Meet the Artist: Kenn McDonald

#40

Hi Kenn,

Thank you for taking the time and answering our questions!

When you take the most challenging shots the animators had to work on - what do they have in common? - or maybe more simply put - what are the common causes for complication and inacuracies in the motion capture process?

Working so closely on CG characters based on real actors and their performances, were there any difficulties resulting from the gap between wire-frame and flesh-and-bone?

…I can’t help asking what was the weirdest sentence that had the word “Angelina” in it.

Finaly, what was Bob’s involvment in the process? Was he present at the mo-cap stage when the lead roles were captured?

I have many more questions but I guess that will do for now!

Cheers mate!

#41

i haven’t seen the movie yet(my mom doesn’t care to see it), but from the trailers and previews i’ve seen, you got angelina jolie’s lips perfect! (1) what was the process of modeling her? (2)did she come and you took pictures and did sketches,(3) and did she hang around while your guys were modeling her? i’ ve seen modeled people like in spiderman but i never knew how it went along. also they werent animated close up as well, just at a distance…btw(4) was angwlina animated by animators or was she face captured? and my last one (5) do you think it was easier for the movie being fully 3d or would it have been better mostly 3dcg?

#42

Hi Kenn,

Thank you for taking the time and answering our questions.

I’like to know what kind of renderer you guys used for characters,
mental ray, renderman , or mixture of them?

Recently, I used mental ray for personal project, but it takes a long time
for subdivision, and motion blur, after I add details to characters I made.

So, I’d like to know how you guys optimize rendering process, even after
you created so detailed characters with very high quality animation.

#43

Hi Kenn, great job ,great done.
Is there any inside or productons views or at least behind thw scenes to get an image how you guys worked on that?
Regards
Piet

#44

Hi Sachin,

Thanks for the kinds words.

I’ve talked quite a bit about the eyes in other posts so I won’t go into it here again except to say that it was a high priority for use to improve on what we had done on Polar Express. Looking back on our approach on Beowulf, I think we made a big leap forward, but there are still shots where the eyes aren’t quite there. I look at those shots and I know we have a lot of eye animation in there or we used the EOG, but they still feel kind of, well, dead. I don’t like using that word, because I don’t think any of the shots in the film are “dead”, but there are shots where the eyes still don’t feel quite right. It’s something that I will look at over and over again and plan to tackle on my next project. Still, all in all, I think the animators did an amazing job bringing the eyes to life.

Your other question is a really good one and something that doesn’t get talked about much. There are 4 phases in an animated film production, in my world anyway.

In the first phase you are setting up the rigs and pipeline and doing you first test shots. These shots tend to get worked over a lot and usually end up looking pretty amazing.

In the second phase the rigs are mostly done, the pipeline is coming together and it’s time to start making the movie. This phase starts out slow and some of the shots might not be as sweet as the earlier test shots. The rigs are being shook down and tweaked and the animators are getting in sync with the characters and the supervisors.

Then comes the third phase. Everything is rolling you hit that sweet spot where you aren’t going too fast, everything is working pretty well and the characters are well established. This is usually the longest phase and is where the great stuff gets done.

The fourth phase is the “get 'er done” phase. Everyone puts the petal to the metal and works as fast as they can. On Beowulf we had some of our most complicated sequences to do at the end, but the animation team really stepped up and kept the quality coming. I’m really proud of them for doing that. Unlike many movies where the quality starts to suffer near the end, I think on Beowulf it continued to get better right up to the last day.

I hope that answered your question. At least it gave me the opportunity to brag on my crew.

Kenn

#45

Hi Jay,

Three great questions.

  1. The first program I ever animated in was 3d Studio Max. I took a class and modeled a robot and did a little animated film. Shortly after that I was hired at Imageworks and started working with Maya and since then I haven’t worked with any other program. Except for a little bit of tutorial work in Motion Builder to become familiar with how it works.

  2. This is a great question and therefore very tough to answer. Ultimately if you’ve got a great reel it doesn’t matter what program you’ve used to create the animation. Your reel shows that you can learn a program and can animate. However, if you want to work on your scripting skills, which I highly recommend, you might consider switching to Maya if you plan on working for one of the big feature or FX studios. Most of the biggies I now of use Maya. There are a lot more jobs in games however and they use Max and XSI a lot. Your final choice is going to come down to what you really want to do in your career. How’s that for a whishy washy answer?

  3. I’ve worked with many mocap artists over the last 5 years and one thing is very clear. The best mocap artists have a great eye for detail and an understanding of the dynamics of human locomotion. They don’t necessarily need to be good keyframe animators, some are and some aren’t. But they do need to understand performance.

Kenn

#46

Hi Matthew,

Thanks for the kind words. I’ll be sure to pass them along. It really is great to hear that people are enjoying the film. We thought we were working on something good while were were making the film, but you never know until it comes out.

Now on to the questions.

I personally didn’t have a lot of face to face time with the actors. I met many of them and had a memorable night drinking with Ray Winstone, but my on set time was very limited. Bob Zemekis worked very closely with the actors. Capturing a scene was very much like a run through for a stage play. The entire scene was captured in one take. They might do a scene 3 or 4 or 5 times then move on. No need for pickups. Bob would place the cameras later. Many of the actors really liked this approach. Anthony Hopkins was the first person I heard who likened it to being in a stage play. Angelina Jolie in paticulr enjoyed the process. I’m paraphrasing here, but basically she said that it was freeing. She didn’t have to worry about where the camera was or how the light was hitting her. She could focus entirely on the other actors and her performance.

Bob worked very closely with the actors. Instead of being behind a bank of monitors, he was sitting just a few feet outside of the capture volume. He watched the performances first hand. After a take he could just step into the volume, talk with the actors, give notes, whatever, then they could go again very quickly.

A set could also be struck very quickly and a new environment put together. No moving cameras around or relighting a scene. All of that came later. It was really amazing to watch the actors work on the days that I was on the set. They were acting for the whole day. There was very little sitting around and waiting.

Previz was done using the performance capture. We basically captured everything in the script. Bob chose the takes he liked based on what he saw that day on the set. The body motion was then tracked and solved and put on low poly characters with video game type textures on them to identify who they were. Bob place virtual cameras in these scenes and started breaking down the film into shots. It’s a very organic process and I think one of the big reasons why Bob likes doing the performance capture movies. It gives the director an amazing amount of freedom in how he wants to tell the story visually.

The approach we took to animating over the performance capture was based on methods we developed on Polar Express. I had worked very closely with the rigging and pipeline teams on that film and what we did on Beowulf grew out of those experiences. Without going into too much detail, the approach is based on layering. Animation on top of performance capture. The body and facial rigs are designed to allow the animator to blend performance capture with animation, turn down the mocap, turn off the mocap, switch from mocap to animation and back to mocap again and so on… I’m not a big fan of just automatically grabbing a few frames from the performance capture for timing and poses then blow away the mocap. I think you really lose something there. There are times when that is appropriate, but most of the time, with a little thinking and planning you can combine the mocap and the animation to get a result that neither could achieve alone. That’s what fascinates me about these films. The possibilities that are still untapped. We haven’t even scratched the surface.

Kenn

#47

Hi Kenn,

I am currently a senior in highschool and am planning on going to the Art Institute in Orange County California to major in Media Arts and Animation. I reside in a small town in Oklahoma that has very little information on the art field and so I was wondering if you could possibly clearify things about two college majors. I want to work in the field of Visual Effects and I was of course looking at the Visual Effects and Motion Graphics major. In California, however, they no longer have that major because of lack of interest. So, I started to look at the Media Arts and Animation and saw that you can become a “Computer-Generated Effects Artist” through that major. Is that the same thing as a Visual Effects Artist or are they different? If they are different, will the Media Arts and Animation major allow me to become a visual effects artist? What are the differences in the Visual Effects and Motion Graphics and the Media Arts and Animation majors regarding the visual effects field?

Any information you caould give me on this would be highly appreciated.

Thank you for your time,

Liz

#48

Kenn,

That’s much more than I expected to get answered! Glad to see the enthusiasm in the reply.

All the best to you and the team on the next project!

Cheers,
Matthew

#49

I just watched to movie last night. It wasnt like the graphics were all show-offy and spectacular, but they were still amazing and realistic. I think the story was quite moving and emotional and the fact that it was in 3D didnt take away at all, and I felt emotionally connected to the characters. I loved the movie.

Such a Good job!!!

#50

Hi Kenn, again… lol i usually limit myself one post on these types of Q&A’s but i have to say unbelievable work! as this could be the 3D glasses and the imax talking but i was in total awe.

His maybe a big ask and also the fact you alreadly answered in previous but i replies but in But…

1a) your own words could you describe what a average day(or even a week) in your life and whats its is like ?

1b) What you expect the recieve as a animator and what is expected of you and your department?

2a) Also if you can explain Sony imageworks pipeline as best you can do give a better picture for everyone as its a rare chance for a person with your talents is here online to answer questions, even if it just animation department but any other areas eg modeling, texturing would be fantastic?

2b) Like how many numbers of people work under you (and vise versa)and the leadership structure?

im sorry if i got cared away with the question but after seening the BEO movie i was so humbled by the fact your here with us answering the community’s questions.

And want to pass on a big thankyou on to you for your time and for making a cool cg movie.

jay(your humbled admirer)…

all the best for 2008

#51

Hi Kenn

I was interested in your earlier comments about the transition from the 'toon-styled character animation in Open Season to the realism of Beowulf.

I remember listening to Randy Cook give a talk about the character work in Lord of The Rings. One of the things he touched on was the need to avoid many of the established principles of 2D animation that have become instinctive in most animators. For example, anticipation: if a character needs a “surprise” expression it would be a mistake to drop the eyebrows momentarily in advance of raising them into the full “surprise” pose, the way we normally would for a 'toon character.

Another theory, the origin of which I’ve long since forgotten, says that the 'toon style places most of its emphasis on the held pose (and progresses by means of the pose-to-pose principle), whereas in realistic work it’s the movement between poses that requires special attention. I’ve always placed a lot of merit in this philosophy; in realistic stuff you always need to watch out for “spliney” character movements and tweak those curves ever-so-carefully. You can’t hide that splineyness with 3 or 4 or 5 frame “snaps” between poses, something that in traditional work is the lifeblood of energetic, dynamic characters.

Do you have thoughts on this?

Moving on a bit: I’m a 3D character animator (with 2D training) thinking about mapping out the future of my career. I think mocap and performance capture is going to become a vast new territory for animators. If you were auditioning reels for character animators in a realistic show, what special things would you be looking for as opposed to a “toon” production? Do you think the required training or background of candidates is different? If someone wanted to specialise in the realistic form, what advice would you give? Or is specialisation not even advisable or necessary?

Thanks

Tim

#52

Hi Ruben,

The models were based on measurements and scans of the actors. In most cases the modeler stayed very close to the actors measurements. In the case of Hrothdar, played by Anthony Hopkins, the modelers went in and adjusted his mass to make him, well, basically into a big fat jolly guy. Beowulf’s model was based on a scan of a body double, but was extesively reworked.

I don’t know what the legal issues were involving the actors. That’s way outside my department.

Kenn

#53

Hi Alper,

In the animation department we use a lot of mel scripted tools to work with the mocap and animation, but really don’t do many simulations. We had a rig based jiggle simulation that we used extensively on Beowulf , Grendal’s Mothere and King Hrothgar. However it wasnt based on scripting.

Kenn

#54

Hi Tim,

I agree with Randy to some degree. The thing is that those 2d, cartoony principles came from studying real life. They are exaggerated in toony animation, but they are still there in realistic motion. We used them quite a bit in Beowulf to give the performances a little extra life. Much of the time those little anticipations were already there in the actors performances. The performance capture didn’t always pick them up, but if we often saw them in the video reference and added them in the animation. They do have to be subtle though. If you telegraph too much then it becomes a cartoon.

I agree 100% on the concept of pose to pose versus animation that is constantly in motion. It’s one of the aspects of realistic motion that makes it so hard. That splineiness that you talked about. You need that ‘noise’ I spoke of in another post that performance capture gives you. It’s not really noise and it’s certainly not random, but there is a low level hum to all of us that must be taken into account if you want your animation to be believable.

Yes. performance capture is going to be around for a while. I don’t think it will supplant keyframe animation, but there are a lot of mocap projects getting started. When I look at an animator for a mocap show the first thing I want to see is just good solid animation. It doesn’t matter if it’s realistic or cartoony. I’m looking for a solid understanding of basic animation principles and an eye for detail. Seeing mocap related experience helps, but solids skills and a good attitude count for a lot more. I always ask a candidate how they feel about mocap. Their response tells me a lot about how they will do on the show.

Kenn

#55

Yes.

Well perhaps I over-paraphrased what he was trying to say. The way he really framed it was as a warning that in photorealism you can’t rely on 2D cartoon “tricks” to bring your characters to life. As you said, a lot of it is exageration. Stretching the length of an arm in the act of throwing a ball exagerates the effort of the activity. Backing your character up before walking him forward exagerates the initial thrust.

But some tricks are substitutes for realism that, for whatever reason, is not able to be achieved. For example, in Jurassic Park the T. Rex ran by pounding its feet into the ground. I don’t know about you but I thought it looked pretty good. It sure looked massive and powerful and dangerous. But then I saw the spinasaurus in Jurassic Park III and that looked so much better! What happened between those two movies was the development of dynamic muscle sims. Apparently, the exagerated pounding was a deliberate artistic conceit to make up for the lack of dynamic musculature. Once their rigging technology grew to include rippling muscles they could forego the “interpretive” animation and go for the real McCoy (and feature the dinosaur in daylight, to boot).

I think that’s what Randy was trying to say. In cartoons (and dinosaur scenes that take place at night), where some things are lacking you can make up for them by substituting something else. But you usually can’t get away with that in photoreal stuff. Or you can try it on anyway and risk a trip to Uncanny Valley.

Speaking of which, I have a theory!

I think the Uncanny Valley is not a chasm that needs to be leapt over in a single bound, but rather, it’s more like uneven ground that requires the CG team to form a line abreast and advance in synchronised paces, one step at a time :o)

Let me explain:

As a representation of a human being, Homer Simpson is not much better than a stick figure. Let’s say he rates only 10% on the realism scale. But he works. That’s because his proportions are 10% real, his texturing is 10% real, his movements are 10% real, his behaviour is 10% real, etc.

Similarly, you can look at a more realistic animated human like Max Steel (from the kids TV show). He was much more realistic: say, 60%. But once again, he was 60% all the way through: 60% realism in modelling, 60% realism in texturing, 60% in facial animation, etc. So, for me at least, he worked.

The problems start to occur when there’s a mismatch in one or more of those realism factors. If you look around at some of the CG humans viewable on the net, there’s really nothing more creepy than a cartoonishly modelled human (even if only ever-so-slightly cartoonish), but with photoreal skin texture… br-r-r-r-r! What happens when you get even closer to the photoreal end of the scale is that the matchups between those component parts need to have so much more parity.

So steering realistic characters around the Uncanny Valley doesn’t mean a sudden exponential increase in realism on all fronts, merely that you maintain tight control of all your design factors and don’t let one aspect get ahead of the others - even if a technological advance in that one area allows it to.

As you can probably tell, I spend far too much time thinking about things like this. Anyway, that my theory, of which I reserve the right to disown at any time :o)

cheers

Tim

#56

Tim,

I like your theory about crossing the uncanny valley. A shortfall on any front will pull the viewer out of the movie. He will become aware that he is watching a cg character. I’m still not sure the uncanny valley exists per se, but I do think you’re onto something there.

Kenn

#57

Ok, so here I am half way through my two year technical art course, but I have the feeling that I could probably have learn’t everything so far on my own. My question is thus, how acceptable is self tought students as opposed to those more traditionally tought?

In my opinion, I’d have thought that having the drive to learn the techniques used in the field through your own self discipline would be more impressive? Any thoughts on this would be greatly appreciated.

~Thanks

#58

Hi Everyone,

I’ve been getting a lot of questions about rendering, textures and cloth and hair, which are outside my area, so I’ve gone to the experts for some answers. I spoke to Theo Bialek and Francis Liu, two of the CG supervisors on Beowulf, and they filled me in.

Cloth and hair were done in Sony Imageworks proprietary software.

Lighting was done in Katana which is proprietory software as is Bonsai, our compositing program.

Cinema 4D was used to handle our matte paintings.

Now on to render times. This turned out to be a pretty complicated question, but Frankie broke it down for me very nicely. Environments generally took around 1 1/2 hours perframe. An individual character was between 30 and 60 minutes per frame. On top of that there are shadow passes and effects and lots of other stuff that is a mystery to me, but make the shot look really cool. All together the render time for a frame, with everyting in it, varied widely. For a simple one or two character shot it could range from 20 to 30 hours per frame.

Thanks to Theo and Frankie for the information.

Kenn

#59

Time to wind it up!

I am sure everyone reading will join me in thanking Kenn McDonald for giving his time in this enthralling ‘Meet the Artist’ thread. His work as Animation Supervisor on Imageworks’ ‘Beowulf’ is part of film-making history, as are many of the techniques and technologies we have discussed here.

Thank you all who took part and thanks also to Rachel Falikoff from Sony Pictures Imageworks for helping organise.