I love GDC. This is my second year attending the conference, and I doubt it will be my last. These are some blog/journal entries I made during the conference a week ago. I'm cleaning them up and posting them here to help a reader understand what they can get from GDC on a professional/educational level and to remind my future self what exactly I did each day (and to actually have some content for this blog). All words like "today" or refer to about a week ago from time of posting.
Yesterday was the animation bootcamp. Jalil Sadool of DreamWorks gave another great talk on performance, acting, and how to acquire and use reference (something I need to get better about). He had great examples from Guardians, Avatar (blue cat-people), and Rise of the Planet of the Apes. He tried to update the original 12 principles of animation with 4 modern ones: creativity, logic, method, and critique. I'm not sure how applicable this breakdown will be for me in the future, but I'm glad to have a new perspective to consider. He had some great 3d eye reference, though, accompanied by a breakdown that sorts eye movements into looks, darts, and even smaller 'keep alive' which he demonstrated on Tooth and Neytiri.
Mike Jungbluth from Zenimax Online tanked about techniques he uses to help understand different types of NPCs based on Maslow's hierarchy of needs, relative intelligence, and x-y plots of character traits. He found organizing NPCs in this way made it easy to compare them to each other and to the player. Keeping NPCs in diverse niches helped him to fill out the non-player world in Elder Scrolls Online.
Mariel Cartwright, lead animator on Skullgirls gave a talk on how to get good hots and clear actions in spite of -- and sometimes because of -- frame restrictions. This is very close to my heart. At SCAD, I often saw students and professors doing things on ones that really could be done on threes. She gave the first good advice I've ever heard on getting anticipation into game animation (get what you can; even if it's one frame, it will help sell the motion). She also advised favoring your keys (classic) and using follow-through to fill gaps left by a lack of anticipation. Smears help sell swings and overshoot helps sell directional movements in the frame before the hit frame (before this advice, I would have thought to put the overshoot on the hit frame), this is especially useful when programmatic hitstop freezes the screen -- try not to let it focus on a weird-looking pose. Also, never forget what you're making. Motion is more important than the pose, and gamestate clarity is more important than the motion. For further research, she recommended Zweifuss and Fighters Generation.
Next was Kristjan Zadziuk of Ubisoft; he gave a talk on getting performance inspiration, wrangling the many many animation states, and fitting it all into memory. Who knew they only have 23mb (!) for all their animation? He also gave some advice for when you're going for a mocap shoot: have as few people as are necessary, have clear direction, collaborate, respect the talent (opinions and wellbeing), and have an ambitious list of shots to shoot (it lets you skip a shot that isn't working). He also advised one not to take more than 5 seconds to do any action that the player can't interrupt.
Jay Hostfelt gave a talk on the role of animation and art in prototyping for AAA. I love that he made a strong case for generalist members of a prototyping team. It helps keep teams small which simplifies communication and expands flexibility when those features are paramount. A good, well rounded artist/animator can help keep a prototype from being distractingly ugly, provide timing to basic actions, and help clearly communicate gamestate to the player. Go for simple direction that will fit the graybox levels and unpolished mechanics, and be sure that the team is made of members who are okay with having their work thrown out -- you can't get attached to the prototype, it only exists to be thrown out. Jay advocates we try to go back to thinking of ourselves as "game developers" instead of just representatives of our individual disciplines.
Then came my favorite talk of the day: David Rosen of Wolfire giving a talk on programmatic solutions to solving animation production problems. Most of the animations in Overgrowth are apparently made of individual keyframes that are interpolated programmatically using bicubic and spring blends. The entire locomotion branch of the state machine is something like 13 frames! He also went into how these techniques were used in receiver. I had never considered you could get so much movement out of so little data. It makes sense that this sort of logic would flow from the mind that does so much animation work (leaning, jumping, ik, looking) programmatically whenever possible. I'd love to see how this workflow would scale up to a larger production that required more nuanced acting and motion.
David Rosen talked about using layered animations in bf4 to support dlc weapons without breaking the existing animation budget. Not to diminish the work of the DICE team, but when compared to the previous talk, AAA problems start seeming like a misallocation of resources.
Animating cameras. Simon Unger talked on animating the camera in a game. He focussed a lot on using reality-inspired camera techniques and rigs to help maintain a believable presentation. He also recommended keeping the horizon as stable as possible so as to mimic the stabilization that our eyes usually do. Doing otherwise can cause motion sickness.
Pixar / costume quest. Finally, Tasha Harris talked about her experience going from animator to game director on costume quest. She touched specifically on using visual and playable examples to articulate vision to the team (for optimal clarity). She found dailies to be great for quality and morale especially if the group got in the habit of acting out suggestions. She also found it important to let empower the team and trust them to do things.
After the talks, we all wandered to the wrap up room to ask follow up questions and I got to chat some more with Mariel Cartwright about 2d bone animation, game jams, and eventually voxels. I also got to talk to an animation director (sorry I forgot your name) from EA who's working on UFC. There are a lot of interesting technical and political hurdles to deal with in a 'realistic' sports/fighting game like that (wouldn't want those punches to be too snappy, that's not realistic), and it was good to hear some inside perspective on it.
On my way back to the hotel, I got to talk a little more with David Rosen. His work is really inspirational to me and it was great to be able to tell him the impact he's had.
No comments:
Post a Comment