This post was originally written by Matt Kidd for the Incubator Games website.
While some movement was best handled programmatically, Feeding Time's extensive animal cast and layered environments still left plenty of room for hand-crafted animation. The animals in particular required experimentation to find an approach that could retain the hand-painted texturing of the illustrations while also harkening to hand-drawn animation.
An early pass involved creating actual sketched frames, then slicing the illustration into layers and carefully warping those into place to match each sketch. Once we decided to limit all the animals to just a single angle, we dispensed with the sketch phase and settled on creating the posed illustrations directly. When the finalized dog image was ready, a full set of animations was created to test our planned lineup of animations:
The initial approach was to include Sleep, Happy, Idle, Sad, and Eat animations. Sleep would play at the start of the stage, then transition into Happy upon arrival of the delivery, then settle into Idle until the player attempted to eat food, resulting in Sad for incorrect choices and Eat for correct ones.
Ultimately, we decided to cut Sleep because its low visibility during the level intro didn't warrant the additional assets. We also discovered that having the animals rush onto the screen in the beginning of the level and dart away at the end helped to better delineate the gameplay phase.
There were also plans to play either Happy or Sad at end of each level for the animal that ate the most and the least food. The reactions to this, however, was almost overwhelmingly negative! Players hated the idea of always making one of the animals sad regardless of how many points they scored, so we quickly scrapped the idea.
The Happy and Sad animations were still retained to add a satisfying punch to a successful combo and to inform the player when an incorrect match was attempted. As we discovered, a sad puppy asking to be thrown a bone (instead of, say, a kitty's fish) proved to be a great deterrent for screen mashing and worked quite well as a passive tutorial.
While posing the frames one by one was effectively employed for the Dog, Cat, Mouse, and Rabbit, a more sophisticated and easily iterated upon approach was developed for the rest of the cast:
With both methods, hidden portions of the animals' faces such as teeth and tongues were painted beneath separated layers. In the improved method, however, these layers could be much more freely posed and keyframed with a variety of puppet and warp tools at our disposal to make modifications to posing or frame rate much simpler.
For Feeding Time's intro animation and environments, everything was illustrated in advance on its own layer, making animation prep a smoother process than separating the flattened animals had been.
Because the background elements wouldn't require the hand-drawn motion of the animals, our proprietary tool "SLAM" was used to give artists the ability to create movement that would otherwise have to be input programmatically. With SLAM, much like Adobe Flash, artists can nest many layers of images and timelines, all of which loop within one master timeline.
One advantage over Flash (and the initial reason SLAM was developed) is its capability to output final animation data in a succinct and clean format which maximizes our capability to port assets to other platforms.
Naturally, SLAM stands for Slick Animation, which is what we set out to create with a compact number of image assets. Hopefully 'slick' is what you'll think when you see it in person, now that you have some insight into how we set things into motion!