sheepfilms
Xuebing Du
hello vonnie
Mike Driver
Cosimo Galluzzi
RMH
taylor price
occasionally subtle
noise dept.
No title available
cherry valley forever
todays bird
macklin celebrini has autism
No title available
JVL
Three Goblin Art
Origami Around
YOU ARE THE REASON
tannertan36
$LAYYYTER
seen from Germany
seen from Türkiye
seen from Türkiye
seen from Malaysia
seen from United States
seen from Ecuador
seen from United States
seen from Malaysia
seen from Türkiye
seen from United Kingdom
seen from Japan
seen from Netherlands
seen from Mexico
seen from United Kingdom
seen from Brazil
seen from Spain
seen from Türkiye
seen from Indonesia
seen from Netherlands
seen from United States
@lawrencehoward
Dissertation Testing
The past few days have been spent organizing the testing event for my MOG.
Testing will take place this Friday, giving me the Easter break to parse the results through the tool I am developing (leaning towards a MatLab implementation).
I should also be able to complete my write-up, as well as finishing other assignments. There's a lot of work to be done, but I like to keep busy and pushing myself to overcome these challenges.
Test Phase Commence!
My Node.JS server is now recording all player actions relevant to my research, and outputting the data in a tokenized format.
The data is collected from the server in a rather crude manner, it is simply written to a .TXT file server side, and served as a response when accessing the relevant page. Not a very elegant approach, but it's simple, effective, and allows me to focus on my dissertation research.
There are two stages left now, gather players to test the GUI on Mobile Devices, and create the tool to parse the logs for data. If time permits I will also implement a 'Dynamic GUI', and run a second test to compare results, and finally my write-up!
Tools Development
After months of implementation , I have a playable scenario to test feasibility of Cooperative Massively Multiplayer games on a mobile device. I am currently modifying my clients to log and output data, ready to be parsed by a tool I am developing to make sense of the data and provide an insight into: Player error. Ease of interactivity. Qualitative analysis.
Circles! Circles everywhere! The GUI is moving along nicely, and I expect to meet my target and begin testing next week.
Looking through the data I collected in my initial survey, I may not implement the triangular GUI, based on the negative views expressed throughout the results.
I also intend to implement a basic AI to interact with the players, this way I can populate the game and give my human testers more to interact with.
Population will occur to fill up the screen with 20 players, so if I get 17 human testers in attendance, only 3 AIs will be generated.
This week needs to be my final weeks work on the server and client. By the end of the week I need a few options for players to test in regards to GUI, and then I will begin collecting my quantitative data on player performance.
Just a few more GUI to program today!
Further Asset Optimisation
This morning I have implemented the same optimisation fix to the loading of player models. This greatly decreases bandwidth from a client/server perspective, and decreases processing time from a client perspective.
Upon inspection, RAM usage remains unchanged, which contradicts my earlier theory that this method of asset handling would increase RAM usage.
I have one final set of optimisations to implement involving the frequency of messages sent from client-to-server, and, if time allows, a complete reformat from JSON encoded messages, stripped down to binary messages.
Optimisations and Research
As you'll have noticed, updates have been scarce. The majority of my time has been devoted to my literature review for my research.
I have been reading through the various relevant materials. Citing work from Chittaro, Campbell and Tarasewich, and Matt Jones. All of which has been very interesting, and geared towards mobile GUI findings.
In terms of development, I have made some very influential tweaks to the way in which model loading is handled. Before a very poor system was used, in which each time an asset was required it would download a copy from the server. Now one copy of each asset is downloaded, and broken down into Geometry and Materials. These are then stored in a buffer, and can be copied from to instantiate multiple copies of the same object.
This is a massive performance increase, and to give you an idea of the differences, a particular asset weighed in at 2MB, let's say it was required 12 times by each player at a given update, one player alone would rack up a bandwidth load of 24MB from the server - PER UPDATE. 10 players at once would burden the server with a 240MB request - PER UPDATE.
Now that load has been shifted to the RAM of a connected device, impacting the bandwidth by 2MB per asset - PER CONNECTION (Once only), and requiring 24MB of RAM per device, a small portion of the expected 1GB RAM high-end mobile devices carry.
MY CREATION! IT LIVES!
Which is just an over excited way of saying, my dear Rock Monster now does as he is told. Upon command from the server, the Boss fires off one of his 3 abilities. This needs some tuning to follow some kind of basic AI pattern, as currently he chooses one at random. This paves the way for the abilities to actually do something. And by something, I mean horrifically injure our Client Players.
Another day or two and I will begin working on giving the poor Client Players some tools to defend themselves, and launch an offensive of their own. Finally my server/client framework is beginning a game!
*Demo Video to Follow - Watch this space!*
Of course, in the meantime you could always head on over to The Bound MMOG, register, and have a look for yourself!
All Caught Up and Ready To Finish!
My project has seen nothing but negligence these past few weeks. I have had so much to catch up on after my six weeks of recovery, I've had to devote all my time to other assignments.
Today however, I've pushed forwards, and with the help of my 3D teacher from college, James Jackaman (http://laserred.co/) I have successfully rigged my models up to my Motion Captured animations.
The MoCap magic is all thanks to the Brekel Kinect project, and the results (whilst in need of some clean up) are fantastic.
And look, someone dropped a big bad rock monster into the game!
Server Update: Stable Connection Processing
Resumed some work on the server today, and a new build has been published to http://lbhoward.bound-nodejitsu.jit.su/.
The following changes have been made:
Players attempting to log in from concurrent devices will be logged out from previous devices.
Players disconnecting are now correctly removed from connected clients.
Players connecting are now correctly added to existing clients.
My previously planned implementation of more responsive methods of client-server communication have been put on hold. In today's world our connection speeds are just fast enough that responsiveness is not incredibly noticeable. I do hope to find time to implement these, but even on my slow 1MB home connection, there is no noticeable impediment.
My priority must remain on connecting 20 players from mobile devices to test mass-player GUI. This is currently possible with the server as it is now.
The next stage is to implement a method of gameplay. A short combat scenario where players may work together to defeat a terrible beastie!
Development Break (Server Focus)
Realistically, I have reached a point in development where I should stop.
Firstly, the 2D Prototype (GUI Tester) is in a very good state. The Raphael Library provides all the functionality a Mobile GUI could need. Each element has a TouchStart function, allowing ridiculously simple interaction from a development perspective. The only issue is the FPS. Or more specifically, iterating all 20 players on a Mobile Device each frame takes its toll. Aside from needing that fix. It could be ready to deploy within a few days of resumed development.
Secondly, that dreaded Network Improvement on the server. Originally (and currently) - I'm working with an Authoritative Server/Dumb Client model. This isn't a bad practice for slower games, but throw Latency into the mix and you have a noticeably slow experience with faster paced Multiplayer games. This is the main reason for the Development Break. Reading, and lots of it. I need a firm understanding of the concepts of Interpolation, Client Prediction, and Server Reconciliation. I need to develop separate prototypes of these, and then port the existing game onto it.
For this reason, the tone and content of this blog will likely change again. Hopefully I will grasp and implement these concepts quickly, and finally start really building the game.
Prototype GUI Application (Early Alpha)
Today I began some work on my prototype application to test the GUI that will be created from the trends found within my preliminary survey.
There are a few kinks to iron out, and some tweaks to be made to the way the AI behaves. The application scales uniformly to any device size.
AI pieces can only take damage once every 0.5 seconds, and lose 4% of their Health each time they do take damage. These numbers may of course need to be balanced later on. Damage is taken when inside the Red Zones.
The application is web-based, and is built using the Raphael Library. The idea is that after a specific amount of time (or when everyone dies!), collected metrics will be sent over to my database.
The players job is simply to move out of the red zones, whilst tapping the health bars (soon to be added!) to heal that AI's health up a little.
Ideally, I'm interested in things like:
Length of time an AI group member is low on health, without receiving a heal.
Number of AI group members died.
Number of times the player healed a group member with a plentiful amount of health.
I also intend to deploy a second survey which will give players an opportunity to give their thoughts on how they felt the GUI handled.
Login Well and Truly Fixed!
4 hours later and the server isn't denying anyone access to the Database! Hoozah!
I am now finally able to move on to optimising the server-client communication.
Tomorrow also sees my student loan payment, so I will be able to access my survey results one-by-one. Once I have analysed the preliminary trends I can begin building the prototype GUI onto a 2D representation of the game and begin to collect actual player performance data.
Login Issue... Fixed? *Wince*
I can't believe I'm even bothering to post this. It's painful, I'm tempting it to break on me again.
That last issue was a false alarm. The server changed IP slightly and I hadn't accounted for it in my Remote Access on the MySQL server. I think it's finally fixed... I'm praying I don't have to counter post that in an hour. I didn't actually restart the server either, so there technically doesn't appear to be an issue.
The Node.JS server is picking up any toys it throws out of its pram now.
Login Issue - Not Fixed
Angry face! And the guys hosting the Database are just as confused. Persistent connections are enabled and there is no time out.
This is going to be a pain to fix. I don't think I'll be posting on this issue until I have a confirmed fix.
Login Issue and New Bound-MMOG Location
Firstly, the Bound server has now moved and can be found at http://lbhoward.bound-nodejitsu.jit.su/.
As suspected something over at the AppFog site has changed, and is no longer allowing my app access to WebSockets.
NodeJitsu seem to offer a more WebSocket friendly approach, and have Pricing Plans available for projects like mine.
Lastly, the Login Issue which I believed was fixed is now being tested. I'll know within the hour whether progress has been made. Failing this, I am in talks with the support team at XILO who host my Database, so I'm sure we'll come to a solution soon.
Further Login Issues.
Unfortunately, some new issue has reared it's ugly head. The Gods of programming are putting me through my paces today.
Access to the server is technically working, however, for some reason WebSockets is no longer behaving, and the client is reverting to Polling.
This occurs remotely only, hosting the application locally does not produce this reversion.
Reverting to the previous version does not solve this either, it still occurs. Why did I not just attempt something easy! Stupid love of challenges and learning.
Still, can't help but want to throw it out the window right now: "Firefox can't establish a connection to the server at ws:/___________"