Presentation slides and notes

This is long overdue. I got really bogged down with other coursework which is no excuse.

Slide 1
As written my Aim is to explore methods of retargeting motion capture data to non human characters in the hopes of discovering a more efficient pipeline than those currently available.

Slide2

This slide shows an example of the main thing i'm hoping to avoid. Especially when it comes to biped characters a real problem is getting them to move as if they aren't just a guy in a suit at a theme park.

Slide 3

This is the first of the existing examples i found. It's taken from a short called "40 Years" and the stills on screen are taken from a breakdown video they made. While it's very nice to see it visually I haven't been able to find any documentation on what processes they needed to go through to get to this point

Slide 4

The other example is Ted, They had Seth Macfarlane on set in the motion capture suit (the one that uses sensors rather than markers which eliminates the problem of occlusion and marker swapping as they resonate at different frequencies) As you can tell by the image there's a problem with proportions when its first captured which then needs to be fixed in cleanup.

Slide5
So as a result of the research (that i have really badly documented on this blog, but i'll get better!) I have four methods i can look into to try and improve the pipeline a little. I think that it would be best if i focused on this four alone to narrow the scope of the project but at the same time if i find something interesting that's relevant and worthwhile Its better that i follow that.

Slide 6

This is an example of how props can be used. I won't be able to do something as grand as these but I would be able to get the actor to hold weights to help shift their own weight. The obvious problems that come from using props to adapt a performance are listed in the bullet points. The center of gravity might appear wrong for the character if it's a quadruped but also marker occlusion/shifting and confidence. If you're in one of those rigs i'm not sure how confidently you'd be able to move.

Slide7

so a big chunk of my research currently has been looking at the procedural methods of animation found in spore. The animators had a massive task of having to produce animation for creatures they'd never seen so it's an interesting if not very wordy read

Slide 8

Tasks that i'd need to do to complete things

Slide9

A thankful hippo

Comments

The main comment i got after presenting was that it seemed like my scope was too big for the time we have. I think this was because my slides didn't really express that i only wanted to find ways of improving it slightly using techniques i found rather than fixing it entirely. Because of this i should probably look at revising my aim and objectives to reflect that more clearly.

Research Structure - Early research

Limitations within motion capture

Develop, Will Freeman, 2013http://www.develop-online.net/tools-and-tech/motion-capture-moving-with-the-times/0117523
Markers - Limit movement possible
Motion Builder only allows for retargetting between characters with the same amount of bones. The video a little further down shows that this doesn't need to be the case.

constraints highlighted by Michael Gleicher at 1998 siggraph

"We motivate our approach with an example: retargetting motion capture data of an actress walking up to, picking up, and carrying away a box. During pre-processing, we augment the motion data by specifying constraints that are essential to the action: the hands must grab the box in the middle frame, the hands must remain the correct distance apart while carrying the box, and the feet must be planted and not skid when they are on the ground."

NOTE: Consider the affect props have on the capture data. 

Expand

Examples of Non-Human Biped Characters

 (and possible free rigs and meshes i might use during testing)

Could be characters with counter weights like wings, tails or asymmetrical features

Could be characters with legs set up more like a birds

Different centers of gravity

Proportions - Look at Ted and how they applied real motion to the short limbs of the bear

Human Rig for comparison

How to Judge success

How well the data applies to the new hierarchy without any adaptation?

consider how long the adaptation would take? 

Is the method efficient enough to improve existing pipelines?

Existing Work

Havok Engine retargetting research: 2009 

What I'd like to produceWould like to produce something like the above video for final report. Demonstrating the same motion capture data retargeted to different biped creatures. Hopefully looking a little less stiff than the examples in this video. There should probably be at least an example of a walk cycle and also maybe an interaction with an object.

Things to look intoSomething something.. methods of creating additive animation used to help find a balance between mocapped data and whats needed for the motion to be believable for the character

Chain Maps? (retargetting between different character resolutions. arms, legs, spine.) Important at all?

Motion Capture Monsters

So this is now my project. 

I'm going to look into retargetting motion capture data onto non human characters. As far as i can see, very little has been done to explore this area with the majority of research done been into realistic human motion and facial motion capture.

http://www.youtube.com/watch?v=JGNnhkk4bfU<--Using extra props such as stilt shoes and crutches to get more realistic data.

http://www.youtube.com/watch?v=xr6Ddvi0wPE <-- Example of how a person changes the way they walk to try and add a different sense of weight to performance capture. (also 2D motion capture)

Facial retargetting

Facial mocap using a phone and drawn on markers -->http://vimeo.com/27678634

MoCap possibilities?

http://www.youtube.com/watch?v=KBd2ToLt8xQ

Guy from mixamo speaking about their process.. Their markers don't stick out and each marker has it's own frequency which makes marker swaps much less likely. I'm jealous.

What are the limits of the motion capture set up? Push it to it's extreme? what can't you do with 8 cameras that you can do with more. Find examples and mimic them with our set up?

Retargetting motion capture data to non human meshes. Monster motion capture
http://vimeo.com/13132912

Motion capture fighting game? Tekken did it? Bit of a tricky one with the whole not knowing anyone who can do flips and things...plus they had the capeaibara

Motion Capture - Ikinema

This module is hard.

I want to do something using motion capture but honestly my brain shuts down when I try to think at the moment.

Idea's that have gone nowhere so far..
-Motion capture to create an environment somehow

WAIT
-2D animation using motion capture (was a nowhere but now maybe not?)

Ikinema has a free version that works on browsers. I've figured out little bits using various documentations that i've found and produced this using the tutorial assets that they provide.




I'd like to think about possible applications for 2D motion capture. I want more of a reason to play with this other than just helping massively with the motion capture pipeline.

Maybe making tutorials for creating game animations using this process? Go through the entire pipeline?



Also sidenote.
This took less than half an hour in total. My honours project is silently weeping.

Animation instead of Health Bars

Animations instead of health bars?

Attack/Defense animations

Drawing on fight and flight body language already done

Create a series of animations that replace the need for health bars but give a believable reaction to being attacked.

Maybe have additive animations for different parts of the body that change depending on how damaged that part is similar to the buildings damage in Starcraft. health bar for different Parts of the body.

"Is your opponent right-handed? Then why shouldn't a well-aimed blow to their right arm decrease its efficiency, forcing your foe to switch to his or her weaker arm? Are they fast? Then a blow to the head (concussion) should slow their reactions and decrease their accuracy." -Escapist forum

Starcraft Barracks Health

Rough Person Version

Weak spots affect on animation. Use the defensive nature some bosses have over their weak spots to help inform possible changes to regular animation?

Narrow down the topic by keeping animations to just a silhouette to show whole body gestures rather than details.


Articles related.
http://games.kitguru.net/featured/jon-martindale/lets-kill-off-health-bars/

Introduction

In this post I'll talk a bit about my past projects and areas of interest.

My main interest is in 3D characters. While I enjoy all aspects of the pipeline I am most experienced in the animation portion of the process having used both key frame and motion capture to produce final results.

I worked as the animator for my 3rd year group project which was an infinite runner, featuring a ninja running along rooftops sliding and jumping to navigate obstacles.

In the summer between 3rd and 4th year I took part in the Dare to be Digital competition on the project "Grav Tech". As part of this project I was responsible for the majority of the concept work, some asset creation and anything relating to the character.

My 4th year project was titled "Credible performance of NPC's within Story Driven Games" which aimed to devise a system to create a more believable performance taking into account current technological limitations and timescales. To demonstrate my research I created clips using motion capture to show a variety of possible interactions the player might have with a non playable character.

I played a small role on the art/empathy game "9.03m" created by Space Budgie as well as working on some of their other projects.

While I enjoy many of the art roles in the game development process, I have become more focused on learning more about the animation processes including more technical art roles such as using python script to produce helpful tools and Rigging.

In the future I would like to be more competent at 2D animation for games as well as 2D and 3D asset creation.