Next-Gen Tech and Selling Ideas

by Jason Della Rocca (October 2005)

Ever the popular format, this was the chapter's fourth members-only roundtable style discussion night. About 50 developers and industry professionals showed up to discuss and debate on various game development related topics (see summaries below). Unlike the chapter's usual beer infused social/presentation nights, the evening was somewhat more serious as developers were locked in debate for about an hour and a half.

After the roundtable sessions, everyone hung around for snacks/coffee and continued discussion. And, we raffled off a few game development books along with t-shirts.

Roundtable Summaries

Selling Your Ideas... Internally

by Vander Caballero, Lead Game Designer, EA

When the roundtable finished I thought "O crap! This round table was a disaster, the topic was too broad". People spoke based on their experiences, but the experiences were so different among big and small developers. The strategies people shared were all so different when selling your ideas inside, or outside, the team.

It was almost impossible to close the roundtable with a conclusion that made sense to every one, but at the end of the roundtable some people came to thank me for the information they got from everyone's comments. This ranged from simply having opened their eyes to the often complex task of convincing other people, to notions of getting a champion behind you or trying hard to empathizes with others so you can understand what drives them and what motivates their decision making.

Next-Gen Intersections: Physics/AI/Animation

by Martin Walker, Technical Director, A2M

Next-Gen platforms will provide for an unprecedented level of integration between physics, artificial intelligence and animation. These advances will open the door for greater levels of realism and immersive gameplay. But, how do we manage the intersection of these complex aspects of game development and get everything to work seamlessly? Or even convincingly?

Interestingly the better part of the conversation revolved around integration of physics simulation and the animation system. AI was mentioned but briefly during most of the discussions and although it was said that AI should be driving gameplay, it should be taking its cues from physics.

One aspect that was discussed thoroughly was the predictive nature of physics simulation used for driving animations. Predicting the outcome of a set of actions will allow for better blending of animations to reach the expected behavior. Knowing in advance that we are going to hit a wall we naturally have a tendency to brace ourselves. Being able to predict this in a game will help blend in the proper animations ahead of time to make the character more believable.

Real physics is actually boring to gameplay and in most situations the player is asked to contribute to determine the outcome of physics dynamically, contributions which in real life would not be possible. We used sports as an example and more specifically hockey. In real life physics the outcome of whether a player scores is determine from the time the player shoots the puck. Although the hockey player can have some “effect” on the puck at the moment of impact, he cannot change the trajectory while in flight after the puck is launched. In a game scenario we will add puck trajectory change in real-time to enhance the participation of the player rather than waiting passively for the outcome of the slap shot.

So how far do we push physics simulation and how much can we just fake. The biggest decision factor was contribution to gameplay. Although cloth and water simulation will be feasible on Next-Gen, how much of it will help make the game more appealing. We agreed that it might help sell the game from a “feature list behind the box” perspective but that is where it stops in most cases.

One of the main intersections of all three aspects was centered around environment destruction or other types of dynamic modification of the environment and its influence on AI. As an example we discussed the impact of destruction rubble on NPC behavior. Should the NPC navigate around the rubble or walk over it? If walking over it, how much influence should the rubble have over its ability to walk (ie tripping, sliding or otherwise stumbling)? Curiously enough we came back to the conclusion that true simulation would be difficult, costly, too unpredictable and often not desirable. Faking it by adding a random stumble or slide would do the trick and it would be much simpler to implement.

The last 20 minutes (out of a 90 minute roundtable) was centered specifically and only on AI. Several experimental games were discussed as well as some research (Ken Perlin's research, AI in Indigo Prophecy and Facade). Crowd simulation was discussed but mainly seen as yet another feature to put behind the box to sell the game.

We concluded by stating that although it seems Physics, AI and animation should be part of one seamless integration, even our discussions had a tendency to separate and isolate AI out of the bunch. Physics and Animation are known quantities. As animation is becoming more and more procedural it is blending into the physics algorithms. Our impression is that eventually, artists will be determining character moods in a general sense based on behavior parameterization rather than placing vertices or moving bones in a skeleton. AI being more of a non-deterministic set of behavioral algorithms, it becomes more difficult to blend it into these known quantities unless a particular AI is well defined.

Scaling Production Capacity
Feeding the Beast: Efficient Art/Asset Creation
Questionable Content: Dealing with Spilt Coffee

Sadly, these three roundtable were cancelled due to lack of interest... I'm not surprised about the visual arts session as the chapter has not had much success at attracting artists in the past. But, I am disappointed that folks aren't concerned with production scalability and challenges related to mature content.

October Gathering Pictorial

...here are just a few shots I took during the October gathering (all persons are identified from left to right):

Vander Caballero (EA) leads the discussion from the center, with Matt Sifton (Ubisoft) and ? on his flanks.

Chris Crowell, Remi Hachoud and Florent Dotto, all from Ubi, all have "ideas to sell"... Hmm...

David Benson (McGill) and Elspeth Tory (Ubisoft), being the only woman in the room, discuss the recently released diversity report.

Chapter Sponsors	Partners