How to create a contest-winning animation
Staff Writer By: Bryan Steagall
Hello Renderosity members!
My name is Bryan Steagall, and I'm the creator of "The Monster Race", the animation grand prize entry in the recent Generation Horror Wars contest at Renderosity.
In this article, I'm going to go a little into the processes that I used to create this animation, with video tutorials coming in the next month or so.
Start with a story
One of the things that I realized over the years, is that if you want to create a really good story, there's a lot (and I mean, A LOT) of planning you have to do before you even start to animate. (Trust me, it's worth doing this in the beginning and saves you work later on.)
If I had to break it down, these would be the steps in the planning process:
1. Come up with the premise of your story.
In this case, it was rather easy, as Renderosity gives you a general premise, "Generation Horror Wars." Hmmm ...
2. How are you going to interpret that premise?
I toyed around with some traditional horror ideas (like Baby monsters who eat their parents! Chest poppers vs. giant spiders! Blood, guts and gore!). I realized not only did these ideas not really interest me very much (done too many times) but they also would require some resources that I don't have to accomplish them as I would want to (fluid dynamics, anyone?) and a lot of time I really didn't have. So, in other words, sometimes simple ideas, if done right, are much more effective. I was leaning towards something more cheeky, kind of funny.
Discussing my ideas with my 18-year-old son, he suggested a different interpretation, a race or competition of some sort.
Hmm, this is interesting.
Maybe, a dance-off? Nah, I'm a white guy, I can't dance worth beans.
A race perhaps?
OK, a race sounds interesting but how do I meet the criteria? How about different monsters? But what monsters? Something popular that everyone knows. Dinosaurs? Zombies? Aha! Eureka!
The lesson here is to talk to someone else, your kids perhaps, they might have better ideas than you.
3. Outline and storyboard the actual action (Even if it's just in your head).
Now you have the idea come up with an outline of the actual action and write it down.
If there is dialogue, write a script. If you are lazy like me, imagine the actual shots in your head (How am I going to introduce each monster? Are they going to run down a street or a track? Are there going to be spectators?).
In real productions, this is called storyboarding and a shot list. Although I didn't, in this case, create a storyboard, I did write down how I was going to do all of it. This included how the cameras would be looking at each of the characters to introduce them, how I was going to generate interest by first introducing the zombie, then the dinosaur, then going to a different view to show the whole background, showing that they were competing in "monster races," then creating some tension and a "what happened here?" by having another dino attack the cameraman just as the running dino decided to cheat.
Now, this outline, storyboard or whatever you are using doesn't have to be rigid, but it does need to be the framework on which you will build the story.. as always, things come up that don't work and you have to adjust somewhat, but this frees you up by having a good idea what you want to do and what you will need to accomplish it.
1. Choose your assets
Now that you have your framework, start choosing your "assets," like characters, backgrounds, etc. I have a huge collection of props, characters, scenes, etc. (I don't know how many thousands of dollars of product I've bought over the years.) or buy what you need (or if you are inventive and have the skills, make your own! I certainly don't.)
2. Stock your toolbox
Decide what tools are you going to use to create your animation.
What are you going to use to animate or what do I have that can be used to animate? Will it be 2D, 3D, or a combination? This step might be combined in your coming up with your idea portion or during choosing your assets.
In my case, I knew from the start that I would be using motion capture, both facial and body, so that required MotionBuilder.
What am I going to use to render? Will it be Iray? Reality? Firefly?
I didn't like any of these. As I know, although they can produce good results, it would be very, very slow to render (even with a powerful PC with a good video card).
What is this Marmoset toolbag? Hmm, let me download a demo. Cool! This is exactly what I need. It's fast, uses fbx natively and looks terrific. Yaaay!
What am I going to use to put my scenes together once I render them? What am I going to use to add sound? (I cannot stress enough how extremely important sound design is. IT MAKES OR BREAKS AN ANIMATION!)
It is commonly said that a good movie is 80 percent sound, 20 percent visual. Think about it. Take a movie like Cars or Finding Dory. Watch it without sound and tell me what you think.
In my case, I have Adobe CS6 Production premium, with AE, Premiere and Audition.
Freesound.org and Incompetech are great resources for sound effects and free music. There are also a few free tools out there for nonlinear editing that can be used if you don't have one.
1. Capture some motion.
So now that you have all this decided and worked out, it's mocap time!
A. Motion capture.
One of the things I drill into everyone I teach about mocap: Do not go into the mocap volume without having worked out what you are going to capture.
Use your story to decide what movements you will need, not only of the main characters but any secondary character you might have. In my case, not only did I have two main characters, but I also had "spectators" on each side.
This meant at least three different takes of movement for each character in each scene.
You also have to think how each character moves or how it should move. Zombies, for example, should I use the traditional slow "I want brainzzzz" or the mad dash World War Z type?
Beacuse my mocap volume is not long enough for a run, I asked some friends in the industry to do a simple capture for me. My instructions were "a slap dash.. neurotic zombie run" like in World War Z.
The dinosaur, a deinonychus to be exact, was somewhat of a challenge. I read a breakdown online on how they did some of the mocap for the latest Jurassic Park, where they had these guys running around in dino suits, funny but effective. Then I went and watched the movies to see how they moved. I got it close, but even in the end my oldest son (the animator) kept telling me they needed to be "more birdlike, more birdlike, butt out, neck out, bend those knees more!"
B. Am I going to need facial mocap?
To keep it simple, only the two main characters needed facial mocap (and really only the zombie, and only during their close-ups, all the rest was hand animated, just open close the mouth).
I even planned for animating the camera movement.. by adding markers to a cardboard box and pretending it was a camera.. with my son punching me to simulate a hit (ouch)
If you are going to interact with your set (such as opening doors, climbing steps or using props), you need to add representations of these into your mocap space so that you can adjust your acting accordingly.
If your character is to grab a sword or staff, use a broomstick. If you are going to climb steps, use boxes. If you are going to grab something heavy, actually pick up something heavy (your body mechanics are very different when you pick up actual objects versus pretending).
Look at this picture of a mocap stage, taken from the "LA Mocap Summit" I attended earlier this year. Note the props such as boxes on the stage:
Now that's done, it is time to bring it into MotionBuilder (or your animation software of choice) for animation!
If you planned your scene breakdowns, camera shots, and your mocap right, you won't have to blend any of your mocap together (plan if you can to switch camera views, to hide stuff you know is hard to blend, transitions, effects, hits, grabbing stuff, sliding feet, etc.)
And cheat whenever necessary.
Sometimes though, it cannot be avoided. This is where MoBu shines. It has layers. You have a control rig that is easy to animate and change movement, constraints, which is how I animated the tail on the dinos without having to do anything and it is designed specifically for animation, whether mocap or keyframe.
First, export all the characters you are going to use in MotionBuilder as fbx, make sure you include props and morphs (if you are using DAZ Studio, edit the export rules to only export the head morphs, otherwise you will get a large, bloated mesh that can be slow loading in MotionBuilder). Most textures are also quite large when exported, I normally go in and resize them to keep things fast.
Second, import, characterize (which is adapting the existing rig to Motionbuilder's internal rigging system for retargeting and animation), and then save them as individual fbx files.
Keep a copy of these somewhere in case you mess things up, then open as needed and save in the scene you are working on.
Third, prioritize which animations to work on first. In my case, I decided to work on the one-character scenes first and all the secondary characters, as I knew that I could finish those quickly allowing me to concentrate on the difficult third scene.
Fourth, bring in your background set to gauge where the characters are going to be and see what type of adjustments you might need to do your animation if any. Once this is done, you can delete it or make it invisible so you can concentrate on your work.
Screenshot of scene3 in mobu, with primary characters, camera and secondary character.
Finally, even if you planned ahead of time, sometimes you might find yourself with a situation where one or more of your mocap files doesn't work like you had planned it to, or you change your mind and decide to add something at the last minute. Here you have a chance to go back and do some more capture to add to it.
Once your animation is complete, it is time to put all of it together for render!
Originally, I had planned to render the project in DAZ Studio so I could use Iray. Frankly, I dreaded the idea because I knew it would take a long time to render with having to deal with importing the fbx animation, saving as a duf, editing the file manually to use the facial mocap and so on.
For a long time, I've been looking for a way to render native MoBu scenes without having to bring them into another program for render. Out of curiosity, as I had never heard of Marmoset Toolbag (and seeing that it was one of the prizes), I decided to look it up and see what it was about.
I'm glad that I did, as it turned out that it has most of what I've been looking for: real-time lighting, good texturing, additional animation capabilities and, best of all, it's fast.
This was my process with Marmoset:
First, bring in your background set. In this case, Urban Sprawl 2 by Stonemason.
As Marmoset interprets textures a little differently when using fbx files, I had to go in and tweak the textures so that they would look right. This is relatively easy, and it also allows you to add maps that DAZ Studio and other programs don't include when exporting fbx format, such as normal maps, specular, etc. I copied all the textures from the DAZ Studio textures folder for Urban Sprawl to the texture folder created by the fbx export.
Second, bring in any additional props you are going to use that are part of the background (such as the barriers seen in the final animation and the "Monster Race" banner, created by my good friend Brett Martison).
Third, light the scene according to the plan you had and tweak, tweak, tweak.
Fourth, bring in your main characters for the first scene and tweak the textures as with the background set. Here you can really take advantage of textures such as normal maps, etc. Bring in any additional characters that might be required (such as the "spectators").
Fifth, setup your cameras and set effects for the camera as needed (such as DOF), render, save the file as scene1.
Sixth, open scene 1, delete the characters you used here, add characters for scene2, change camera view, render, save as scene2.
By doing it this way, I could always go back and change things if the render didn't come out as expected, which happened many times as I was learning the peculiarities of Marmoset lighting.
Seventh, open scene 2 and delete everything except the backgrounds and props, to bring in the files for scene 3.
Scene 3 was the most challenging of all because now I had two main characters who were interacting, around six or seven "spectators" on each side, and an additional character that would "bump" the imaginary camera person as they were attacked.
Here I found a flaw in my production planning that I didn't think of, plus a few quirks of Marmoset that made me have to change a few things.
The thing that I didn't think of was this: I did all my mocap on a flat floor, however, the floor of the set was neither flat, nor regular (the ground was slightly peaked in the middle, sloping towards each side and with an incline), so what happened?
My characters started out good but after a few steps, they were running on air! Oh, no!
Fortunately for me, Marmoset does allow you to move around and add animation on the individual objects in the fbx files, so I ended up cheating by adding keyframes to bring back each character back down to the floor, which was not perfect, but enough to make it look OK (And if I ever go back to fix it, I will adjust this by adding an inclined floor in MotionBuilder and adjust accordingly).
The other problem was a quirk of the camera in Toolbag, called a "clipping plane," which does not allow you to get a camera too close to any object.
When you do, it starts cutting into the mesh of the object (sort of like an MRI slice of the brain). So, I had to adjust by changing the zoom of the camera to fake it coming up close and bumping the camera. Since the action was really fast, it was not that noticeable.
2. Visual effects
Before bringing the animation into Adobe Premiere for final composition, I had to see if there were any effects that were needed on my renders. Here is where you would add effects such as gun blasts, lightsabers and stuff like that.
As Marmoset, at the moment, doesn't have motion blur, this was the only thing that I decided was needed, and only on the last and final scene. So, I added a "time warp" effect in AE to add the motion blur that I needed on the final scene and re-rendered that out.
3. Final composition
I used Adobe Premiere for this project, but there are many free software packages out there that can do the same (Try Bedroom Porducers Blog) some of which are professional quality and very good.
As I mentioned before and I will stress it here again, sound design makes or breaks an animation. So you should pay some attention to it here.
I used a combination of sound that I captured during mocap (such as the zombie scream and teeth clicks, captured by a cheap lavalier), and high-quality sound FX from Freesound.org.
When designing your "soundscape," you have to stop and think of sounds that you normally take for granted during everyday activities. For example, when somebody is walking, the sound the shoe makes when it hits the ground, is it a gravel surface? Wood? Concrete? What type of shoe?
You have to think of distance to the "camera" (who is you...), a sound will be faint the further it is away from you and gets louder as it comes near. Take a few hours to listen to things around you or watch a movie focusing on the sounds and how they were used.
If you can't find a sound, make your own, you would be amazed what "Foley" artists (those that create sounds for movies) use to create the sounds you see in movies.
Create layers in your editor for each of the sounds or groups of sounds you are going to use. For example, all the shoe sounds for the zombie as he is running go in one layer, all the foot sounds for the dino in another, so on and so forth. This makes it easier to go back and change if needed.
The final part of my composition was the background music, which can really set the mood of your animation. Here my go-to site is incompetech.com by Kevin McLeod, who has a whole library of music that he creates with different "moods" and allows their use for free (with suitable crediting of course).
In my case, I had narrowed down my music choices to two wildly different styles, one that was cheeky and kind of funny, the other a more mysterious and eerie choice, both of which fit the animation. After consulting with several people, the consensus was that the eerie one would fit the best.
Wow, if you were patient enough to get to this part, congratulations! This is it. We are done!
Thank you very much for reading my long and rambling article, I will be following up in coming months with detailed tutorials for some aspects of this article for those interested.
About Bryan Steagall
I'm 48 years old, married with two grown kids, a full-time day job, a part-time company of my own and involved in my local church's kids ministry. My work is primarily IT, although I studied fine art in college (as a metalsmith)
I started creating animations almost 15 years ago when I started looking for a way to tell interesting stories to my Sunday School kids.
I had grown up with video games and stumbled across Poser 5. I quickly learned that as an animator, I would probably starve to death (I was very impatient back in those days and my animations were terrible!)
After a few years of soldiering on with Poser 5, 6 and 7 (and getting a little better with each year), I thought to myself "There has to be an easier and faster way of creating animations in the timeframe I'm looking for." This is when I found motion capture!
So I started using free mocap files from the internet. And again, quickly found out that they weren't exactly what I was looking for, mostly walk or run cycles (or "sexy walks" ) and I couldn't find the specific movements that I needed. I discovered MotionBuilder (back when it was Kaydara), which helped a little, but still not enough. So I tried contacting some motion capture providers (and almost died from sticker shock!).
Then I thought, "how expensive can a mocap system be?"
Talk about a rude awakening. I think the cheapest I could find was an exoskeleton mocap rig called the Gypsy 5 at around $25,000.
I already had my wife convinced (99% of the battle) and was just about to purchase it when I stumbled on a little company in Oregon called OptiTrack. This was right before they came out on the market with their first mocap system.
To keep this story short, I bought the system for about $6,000, I believe. Since then I have added a whole repertoire of software skills to enhance the stories I wanted to tell (such as Particle Illusion, Vue, Adobe After Effects, Premiere, DAZ Studio, Adobe Audition for sound, etc.). I have learned how to act for mocap (and used my kids as actors, one of which now is studying animation in college, I wonder why?), all by using mostly free tutorials.
My part-time company is now the Latin American reseller for OptiTrack and for Dynamixyz (facial mocap). I also install and train people on how to use mocap systems all over the continent.
I guess what I'm trying to convey with this is: Don't despair, keep at it, there are a lot of resources out there to learn.
I don't consider myself a professional by any means (my son's skills far outstrip my own), but I keep learning anyway, and I'm more than happy to help anyone along the way.