Capstone - Week 7

By: James Keats | 18 October 2017

Weekly State of the Repo

For the first time this past week, we failed a challenge in Capstone.

We challenged to get out of Deep Dive and into Proof of Concept last week, and our professor gave us the thumbs down. All is well though, as he gave us a list of very specific feedback and criteria to meet to succeed and make it through to the next stage, and we have addressed all of it and plan to challenge again this week.

Again, the visible changes to the prototype for this week are fewer in number than some of the previous weeks, because a total of 20+ hours on my part this week were spent refactoring and overhauling all of our networked code. Almost everything from the ground up, from the player movement and controls to all of the weapon code, was re-written with networking in mind from the start this time–as opposed to before, when the code was written for a first person game but I had stuck the bare minimum on top of that to get it networked.

In the overhauled framework,we finally actually do hit detection on the server instead of on the client that gets hit. Projectiles actually exist on the server, and their position and collision detection is consistent across all clients. Things like scores and the game time are now properly networked as well. I experimented with making the server authoritative for when the player is able to do things like fire their weapon, but found that the way I had it set up originally was too laggy, making the game nearly unplayable across different networks and still suffering from a noticeable couple frames of lag on LAN. As a result, the client is currently allowed to decide when it can shoot and the performance of its weapon. This is acceptable in a QA environment where we know players will not be cheating, but may be an issue later on. There are solutions to this using prediction, but that would be overkill at this stage.

I also reworked the magnet arm for this sprint, and its performance is now much more similar to how it was in the single player versions. The client applies a local force on the physics object (the gun part) and requests that it be reflected on the server. The client then uses interpolation provided by Unity’s built-in NetworkTransform component to resolve any conflicts between the two. When the part reaches a certain distance from the player, it snaps to their hand, and is locked in that position until they choose to equip, drop, or throw it. Players are no longer able to steal parts from each other with the magnet arm, but can still walk up to the other and equip it directly. The result looks much better than the previous and is significantly more consistent. After a talk with Charlie, our designer, we decided this minor loss in gameplay functionality is worth the gains.

Updated In-Game UI

This new in-game UI is based off of a mockup that Charlie did in Photoshop and then implemented by me. The main goal was to help players understand what different parts are when the models are partially obscured or when the placeholder art is not able to convey the full meaning of a part. Many QA players specifically requested a visual of what the stats of a weapon are and how they will change from a particular part. This panel was the first response we have to that. This coming sprint, we are going to iterate on the screen location of the UI and how it reacts to the player and environment.

Updated Animations

We also did a second pass of our animations! I worked together with our artist, Tyler, to make sure that all the poses worked and blended together as we wanted. Together, we were able set up the animator controller above that is properly called from the gameplay code, although there are still some kinks to work out with networking. For now, the above solution is a significant improvement over the animations we had at the end of the last sprint.

Overall, this was another very successful sprint despite the setback of failing our challenge. Our plan is to challenge again this week and get into Proof of Concept, where we will continue to iterate and move along. There are only about four and a half weeks left until the midyear show–a fact which is both exhilarating and terrifying. Here’s to another successful sprint!

Back to blog