Friday Flashback #452

From the Jan 2001 issue of Computer Graphics

Men in motion

by Karen Moltenbrey

To achieve more realistic animation in games, developers have been using standard motion-capture techniques to create unique, compelling character movements. Links DigiWorks of Tokyo, however, has taken motion capture in games much further-six times, in fact-while creating an opening cinematic sequence involving six characters for Capcom’s upcoming Onimusha: Warlords game, expected to be released this month, for the PlayStation 2 console.

Using a Vicon 8 system from Oxford Metrics, Links DigiWorks simultaneously captured the motions of six actors for integration into samurai battle sequences that appear in the cinematic. “With the Vicon 8, which was introduced last year, we are no longer limited to capturing the motion of just one person at a time,” says Koji Ichihashi, president of Links DigiWorks. “Previously, when we needed two or three people in a battle scene, we’d have to capture the motion of each one separately, and then try to match them up in a scene. By shooting all the movement at one time, the actions and reactions of the actors during a fight scenario remain natural, fluid, and realistic.”

Onimusha’s opening cinematic sets the stage for the game’s story line, depicting agile warriors battling to the death on behalf of their powerful, fearsome warlords during 1560 feudal Japan. The video console game, which is loosely based on historical events, blends fiction with fantasy through extraordinary 3D images that take advantage of the Play Station 2’s highly touted graphics capabilities.

Capcom wanted the game’s opening sequence to be of similar quality, so Links DigiWorks hired Japanese dramatic feature-film director Shimako Sato to help stage the drama. Using a Vicon 8 system, the Links DigiWorks team shot several takes of six actors, specially trained in samurai movements, performing various fighting actions during a four-day motion-capture session. To capture all the motion at once, the group used 12 cameras set up around a 20-meter square area that contained rigs, props, and wooden hills to help depict the terrain where the action occurs in the cinematic sequence. The movements of the props, including spears and other weapons, were also captured during the session.

The most challenging part of the project, according to Ichihashi, was sorting and analyzing the motion data from the 100-plus reflective markers placed on the actors and props. “We had to determine which markers belonged to which characters in the scene, which was an extremely complex process, because the actors were moving around so quickly and hidden by each other’s bodies.” Once the data was sorted, the group mapped it onto low-resolution Softimage character models for review and additional cleanup of the motion.

Although using motion capture proved arduous-it took nearly eight months to complete just the opening movie-Ichihashi still believes it was the best option. “We chose motion capture over keyframing because we wanted natural, performance-style animation, not Disney-style animation,” he says. Also, Ichihashi believes that motion capture will enable the studio to create animation data, particularly for feature films, more quickly and less expensively than they can with keyframing. This project, he estimates, likely would have taken two to three times longer using keyframing.

Creating lifelike motion, however, was only half the battle. To achieve the desired realism for the cinematic, the group needed photo-quality characters on which to map the movements. Using Softimage 3D running on a variety of hardware, they created the main character models, dressed in period clothing and textured with Adobe Systems’ Photoshop.

With the company’s customized software, the artists also incorporated hordes of warriors and riders on horseback into the animations, as well as special effects such as fog, smoke, and raindrops. “The crowds were especially difficult to control,” notes Ichihashi. “At times there were hundreds of people doing battle in the fields.” To complete the scenes, the group added backgrounds and non-character models, as well as special effects such as fire and fog using its particle software.

Rendering the sequence, which contains 40 to 50 layers, was done in Softimage, Discreet’s 3D Studio Max, and Links DigiWorks’ proprietary rendering software, which was used mostly for the effects. According to Ichihashi, the majority of the layers were rendered overnight using a variety of hardware, including SGI machines (O2s, Indys, and Indigo2s), Unix boxes, and NT workstations. Some layers, though, took days to render.

Using Vicon’s Body Builder, Links DigiWorks sorted more than 100 mocap markers before mapping the motion to low-resolution Softimage characters for further review

Friday Flashback #450

3D Festival 2003 Report
Read more about the new features of Softimage|XSI v.3.5, the crowd animation system Softimage|Behaviour and about the interesting seminars at the 3D Festival in Copenhagen, Denmark.
May, 12th, 2003by Raffael Dickreuter

Softimage launched XSI version 3.5 and Behaviour 1.1. at the 3D Festival in Copenhagen. There were a lot of interesting seminars about the current situation in the industry, the history of 3D graphics and even an entire day about the making of Lord Of The Rings: The Two Towers.

Softimage|XSI v.3.5
Here’s the scoop about the new features of XSI 3.5: In the preferences you can now choose between two languages: English and Japanese, so you don’t need to buy a Japanese Version, you get it as a language package. You can now also set prefix-names for your models.

New features are the transform manipulators which may please especially former Maya and 3ds max users. On right mouse click you now get a little popup menu the options of “local”, “global” , “view” etc.. for faster access.

A new pivot point that has great snapping options makes it way easier and faster to rotate, move or scale a certain component of an object. Till now if you had a complex model and wanted to rotate an edge it was difficult to achieve the desired effect. Thanks to the new pivot point you can now work more accurately on a certain component. However it looked like it wasn’t yet animatable.

In the render tree you can now rename all your nodes. Mental Ray 3.2 now writes incremental .mi files. OpenGL accelerated scanline rendering and shadow maps are another improvement.

When rigging a character you will get a “move joint” tool in XSI 3.5 that makes it easier to edit your joints after they were created.

An impressive feature is that you can quadrangulate triangle objects. If you have a complicated object that consists of a huge amount of triangles you can easily convert them into quads. All triangles, as well as all n-side polygons will then become quad.
When you symmetrize polyongs it will symmetrize also the UV and keep only one UV map.

Two known features of RC Tools can now also be found in XSI: Select around corner and select edge loop.

Texturing a tree or a road will become very easy in XSI 3.5 You simply define four quarter points that will be the quarter points of the UV and you will see your texture to follow accurately the geometry. This feature is also animatable and fully interactive.

The animation editor now allows negative scaling. So you can move an entire curve into the negative area. The F-curve edtior got Softimage|3D’s slope handling tools.

Several improvements in the hair panel is going to make your life easier: A slider helps you matching the hair with the level of subdivision. The amount of hair segments becomes animatable and you can texture along the lenght of the hair. The tickness can be edited with proportional or absolute SI units. In the Shematic view you get memo cams that become very handy for complex scenes.

A new button allows you to clean up structure changes. An option lets you actiave it permantently so this process is automated. When using directX realtime shaders in XSI 3.5 a window no longer pops up but will also be displayed in your viewport just like the OpenGL realtime shaders.

Softimage’s Managing Director Michael Stojda (left) and Europe Business Manager Richard Craig-McFeely could also be found at the festival.
Softimage’s Erik Beaumont

Softimage|Behavior v.1.1
Softimage also showed Behavior 1.1 at the 3D Festival. It has a very strong API and through scripting you get full control about a huge amount of characters. You can tell them to move in certain directions. They avoid each other and obstacles in an intelligent way. In Behavior you create the crowds animation and you can get it into XSI or any
other 3d software via dotXSI.

James Rogers demos how to use XSI and Behavior together.
With Softimage|Behavior you can handle massive crowds.

They showed several examples of huge amounts of characters as well as how a rather small amount of characters reacts to it’s environment and to the other characters around it. To use Behaviour you have to know the programming language “Piccolo” which syntax is similar to JScript and C++.

3D Artist Ola Madsen just came over from Sweden to Demo XSI for Softimage.

Ola Madsen (XSI Base member Null) demonstrated also the power of XSI. The 3D artist from Sweden said that each demo session for a few people took often one or two hours. The people simply didn’t want to leave anymore and couldn’t get to see enough of XSI’s power.

Softimage’s women at the booth #1.

3D Festival Seminars

Creating Memorable Characters
Victor Navone of Pixar talked about creating memorable characters. He explained you should design a character from inside out. To get a proper design you first have to know what kind of person that character is. If he is friendly, generous, aggressive or whatever. This should then be applied to the look and design of your character.

It’s important is that your character is interesting for the audience, which means that it should have some type of flaw. Of course your character has to desire something which drives the story forward. A constrast between the different characters will make a story much more interesting. Your character is also defined a lot by his motion, The personality comes often accross to the audience via his motion. The body language also takes a key role when designing a character Of course the voice also plays an important role, so cast a memorable voice that suits your character.

Vitctor Navone also advises the visitors to exploit the opportunites unique to animation. Cartoon animation should be a combination of the familiar and the the fantastic.
A common mistake is that a character has too many flaws.

Rob Coleman
The animation director of ILM held a personal reflection about the evolution of 3d characters and showed how he got th job at ILM. He was the first person assigned to the postion of an animation director at ILM and stated that he loves to combine digital charactes with real life actors. Rob admitted that ILM’s great effects are the result of hard work and why ILM rather should be called “Industrial Light + Brute Force”.

He said that nowadays the 3d artists are no longer slaves to their computerers as they were the last 10-20 years. The todays challenge is no longer to overcome the limitations of the computer software but to teach the animators acting. Therefore the ILM people also have now acting classes.

Another session was the “History of 3D Graphics”. CG Talk held a voting before the event where the users could vote for the most important film effects. The winner was Jurassic Park followed by The Two Towers.

Lord Of The Rings: The Two Towers
During an entire day Jason Schleifer, Matt Aitken and Bay Raitt showed what it took to bring Gollum to life and how they achived the impressive effects like the crowds.

Gollum was modeled in Mirai by Bay Raitt and he developed a great character system that made it very fast for animators to give Gollum that strong facial expressions. They used two different characters as one was Gollum and the other Smeagol. Bay’s job was also the help the animators so that they understand that brilliant system to get the desired expressions. It was based on sliders that let you achieve many different facial expressions depending on their combination and their weight.

A funny moment came at the end of the presentation when a young person asked where you could download that animation plugin…

Shelley Page Special Student Session
Shelley Page from DreamWorks held a special seminar for students on how to get a job in a large studio. She explained that you should try to specialize in a certain area. There was no need to show great modeling if you wanted to be a character animator. You should rather work full time on your animation skills and then show that in your reel. If you create a great short film with your firends then this might be good way to get an award at a festival, but to get a job it might rather be difficult as it was hard to see for a studio which part of the film was done by you. If you want to show a group project in your reel be sure that you state clearly what you did and what is other people’s work.