there is no need to create a vdb, but it works yes
Super taf! ;)
Ted Bundy's car? :D
Aleksandr Maziura talked about the production of the second teaser trailer for his large project COMBINE made in UE4. In this article, he covered droid design, texturing, animation, rigging, assembly, and lighting. In the previous article, you can read about the character and weapon design for the first teaser.
Hi, my name is Aleksandr Maziura and I am a 3D artist from Ukraine. Quite a while ago, during my university years, I participated in the development of the game Neon Commander as a designer and an artist/animator. For a year I worked as an animator on HOPA games and then, 4 years ago, started working as a 3d artist. Right now I’m working as a 3d artist at an outsourcing company.
About the Project Combine
I always wanted to make video games and storytelling, visuals, and audio are the most interesting parts for me. A cinematographic game is very hard to create, almost impossible for one person. But damn, I wanted a bright and brutal cinematic video game with synthwave, adventure, and retrofuturism of the 80s and 90s. Also, when I was younger I liked old Harry Harrison’s novels Deathworld and Stainless Steel Rat with their almost Indiana Jones kind of adventures style. I’ve put all this stuff in my project Combine.
I decided to make a cutscene/gameplay trailer for it, without any time restrictions and with full control over it.
I analyzed, what amount of work I can handle, what tricks can be used to avoid parts where I have less experience, etc. After all, I kept in mind the whole action/adventure story but the trailer was going to be 3-5 min long. What could I put in? And I decided to opt for an action scene on a giant atomic train which is transporting rare ore from the mines to a spaceport.
I started from a draft storyboard:
I had a huge amount of work ahead and started to work on the project 2 years ago. To slowly study UE4 and progress, I set milestones in the form of teaser trailers. Recently, I released the second teaser, and here’s the first one:
In this part, I’ll focus on a droid asset, however, the same principles were used for all the models. You can find more about the gun and character production in my previous article.
The robot featured in the second trailer is used in the private property protection field, something like a battle droid just more industrial. It has this blocky sharp design but with a smooth, almost stylized chamfer that reminds of something between DOOM and Overwatch.
I gathered references. Made a rough blockout and overpainted it:
Next, I move to the legs. In this project, I followed a simple rule for robots and different mechanism designs: the design should be readable from a technical standpoint, even a little simplistic. I designed the feet and tested them with simple animation.
Then I made a sliding part.
After that, I returned to the concept and took new mood pictures.
Then, classical subdiv modeling with Modo’s Mesh Fusion. Then, low poly and bakes.
I had this moody industrial image as a reference when I started texturing.
The process is fairly standard. I split the mesh into 4 objects: head and hands, right leg, left leg and torso-weapons. All those objects have the same texel density, the right and left legs have the same materials with different position maps and seeds in generators, plus some manual paintover.
My main idea was in layering and utilizing procedure generation as much as possible.
In my opinion, it’s Substance Painter‘s “Holy Grail”: masks on top of masks. Of course, you can always stack all the generators and filters in only one fill layer, but for me, it’s easier to manage masks this way.
I start with a fill layer with parameters like color, height, etc. seeing it as a pure material without any wear but with height. Then, I just use masks to layer this material properly. You can stack them, paint over, change blending types in any texture channel, and so on. It’s very flexible and modular.
And don’t forget about gradients! They always look cool.
The hardest part was to figure out a color scheme and gradients. It was an iterative process.
Here’s how the same approach works for different assets, the train rooftop and droid’s dispenser. You can find more details here.
Rigging & Animation
Rigging is the best time to change the pace of work and think more technically. I rigged arms with Inverse kinematics (IK). Weapons aim with (surprise) aim constraint. The rest of the torso and the head are made with Forward kinematics (FK).
I wanted droid to be foldable and with IK legs, so I needed to make an additional rig for them.
To achieve this sliding effect I made a simple math system.
The Measure tool is measuring the distance between foot and thigh and then this parameter is normalized and transferred into bone translation on the X-axis.
Then, the rotation of the thigh bone is constrained to ‘virtual’ IK bones with a standard knee of 0.5 weight.
There are 3 skeletons used for the legs: deformer skeleton will be exported as FBX (all keys will be baked on these bones and all constraints will be removed). Then, IK and FK leg bones and controllers. Basically, I just needed to parent constraint deformers to FK/IK legs and then link constraint weight and visibility to custom channel on the global control curve. For that, I used the Set Driven Key tool. You can do the same stuff in the node editor but it is so much faster.
Skinning, in case of mechanical stuff, is the easiest part of the whole rigging process.
Then I used basic animation principles with default or linear graphs, ease0-n and -out, anticipation and heavy object impulses. If the rig is made right, the character gets animated almost on its own.
Moving to UE4
I’m using UE4 because, in my opinion, making the whole movie in real-time is much more enjoyable.
First, I imported all skeletal meshes, textures, and animations and put the whole gang into one scene.
There’s nothing fancy to the materials, just a default shader. The roughness parameter is exposed to the material instance for real-time tweaking.
After that, I started to explore lighting options.
The actions in the final movie will take place in a desert during the daytime. However, in the second teaser, I decided to go with a more stylized approach and hide the environment which is not ready. I made super quick tiled dunes in ZBrush and scrolled them in front of the camera with different speed to simulate parallax.
Then, blurred everything with DOF, AtmosphericFog and ExponentialHeightFogm placed simple particle systems, clouds, and the sun.
Lightning the scene is an iterative process for me. First, I think what I want to show and hide in the scene, what colors I want to use. Then I put some directional, point and spotlights, made them movable, and placed low-intensity skylight for ambient light.
While I worked on the second teaser I got used to Sequencer and now I appreciate it much more because it’s very flexible.
I animated cameras and made some camera rig in Maya.
Bone CBN_def_cam02 is basically a socket for the camera in UE4. After the camera animation was baked I added looped animation in cam01_shacker to simulate constant camera shaking. Then, skinned all the bones to a single flipped triangle and exported it as an .fbx with animation. Then, imported it in UE4 as a skeletal mesh with animation and placed it in the scene and Sequencer. Attached a camera to the last bone in the hierarchy. This way I have full controls over a clean camera in Sequencer on top of the baked animation.
Basic keying in Sequencer is very useful as you get a decent amount of parameters to animate in real-time. I put all the animated stuff in it, set and tweak the camera and render everything as a .png sequence.
For video assembly, I chose Davinci Resolve, it’s free, quite powerful, and allows node-based compositing, audio video/editing. For the rendered sequence, I did almost all post-processing in UE4: vignette, film grain, and the shooting effect.
For the outro music part, I made a cassette and animated it, then blended it with the logo and background.
I am not an audio specialist at all but sounds and music are the best tools to set the mood. I had an idea of what style I wanted this teaser to have, a synthwave-adventurous beat. I’ve been listening to a neat synthwave radio for a while and found the track for the second trailer thanks to it. I contacted the band 3FORCE on Facebook, told them a little bit about the project and asked permission to use a part of the track. Since I’m not making any money from this animated short, I offered them credits in the video. The first teaser uses music by Confrontantional.
First, I want to make what I’ve planned: an action scene on a giant atomic train. The milestones I’ve set:
- Main character model, texture and rig
- Weapon and the first person mode
- First UE4 scene and test, the first teaser
- Train roof tile, dispenser, the second teaser
- Locomotive and wagons
As you can see, a few of the goals were already achieved, by I still have some work to do. Optimistically, when the goals are completed I will write a script, storyboard, and concepts, put them away for a few years and then look at what I can do with these materials. Making a full length or even a 20-40 min story alone as a side project is way too hardcore for me. We’ll see!
Anyway, I’m having quite a lot of fun working on Combine, and besides, it will be the centerpiece of my portfolio when it’s ready. So it’s a win-win.
My respect to anyone who has read all the way to the end! Cheers!