This week based on Halves feedback, we have fixed the difficulty of jumping and the camera angles to reduce the unsatisfied moments for the players. Since during playtesting, a lot of time the playtesters did not care about the hippos and how natural they are. Most of the feedback we got are “It is so hard to jump onto hippo’s back.” “The jumping mechanics is hard.”
We decided to adjust the physics and giving an extra target arrow of the jumping system and change the camera to a third person fixed camera to reduce the dizziness. Afterwards, we also found out that the environment we had right now is a circle pond and when the hippo is choosing different food, the scene is too chaotic and the player could not observe the scene since sometimes the hippo is going around the circles. Therefore, we changed the environment layout to a rectangle size of map similar to the Crossy Road game layout.
We also implemented the rest of the layer 3 and 4 that we had planned. Layer 3 is the size changes when the hippos eat the food they like and dislike. Layer 4 is the switching goals during gameplay.
We also planned for our final products to have hippo demo as our base for the game and different levels show different layers of hippo complexity. As for the theme, we planed to have 4 different seasons while changing the levels. So, you might be thinking what’s now for the chicken demo? We still like the idea a lot and think that it is interesting to discuss about building relationship with agents. However, due to time limit, and a lot of feedback are unsure the difference between state machine and the chicken RL agent. We want to treat the chicken demo as a extra bonus system on top of the hippo game and maybe it served as a guidance in one of the level in the game or tutorials.
On Wednesday, we presented all the above progress. Our client has a different thought of what we are planning and going for right now. They are happy to see that we are planning different layers and levels to show the player the difference of the RL, however, they want us to focus on designing more variation or interaction on top of the current system. Also, they mention fixing the difficult game mechanics of jumping should not be the main focus, instead is there any interesting behavior of the agents?
After the meeting, we had a lot of discussions. Looking at the client requirements and goals, having a polished product is not the main goal for us. It is true that we have not tried more complex design or discuss any interactions between different RL and we already try to jump into using the layer directly. The learnings of iterations should be crucial for our final deliverables. Next week, we rethought about how to add interactions, and what variables or attributes need to be considered.
