Beta start

On tuesday, alpha phase is done and we move onto beta phase. I've assigned tasks for everyone for this beta phase which contains polish, tweaking, and getting things on the evil list checked off. One of my goal is to test the ai boss fights and I plan to have a couple people fully dedicated on getting those balanced.

1 week to beta

I've been adding more functionality for the capital ship and started creating a basic structure for the space station. For this final sprint I will be working somewhat closely with Pace,Jason,and Wilson to get the turrets on the capitalship and the space station, and also get the capital ship working like it's suppose to for arcade mode. For today's lecture, we simply worked on our project and established how the arcade mode will work. Arcade mode consists of 4 rounds, each with 3 phases. After the 4th round, the game repeats itself but the enemies are now more difficult. The first phase consists of enemies spawning and attacking you for a set period of time at which point you can stock up on pods, gain points, etc. The next phase will allow the capital ship to spawn and spawn enemies for you to kill. The third phase will require the player to use their special pods to kill the boss.

Final sprint to beta

I got the capital ship drawing and performing collision detection. This upcoming sprint I will help finish the arcade mode which includes adding more functionality to the capital ship class so that we can accommodate all the features of arcade mode.

Close To Beta

I've been adding space debris and capital ships into the game. Logic wise, everything is ready for the space debris and the capital ships. However, with the new graphics system. I am having troubles getting the objects to draw. By the end of this sprint, the capital ships and space debris WILL be working for arcade mode. I am also assigning tasks for next week's sprint which will focus on finishing arcade mode up which means Ai for the bosses, turrets, getting the death animation and high score to work in arcade mode. We will also focus on adding any remaining polish that is necessary prior to our beta release.

New week of tasks

We are very close to the entire completion of survival mode. Daniel is tying all the loose ends to the end scenario of survival mode in which the player dies, transitions to some sort of game over sequence and then proceed to a high score screen. I've assigned tasks for everyone and if everyone completes their task, we'll be projected to finish in 2 weeks! I have continued to add optimization to our collision code and testing it on the xbox 360 to strike a balance between accurate collisions and not cause slow down in the frame rate. I profiled our current project and I am happy to say that we have done a wonderful job at keeping our code optimized. We predicted the collision would most likely be the most computational heavy portion of our project and we were correct. However, it only takes about 6% of the actual execution time to run. This means most of the work are being done by things we have no control over such as the xna framework. We have also decided to cap the max number of enemy ships at around 15. We reasoned that by the time 15 enemies have spawned, there is no chance for you to survive. Deploying on the xbox, I got absolutely no slow downs with 15 enemy ships all shooting lasers at the player. Our next task now is to implement a mesh to mesh collision system with the larger objects such as giant asteroids and boss ships. Spatial hash is dependent on the fact that your position is separated so only other entities that share similar position as you will be considered for collision. At the same time, the algorithm is efficient because we can ignore alot of empty space. With giant ships, they could take over hundreds of grids in which case, collision becomes very slow since we are looping through so much junk again. Therefore, we will first use a boundinsphere to perform a preprocess step to weed out all units that are too far. Then we can take all the entities that are being considered and perform mesh to mesh collision with the giant object. My primary concern is the amount of slowdown that this mesh to mesh algorithm might cause, but with enough optimization, I believe it should not cause any major problems.

Planning and Coding

So Pace and I got together and looked through the collision code to optimize the system. Having two people looking through the implementation turned out to be much more efficient in this case because we could discuss optimization techniques/ideas and get an opinion on whether it'll work, or the kind of problem the change could create. We deployed on system to the xbox and perform a stress test to find that with grid dimensions of 1000, we were able to fix 150 enemies before serious slow downs. We'll continue to optimize this since a dimension of 1000 is quite inefficient for our system. The spatial hash system is very dependent on the grid size because it separates ships into grids and only perform collision test on objects within the same grid. Hence a 1000 by 1000 by 1000 grid can fit a lot of enemies which would make the collision system very inefficient. I've broken down all our remaining tasks and we should reach beta within 3 weeks. I think if all goes well, we can complete what we wanted in our beta within 2 weeks and spend 1 week to add some back log stuff that we want initially. Perhaps different types of player ships, or a versus mode.