Death Stranding


I’ve always had an interest in works by Hideo Kojima and recently I’ve become a bit obsessed with Death Stranding.
Having watched the trailers a number of times I was captivated by the little robot which sits atop this characters shoulder.
Initially captivated by its movement and just wanting to rig and animate it I set about blocking out the forms with cubes, then I thought, I’ve come this far, so why not actually model it up. So that is what I’m doing.
I’m currently UVing this mesh and once that is done I will texture, rig an animate it.


Reconnecting with lost animation

Occasionally an animator will be working away in their scene, when a new rig is published. Usually this isn’t an issue and the animator will reload the rig and have it available in their scene.
But every now and then, someone doesn’t have their coffee in the morning and instead will load in the new rig, unload the old, save their work and -shock!- their animation is gone! having saved their work they cannot see a way to retrieve it.
At this point I tend the hear “Aaall, my animation has broken!”
Talking to them and having a look at their scene it is apparent what has happened.
Upon loading a new rig, the animation is not connected to it, and when unloading the old rig the animCurve nodes are disconnected from the rig controls.
Fortunately the solution to this is simple.

1. find all animCurve nodes.
2. discern what control node it was connected to.
3. reconnect the animCurve node to the newly discerned control node.

I’ve attached some simple code below that demonstrates this process.

import maya.cmds as cmds

animCurves = cmds.ls(type='animCurve')
nameSpace = 'CH'

for animCurve in animCurves:
    attr = animCurve.split('_')[-1]
    control = (animCurve.replace(attr,''))[0:-1]
    try:
        if nameSpace:
            cmds.connectAttr('%s.output' % animCurve, '%s:%s.%s' % (nameSpace, control, attr))
        if not nameSpace:
            cmds.connectAttr('%s.output' % animCurve, '%s.%s' % (control, attr))
    except:
        print 'could not connect %s' % animCurve

My latest demo reel

My most recent demo reel is of my work at PlaySide Studios where I have been rigging characters and looking after artist tools and pipeline.
In this demo reel I am showing the final rendered mesh of the asset, the skeleton of the asset, to demonstrate the capabilities of the rig and its efficiencies. Then showing a wireframe overlay to display the deformation of the character clearly.

Getting back to the shoulders.

I recently found time to get into some improvements in my rigging system.
For the longest time I’ve not paid much attention to a characters scapular, often finding that it is off screen or required at all.
But with Games technology improving and limits on bone count and poly count increasing I find my self with the opportunity to include some extra deformation systems.

So I dug back into my old auto rigger and took a look at the scapular rig. Needless to say, there was not much going on, and what was there really, sorely needed improving.

After an afternoon of tweaking, testing and building I developed a solution.

(this is a GIF, you can click and watch it.)

This solution is based off the clavicle rotation, which is then multiplied or divided through a condition node network and used to drive the scapular bone.

A little time travel.

During August of 2006 I rendered this image:

At the time I felt it was the best I could achieve, I’d used all the tricks I had in my bag.
A physics based glass shader, global illumination, final gather and light caustics rendered with Mental Ray from 3DS Max.
I was pretty chuffed with the results!

Recently I decided to come back to the concept, the rules were simple:
– use the same objects, 1 bottle, 1 glass, 5 eggs – two in cups all on a wooden table.
– use the same type of textures.
– I can adjust the camera angle and composition.
– I can use different rendering techniques or engines.
– Everything else stays the same, don’t add geo or effects

With this in mind I set out modelling the simple geo. I learned a few tricks to get good results creating poly geo from a nurbs curve, which perhaps I’ll go into another time.

Mental Ray was my choice as I’d used it in the past and am familiar with the concepts of Physically Based Rendering.
Initially I went about setting up some basic shaders then jumped into lighting, this proved to be slow, and at the advice of a friend I went back to setting up the shaders without lighting.

Once the basics were setup, the almost endless iteration began. Tweaking shaders and tweaking lighting, changing the environment and adjusting the HDR.
Finally I arrived at what I consider an end point, as we could all just keep tweaking until the cows come home.
After two weeks of work I have this image:

I learned a lot during this project and it was a fun return to some rendering, which I’ve not done for years.

..

….

But its not over yet!
Some of my co-workers use the Arnold Renderer and had been saying to me that I should tackle this project in Arnold.
Having started and made some progress in Mental Ray I chose to complete it. Once it was complete I couldn’t help but come back and have another shot.
So once again into the fray of shader setup and lighting. I used the exact same geo, lights and textures. The challenge was to reach the same result.
I’m very pleased to say the working with Arnold was fantastic! The renderer allows for really fast iteration of shaders and lighting.
The render view has some excellent features like rendering selected objects, fast IPR and customization of the UI elements.
To re-create the same scene and image only took two days!

I did find some interesting differences between Arnold and Mental Ray.
Arnold is lightning fast to work with, though I found it to have softer reflections, and the angel of the Fresnel reflections yields a different result.
The bottle glass shader was difficult to setup in each but with Arnold I had to resort to using a samplerInfo node to approximate the refraction through the volume of the glass.
Mental Ray’s reflections are fast and crisp, though its workflow is slow.

To wrap it all up I’ve assembled my work in progress renders into a time lapse and poster it on youtube.

Need for Speed!

After two years with EA – Firemonkeys, my team and myself are happy to launch Need For Speed: No Limits VR

During this project I was responsible for the character rigging, animation, animation planning and artist team tools!

The trailer is attached below but the 360 degree video works best on your phone. :)

Hiding Maya’s output wnidow

I’ve had enough of Maya’s output window, I know I’m not the only one.
Fortunately Autodesk have introduced a proces to hide the output window with Maya 2015+

This is done by setting the System Environment Variable: MAYA_NO_CONSOLE_WINDOW 1
and restarting your computer.

To implement this navigate to System Propities > Advanced > Environment Variables

capture

I get to this screen in Windows 10 by hitting start and typing “Environment Variables”

Once you’re in the Environment Variables screen, click New

capture2

and create the variable “MAYA_NO_CONSOLE_WINDOW” (without the quotations) and set the value of this variable to 1.

capture3

once this is done, restart your computer.

Thats it, you should be done.

I’ve found this a much cleaner way to work, espically when opening scenes which have a lot of .tif textures to load.

The only issue I’ve found is that when done without restarting my computer, Maya no longer accessed the python interparator correctly.
Further more, when I was loading a plugin which was missing some components, the output window was useful to display this information.

Profiling code.

I’ve been wondering lately, when it comes to finding nodes in a scene, nodes which may or may no exist, which will be faster?

using a Try, Except?
or using maya’s built in objExists command?

My theory was that the Try, Except would be faster as objExists would have to loop over all nodes in the scene. At least as far as I assume.
So i put it to the test.

The first thing I neede ws a profiler. A little googling and I turned up Pyhtons Timeit module.

This can be imported using

import timeit

print timeit.timeit(function)

This will print the result of running the specified function one milling times. Its is run this many times to ensure consistency in the results.
That can be reduced by adding the number flag, to specify how many times the code is run, so:

import timeit

print timeit.timeit(function, number = 10000)

This will run the code ten thousand times. Its quicker and consistant enough for my requirements.

Getting into the problem at hand, I setup two functions, one as a Try, Except and another as objExists.
I ran each one in an empty scene and each one in a scene which included a complex character rig.
The code and results are below:

import maya.cmds as cmds
import timeit

def exists():
    found = cmds.objExists('locator1')
    return found

def select():
    found = False
    try:
        cmds.select('locator1')
        found = True
    except:
        found = False
    return found

print timeit.timeit(select, number = 10000)

The results

The Try, Except took:
0.398722180715 in an empty scene and
0.376243066404 in a complex rig scene

where as the objExists took
0.0698589508479 in an empty scene and
0.0415636909748 in a complex rig scene

astonishing, and, contrary to my expectations, the Try, Except took nearly ten fold longer than the objExists!
From this I learned: Using specific commands looks to be faster than using Try, Except! I’ll have to test more commands and find out :)
If you’d like to try this code yourself, you can see the result for each funstion by changing this line:

print timeit.timeit(select, number = 10000)

To this:

print timeit.timeit(exists, number = 10000)

Cheers!