drawing, so I much prefer to talk early on about an initial idea than show drawings / poses or even blocked out 3D sketches that take time to make. I would also show any video reference at this point, if I had recorded any, or had any on the internet that helped illustrate my ideas.


Bringing Ideas Together

So for the Lurker kill human animation, there were a few things I knew for certain ( technical constraints )

• There had to be a "struggle" at some point for gameplay reasons
  
• The whole thing couldn't be too long
• It couldn't cover too much ground, and had to take place on a flat surface
• The camera needed to feel like a hand-held, and could not make big broad movements
• It all had to be in one shot - no cuts

And these were the things I decided I wanted to do in the sequence, following my initial ideas and seeing what my colleagues thought would be cool

• The Lurker pounces on the human slamming him to the ground with great force
• There is a shape change as the Lurker adjusts himself to a good position to start tearing flesh
• The human is seen to have a last heroic effort to hold off the inevitable death by trying to struggle out of the Lurker's clutches
• In response to the human's effort, the Lurker decides to just shake him violently and break his back in a broad contrasting action

My inspiration for these ideas was basically from thinking how a lion or leopard would pounce on it's prey and pin it down, but then instead of the almost surgical way big cats kill their victims by biting down on the neck, I wanted the more brutal way that a grizzly bear might maul and tear at its enemy. So I mixed the two.


Staging and Broad Action

To be honest I really like this part of the process. After I've got a pretty fleshed out idea of what I think will happen in the animation, I try and figure out the very global movements or high level staging that will be seen in the sequence, imagining the characters as single shapes moving the the screen, and how the camera will move. This stage is all about very high level contrast and variety, but also thinking about what angles the characters will look good from ( e.g, you would want to see a good view of the Lurkers face as he bites in )


I didn't do this prior to animation, I drew this up for the post to explain my thinking - I mostly just figure this out in my head. The reason I like this stage so much is because it can really help you get ideas for what you want your characters to do... you can almost think of the high level staging contrasts first, and then think of suitable actions to make them happen... its fun, and helps go a long way towards a decent animation if you get this stage right. You're trying to keep the screen alive too as well as the characters within it. .


These are the main staging ideas for the sequence where human wrestles the large injured Dilophosaurus ( beginning of reel ). There is contrast in the broad action and camera movements.



Drawing

This is probably the point at which I should mention the overall way in which I work. I am very much a straight ahead and layered approach animator. I don't yet know how to properly plan and block out animations... I am working hard on this.
I like to animate things by getting a strong idea in my mind of what I know will look good, then I figure out how to animate it by concentrating on what are the primary driving forces, and the most important motions or actions that I need to do first in order to make everything else fall into place. I animate these things first, then I animate everything else afterwards. I do not animate everything at once, knowing exactly where things are coming from and where they are going ( like our old pal Milt Khal ) ... I am way off being that good or that organised. I really love the progressive control and feedback that blocking offers, but I love the spontaneity and immersion that comes from working the way I do.

But I do plan. Its just my planning isn't always in the form of thumbnail drawings or 2D flipbooks. Sometimes I draw thumbnails, sometimes I act things out and record it. Sometimes I just go with the flow and start in 3D ( but not often ).

For the Lurker attack, I pretty much could see in my head what I wanted to happen. I wanted the camera nice and close to the ground to emphasise the threat from the creature, and I wanted there to be a strong sense of physical connection between the two characters by using a good overall pose that kept them feeling entangled and brawlish.. almost like the lurker was 'hugging' the human... and biting into his chest and stomach. For this section of the animation I only did one small drawing that helped me summarize the kind of thing I was going for and mostly posed the two characters together in 3D until I had something I was happy with. The main problem I wanted to solve in thumbnails was the shape change and follow-through of the two characters when the human is violently dragged at the end, so I drew some stuff for that before I went to 3D:


As far as thumbnails go, I never really try and figure out too much at this stage. To be honest, this has a lot to do with my inability to draw the real details of what I can imagine in my mind.. I can usually draw the broad shapes, but then I hit a roadblock because I cannot draw out the subtle shape changes or poses that I want. Basically what this means is that I only ever really work out the main "story poses" that describe the most basic elements of what happens.... rarely any breakdowns. At this point I'm really just thinking about the kind of shapes I want to see, and from what angle. The drawing top left ( with a zero above it ) was the entanglement feeling I wanted, and the rest are just figuring out how the drag would work.

Here are a few more thumbnail sheets I managed to find for some other animations I didn't put on the reel. I wish I had some that are more relevant, but I seem to have trouble keeping stuff like that around, but at least these give an idea of the kind of detail level I take these to. Below each one I have posted a video of how the animation finally came out so you have some context





Recording Reference

I found that for almost all the animations on the Turok reel, I needed to record reference, particularly to give a much better indication of how the human would move in a realistic way. I believe when shooting for realism this is an essential part of the process - there are too many subtle and even very broad actions that the human actor will do that you could never imagine or anticipate just through imagining or thumbnailing. I find that more than anything else it is a great generator of new ideas, or it will at least prove or break an idea you already had on paper.

For the Lurker kill animation, despite what I just said, I didn't record any reference, so I can't show you anything for this. I think really I should have, particularly for the 'struggle' section, but I decided to figure this out in 3D mostly because the motion of the human would depend so much on the motion of the Lurker ( in fact, for the entire animation), but I think I could have got a better result if I had at least acted out some kind of struggle in a similar position and recorded it.
Below is an animation I decided to not to render and include on the Creature Reel because I'm not that happy with it to be honest, but it does show how I used direct video reference of myself to create the motion of the human actor



and here's the reference I recorded:



As an indication, I probably spent a good 40 minutes acting this out over and over, trying different things ( some within the basis of the same idea, and some completley new ideas ) and doing a good 30 or so tries. It seems to me like you need a good 20 minutes to warm up to what you're doing and to really get into it, and usually I find the best take is always the last one, or very near to the end.
I find it very weird as well that the ones you though were good when you did them often look crappy, and its the ones you don't remember so much that look great.

As far as using the reference, I pretty much used it as close as to what was recorded as possible, changing a few things ( mostly timing ) here and there to get the desired effect. If you're going to work this way its crucial that your reference is dead-on. I think part of the problem with this completed animation is that I think I made a bad decision on the reference I decided to use - I should have gone a lot more extreme with the idea of the human moving out of the way of the stumbling dinosaur and pushed it much more.


Blocking

I'm sure when some of you hear the word 'blocking' you instantly conjure up images of AM Mentor reels, Pixar bonus DVD features, and directors and supervisors all standing around a monitor looking happy. And so you should. I think blocking is great, I just personally find it very hard to work completely in this way ( because I need to practice more ).
I can say pretty safely for a good 4 or so years of my career I would just animate straight ahead - not frame by frame, but I would block very loosely between two main poses at a time. I would then animate to almost final between those two poses, then move onto the next two poses until my animation was finished. To be honest I didn't even know about blocking, and seeing as I am entirely self-taught, I never really thought of doing it any other way. The problem with this method was that I never really had much control over what was happening on a high level, had a hard time making changes, and had no real indication of how my animation would look until it was finished. On the other hand, my work always felt spontaneous, organic and was a real blast to create. My current struggle is finding a balance between my methods and a more constructive blocking approach.

So coming back to what I was saying before, my approach to blocking is the same as my approach to animating...I like to animate by thinking about what is most important and doing that first. when I say most important I'm thinking in this way:

• What action or pose NEEDS to happen at a certain point because it drives the story or the impact of the sequence ?
• What movements are directly driving a lot of other movements ? ( like the Lurker driving the motion of the human being dragged )
  
• what character is driving the execution of the staging and camera ? ( who is the primary focus of the sequence and where are they going in the frame ? )

With the Lurker kill animation, below is a set of images that are a rough indication of what I blocked out in 3D before going into animation. These are really just the story poses, and I didn't take the blocking to any more detail than this. I was just trying to lock down in 3D the high level staging, and thinking of the characters as broad shapes rather than complex ones.


I could not find any version history for this animation file, so the images above are simply screen grabs from the final animation. They represent more the points at which I set poses in the sequence, rather than the fidelity of the poses being set. The real ones would have been much much broader and less detailed than these. As an indication of "priority" of these poses, I felt that 1, 3, 5 and 12 were the most important for getting the staging across as needed ( i.e, these are all the poses i absolutley needed - and i spent the most time thinking about the fidelity of these poses in more detail ) all the other poses are loose indications of how I would go between these priority poses, and I would not worry too much about how these looked. I would also be setting these poses in stepped mode, and thinking about where I placed them in the timeline so they would represent a very rough indication of the timing.

I managed to find some version history of the Dilophosaurus animation, which gives a better indication of how I block things out. The scenes I recorded for the video were saved at random points throughout the time I was creating the sequence, so they don't represent exact "milestones" of the blocking I did, instead they just give an indication of the various stages.



Looking at the video above it illustrates pretty well some of the ideas behind my workflow

• section 1 ) Here I'm starting to block out the motion of the human first ( I used video ref of myself for this ). The initial poses are set up to show the basic relationships of where the characters are in relation to each other at the start of the animation.

• section 2) One of the main actions in the sequence is where the dino bites back at the human, who manages to dodge. At this point the the blocking I am roughing out the human up to this point, and also the biting action of the dino. By doing this I can properly verify where the characters need to be at the start of the animation so they end up in the right place at the point of the biting action, relative to how far the human moves forward and how far the dino head moves. I am also starting to think about how the dino will move his body forward in a lunging action that will add power to the bite ( but also affect where he ends up at the end of the action ) I am thinking more carefully about the human poses at this point but still not spending too much time on them. For the dino poses, I am still thinking of him mainly as just his broad shape, and where he is in relation to the human. The most important thing about the dino at this stage is what his head is doing.

• section 3 ) Still working from my video reference, I am blocking the human duck and jump animation. The human is leading the staging throughout the whole animation, so I work on his actions first. I know that he needs to get back on his feet and force the dino's head down by jumping on it - so there is a key point where the dino's head needs to be in relation to how far the human jumps. This key contact point means that I need to know roughly where the dino will be when the human hits the ground, so I block out this pose too ( I also start to get a good idea of how far the dino is moving accross the floor, and the cool kinds of struggly actions I can animate to make his do this ). The white dots that floating in space are a motion trail that I always have attached to the human root bone, this is most useful in stepped mode, when you are blocking out poses, but cannot see directly the relationship between them in 3D space. Using the trail I have a direct indication if the motion of his root is doing anything weird or has odd spacing on it.

• section 4 ) So this section is actually a good representation of how far I would take the blocking of this animation, before starting to work into it in detail. All the main actions are there, and technically everything works ( with regards to distances and contact poses particularly ). I also have a good indication of the staging, and can work a rough pass of the camera movement into the scene. I could also show the animation at this stage and get feedback on high level changes without too much effort.
  

• Some things to notice that are not blocked out by section 4 - The blocking of the human is still very loose, I'm not overly concerned about exact poses, instead just the overall feel and technical aspects of the animation. The dino has very few poses that really say much at all, at this stage the blocking is just to make sure hes in the right place, and most of the attention on him is based around his head and its position to the human. The number of stabs and the timing for the stabbing sequence is very loose, but the sense of the stabbing action is there. There is also no fancy blocking of any of the erratic movements the dino makes in the final rendered shot.

I just want to point out here that I always try and animate as much as I can in full 3D. Game animators are generally used to thinking in this way because usually every animation we do can be seen from all angles by the audience who plays the game.
When doing cinematic animation to a specific camera view, I always work primeraly to that shot as my "optimal viewpoint" but I try and make the animation work from all angles. This forces you to think more realistically about the space in which the action is taking place, and gives you a clearer understanding of where things are going and coming from. It also means that you have room to modify your camera angles and play around a bit with camera animation when you have the main character movement finished and polished up.


Starting to Refine

My approach to starting the real nitty gritty part of animating is to work in a straight ahead and layered way between the poses I have set in my blocking. I rarely go back and set any breakdowns to my blocking poses. Instead what I have are pretty solid "marker poses" that make sure my timing is basically on track, my characters are going where they need to go, and I can see ahead of time the overall picture of what will happen in the sequence... ultimately this means that I can work in the more spontaneous way that I am used to, but its much harder for me to go totally off track and end up making a mess or not quite hitting the mark as I imagined... because the framework is already there.

What I found interesting about these animations were the sheer amount of movement that affects other movement, which in turn affected the way I animated it. For example, when the Lurker is struggling with the human, the Lurker is the main driving force because hes so powerful, yet there is some force on the Lurker in the form of resistance to the human.

With the Dilophosaurus animation there were a number of these relationships:

• The dodge and duck of the human affects where the dino needs to bite so the actions seem believable
  
• When the human jumps and wrestles down the dino, its head is affected by when and how this jumping move takes place and how much force the human inflicts on it
  
• Towards the end where the dino is shaking its head and reacting to the knife stabs, this affects how the human moves because he is hanging onto the dino's head

Because of these primary driving forces, I animated the Dilophosaurus animation in this structured order

• I animated the human first, up until the point that he has finished his jump, and is in the pose where he has the dino head on the ground and is about to stab. I took the time to work detail into these movements, getting the human feeling good and not worrying too much about the dino's body or head. At this point when I am animating anything I am not worrying about keying all controllers on every frame, or in fact being that tidy with my key frames at all, I am working in auto-key and just posing things around and keying them wherever it feels right
  

• I then animated just the dino's head, all the way from the beggining, through the bites, the scream, and all the way up to the point where the human is about to stab. I was concentrating on the big mass of the head because of its sheer size and importance in staging for the shot. I animated how it would react to the force of the human wrestling it down, and the initial impact with the floor. I still didnt worry too much about the animation of the dino's body, just making sure however that I didn't push the head too far beyond a pose that my rough blocking of the body could accommodate

• Once the human has the dino on the ground, I started to block very roughly the timing of the stabs... so I just blocked out the extra stabbing motions from the rough blocking I did at the start, and got the timing feeling pretty good. I didn't concentrate too much even on the human animation, I just animated the arm stabbing and got the feel for it. I did this because each stab would affect the movement of the dino head in a small way, which would affect the human because he is holding on to it!
  

• Now I had the timing of the stabs, I animated again just the dino's head struggling and roaring in reaction to the attack, all the way up to the point that the dino has died and the human starts to get back up again. I made sure I moved the head enough to get decent reaction from the human holding on, but also so I could get some driving body movement in the dino's body. For this stage I was loosley animating the human and the dino's body, but only loosely to get an idea of staging and broad movement

• With the dino's head struggling animation pretty much done, I could animate the human reacting to the forces, and take the human to near completion, all the way up to where he stands back up and the animation finishes. I also made the stabs fit properly to the movement of the dino's head, but trying to keep to the same timing I blocked out for them. The human is now about 90% animated, with follow through and all the details included

• Finally I did a straight ahead pass on the dino's body. I made the root move so that it kept up with the driving force of the head, and I animated the frantic motion of the legs in a way that they would also appear to push the dino to the places that it needed to be in its staging. I added reactions and squirms in the legs, root and arms to the section where the human wrestles its head down, and finally to the stabs and its eventual death. I just went with this pass by pass adding and changing as I worked through.

With the Lurker kill animation it was a similar process:

• I animated the Lurker jumping all the way through the air and landing on the ground using poses 2, 3 and 4 as a rough guide ( see the poses image from before ). I took this motion pretty far getting the nice timing, arcs and overlap. I was loosley posing and adjusting the human but not thinking past the main contact poses when the the Lurker initially hits him in the air, and then when he finally hits the ground.

• I then animated the Lurker jumping around and repositioning itself to get to pose 5

• After I had the main attacking action of the Lurker feeling smooth and enjoyable, I animated the human reacting to the pounce, hitting the floor and struggling in reaction to the Lurker repositioning itself. I knew where I could flail the human's legs and where he could try and grab and push away because I already had the the animation of the Lurker there and working

• I then animated just the Lurker's head biting into the human's stomach and moving around, up until the point that he just begins to drag him away. Again, not thinking too much about the movement of the Lurker's body or the human, but making sure that I moved the head enough to get some decent body movement in both characters

• Using the head movement as the main focus of the staging, I pretty much did a straight ahead pass on the Lurker's body and the human's reactions to the bites in one go. I added the weight shifting and pulling that the Lurker does mainly as a result of the range of motion on the head animation. I animated this up until the point that the Lurker begins to drag the human away

• I then animated the root and back legs of the Lurker performing the shaking and dragging. I very roughly blocked the chest and head of the Lurker, and some poses for the human just making sure that I wasn't straying too far from the main story poses, and technically that I was covering enough ground to drag the human far enough and keep the staging as planned

• Once I had the main root and back leg motion for the Lurker in the dragging section, I animated the chest, front legs, and head of the Lurker reacting to this driving force

• I then animated the human being dragged, using the head as the main point of force

• Finally I did a straight ahead pass on the limpness of the human at the end, and the final lunge of the Lurker

During this stage of animation, I am usually trying to get visibility and feedback on what I'm doing to make sure anyone else that sees it is getting the entertainment factor I'm shooting for, and making any changes that are not too destructive or time consuming at this point.

For all elements of anything I animate I follow this same principle. So when I say "I animated the human" I am usually animating the root and foot plants together, then all the torso and arm movement that results from the main driving force of the root. I did a post that goes a bit more detailed into this idea here


Polishing and Taking It In

As a final pass on my animation I will go through and add motion trails to most objects and get the arcs feeling smooth and organic. This tends to make a huge difference, and you see improvement in areas you didn't even notice had any problems. I love motion trails.

I then spend a little while cleaning up curves. Nothing intense at all, just making sure there's no weird flat tangents or spiky sections. To be honest I only usually spend much time in the graph editor if I need to solve specific problems with acceleration or deceleration, or technical problems like locking down movement between duplicate poses by setting flat tangents etc... sometimes its impossible to get the kind of precise or complex movement you want without pulling around some curves.

I then play with the camera a little, changing the shot and framing a bit here and there to find the best way to shoot the action ( you get to do this if you work in full 3D as I mentioned before ). These are not vast changes at all, but can make dramatic improvements.

I then sit back and watch the animation over and over. I do this to look for 3 main things

• Do I need to change and poses or timing in any way to make the sequence better ?
  
• Where can I add subtle movements ( reactions in all animated objects to forces and actions smaller than the main driving forces) to make it more believable ?
• Do I need to globally adjust or modify any poses or animation to get a better result ?
• Does this entertain me ?
  

I usually use animation layers to add subtle movements and global changes. This way I can forget about trying to work within existing keyframes and just start fresh using additive keyframes over the top of underlying motion. Its best to do this at the end because our in-house animation layer tool locks off the underlying animation once you add layers over the top. You can get some great subtle movement doing this, and you can also blend in and out of broad changes over time ( for example I added a layer to the Lurker's root that bought him closer to the human during the struggle section and I blended this change from the point that he repositions himself after the pounce ). I love animation layers too! .... I could post this up along with a tutorial if I get any interest.

So... if you managed to hang in until the end, you have my congratulations, and I hope you found at least some of that interesting or useful. Please post any questions you might have or fire me an e-mail.

Its good for me to write this stuff because it helps me clarify things in my own mind too...