Showing posts with label Abertay Uni. Show all posts
Showing posts with label Abertay Uni. Show all posts

Saturday, 11 May 2013

P.U.N.C.H


So this is the showcase for the game we've been working on for our 2nd semester project. The team is called Good Lord! What a lot of biscuits! The team name has a long story about it, maybe a story for another post. P.U.N.C.H is a product of 7 programmers, 1 artist, 1 designer and a producer and was developed using the Havok physics engine.





Limbo: A beautiful and horrible experience

(Artical was originally part of assessment for a narrantology/ludology module for my postgraduate)

I started playing a game called Limbo; it’s a small indie game that’s gotten a good reputation from fans. It’s a beautiful looking game that sets the environment up perfectly, with its minimalist scenery and monotone colours. Everything on screen matches the beautiful but depressing path the boy character follows.

But what surprised me the most was the amount of attachment I had gained in the first 60seconds of gameplay. A boy wakes up in a forest. The smooth running animation glides you across the grass to the next scene. My first reaction was ‘awww how sweet, I’m going to love this game’. I see something on the floor, I go to investigate but get too close. The bear trap snaps shut around the boy’s body which ricochets like a rag doll with the force, decapitating him with a small squirt of blood. It happens so fast it makes me jump. My mouth just dropped, I was in complete shock and within 60seconds of me starting the game, it had managed to make me feel guilty. I shouted ‘IM SO SORRY!’ at the little boy. This theme continues throughout the game but never seems to get old. I felt so attached to this boy; I felt so much responsibility towards his survival. When he fell from a tree and broke his legs, when he got flattened by a rolling boulder, when his convulsing body drowned in a small pond, I had exactly the same reaction each time.



It’s got nothing to do with the graphic deaths. I've played games with much more gore than this. Nor has it anything to do with the death of a child, this concept isn’t one of much horror to me. It is however completely down to the atmosphere the environment makes. The simplicity of each scene draws all your attention to the death, it allows for no visual dilution. The smoothness and fluidity of the boy’s animations are in contrast with the large, sparse surroundings making him seem far more vulnerably and human like. The death events can happen so quick that in less than a second you’re left with nothing, and the boy always dies in one hit. I think this is a fantastic example as to how effective spatial use can be in games, compared to any other medium in which it barely exists at all. This game would not be viable in any other format.

Saturday, 30 March 2013

Elephant Algorithm: They Never Forget


At the moment the elephant memories are represented by a 2 dimensional array but I was actually thinking I might put different types of geographical locations in other that just the oasis, and use a 3 dimensional array for the elephant memories. The 3rd index being a 'tag' to distinguish between the type of location. This might end up as polish though as I want all the basics in by the end of this weekend so I can start on the crowd  algorithm.

Elephants can now store information on where the oasis' are located. This is done by calling a remember function on collision, which stores the current x and y position. An elephant will continue to store coordinates as long as there is collision. The result is the storage of not only the geographical location but also its partial coverage (the area in which the elephant travelled). Originally only one coordinate was going to be stored but this way has given the memories more depth and given the algorithm more functionality, without increasing the complexity. I'm very pleased with how this has worked out even though it isn't as planned. However it is currently using arrays 3600 in size, eventually this should be changed to a vector to increase efficiency.  

I would also rather the arrays size was set on the fly at run time based on the number of various locations, which would be randomly placed and in randomly set quantities. Probably with a limit to the amount of locations present based on the screen resolution. The amount of elephants could be generated at run time too. It's likely that these features won't be in the build I hand in in May but I have enjoyed working on this project and so I will carry on working on next semester.

PS2 Project Proposal


I handed in my project proposal for the PS2 module. I plan on allowing the player to 'cast spells' by shooting a fire ball at enemies and the enemies will catch fire too. I'll need a model of some kind to represent the enemies and ill need a chase function so they chase the player, then some kind of collision detection, box probably, nothing fancy. The fire ball will be the main feature as I want it to be an animated texture on a sphere model. I've done this before in my Direct X module last semester so I understand the theory. I need 3 textures of varying opaqueness and layer them on top of each other over the sphere. Then I have to cycle each texture across the surface at different speeds.

Sunday, 24 March 2013

Elephant Algorithm Project


It was my intention to finish the 1st algorithm for my innovation project by the end of the month. This leaves me with this weekend and next weekend to complete it (at least the code, write up can come later). Managed to get the collision between elephants and oasis in with no problems. Elephants now know when they are at an oasis or not, they just don't do anything with the information. Next I have to store the information so the elephants can use it.

Narrative Presentation


Spent most of my time this week working on a presentation for my narrative module. The task was to analyse the narrative of a game of your choice to an academic level. So I posed the question 'does having a small development team have an impact on narrative quality'. I chose the game Penumbra for this specifically because the company behind it (Frictional Games) numbered just 3 at the time of development and the game originally started off as a tech demo for their in-house engine. All graphics, audio and writing was outsourced. Even now the company has only 5 employees and the rest is outsourced. But the 4 games they've produced so far could be mistaken for AAA games. Anyway, the problem i had is that i gathered loads more information than i needed too. Then found i had to cut loads of it out to fit it into 8mins. So the end result became a jumble of information that didn't necessarily relate or prove anything. The presentation itself wasn't grate either as this my skills in this area are weak. This aspect of the assessment is only worth 20% so im not that bothered and the good news is that i am going to use all the information i gathered to form my essay for this module, worth 60%. So less work for me.

Tuesday, 12 March 2013

Elephant Algorith Project


Started on my innovation project, it's all going to plan so far. I've got the basics for the elephant based algorithm demo in. Elephants move around on screen on their own and the geographical locations (oasis) are in too. The next step is to give the elephants a memory back which will probably be made out of an array. Once that's done I have to work out how to transfer data both ways when two elephants have collided. This shouldn't be too hard anyway as the elephants have their own class and so a change to one will change them all. It's more a case of, arrays can confuse me a lot and collision can be a nightmare. Once these two features are in though the rest should be straight forward. Then that's one algorithm and demo down and one more to go. The aim is to get this one finished by the end of this month, and the second one finished by the end of next month, leaving me 17 days until the hand in for polish, paper work and to work on the 4 other modules that are due in on the same day.

Monday, 11 March 2013

DirectX Terrain Smoothing


If added smoothing into my DirectX. The theory behind it is quite simple. Work out each points height by averaging the neighbouring heights. The problem i had was that the framework works out the height terrain with a single dimensional array instead of a multi dimensional array. So i had to use a triple nested for loop just to get it to work.  One for each pass through the smoothing (the more passes you make the smoother the result is), one for the height and one for the width.

Friday, 1 March 2013

Havok Project


This week we started working on the new prototype for the house wrecking game. I worked on making objects randomly drop a coin when it’s smashed. This code will then be used to update the score when a coin is collected. I had to wait to put my code into the project because it relied on a function that someone else was writing. Rather than waiting I went ahead and wrote the code as I knew it wouldn't take long. I then translated it into Lua as I didn't know if the code I was waiting on would be in a component or a script.  Testing ended up being peer programmed because of version control issues, but largely remained the same. I'm not required too but I'm doing to go beyond my task and try to make a ‘COIN’ object actually appear when an object is smashed rather than it just displaying ‘COIN’ to the screen. I have no idea how to create objects from the code alone without using the engine but it seems like it would be something were going to do a lot of and this feature would be needed anyway so it would save time if I did it now.


Lua, from the little I’ve done, seems to be quite a nice language. I was expecting to find the syntax difficult and to get confused between languages as I am working with two. The version control however seems unnecessarily complicated. I’ve never used any version control before so I am unfamiliar with it. But setting it up was a real hassle and I still doing really know how to use it. A lot of us are having problems using the interface.

Friday, 22 February 2013

PS2 Development


This week I started on PS2 development. I've only been doing the controls so far but I'm really enjoying it. I've covered all the all the buttons, analogue sticks pressure and vibration, but what I really want to do is make a small framework with all the controller features mapped so I can use it for my project. I have to wait until the room is free though. I don’t like not having the equipment available for me to do it at my desk or at home, I'm not use to it. There is one thing now ingrained in my memory though. At the end of the program the controllers don’t automatically unlock from the mode they've been set in (analogue/digital). So you have to remember to unlock it at the end of the code. Whilst implementing the vibration controls I hadn't yet found this out. I ran the program and the controller started to vibrate. But because the controller was locked in analogue mode when I closed the program it carried on vibrating. There was an awkward 15 mins involving 3 people working out how to stop it vibrating and me constantly holding the controller in the air to stop it making a noise against the desk. 

Thursday, 21 February 2013

DirectX Framework


I also got given the new framework the PC and Xbox module. I’ve gone through half of it to make sure I understand what’s going on and made my own comments etc. I noticed the controls were, at least to me, very counter intuitive. So I’ve started converting them over to a more FPS look-move style control. Putting mouse control in was easier than I expected. The camera now points at the mouse curser but now I have to get forward to go in the direction that the camera is facing. There are two tutorials I’ve found but they look really complicated. It looks like I’d have to modify and add a lot of code to make it work. I think I might have under estimated the task and it might never actually get implemented. This means I could end up taking the camera look controls out too.

Monday, 11 February 2013

DirectX Randomized Terrain Generation


Started implementing procedurally generated terrain when the space bar is pressed for my directX module. I wrongly assumed that the code to procedurally generating stuff would be long and complicated. But you can produce basic mountains with only 2 functions. Randomly generating the terrain was literally as obvious as using a random number generator on the y axis.  Mine are really spikey/ridged mountains so I want to add some smoothing to make them look more realistic.

Monday, 3 December 2012

Ants and Bees


ANTS
Ants find the shortest path from their nest to food and adapt to changes in environment all without visual sensing. They will instinctively go for the path with the strongest pheromone and will utilise information gathered from previous expeditions to guide searching when pheromones have deteriorated. This is how they cope with environmental changes. They do this by using multi agent cooperation and use pheromones to modify the environment for indirect communication.
Technique
·         Leave the nest
·         Check for pheromone
·         Follow path with strongest pheromone OR check information to fined the last place there was food AND take the shortest route (shorter routes should have more pheromones)
·         Walk in that direct leaving pheromone as you walk
·         Check the path travelled to count the length
·         Find food and return to nest
Using an ant algorithm allows for minimum individual intelligence and self organisation to complete complex tasks.

BEES
Another algorithm inspired by insect behaviour is the bee algorithm. Developed in 2005 its goal is to work out the best solution to a problem. Its essentially a form of swarm/herd  intelligence because it mimics a group of individuals working together to solve a problem. It can be very useful in finding the best option when there are lots of possibilities to choose from.
Scout
Send out some scouts that randomly search trying to find things that will be useful (relevance, effectiveness etc.), they then goes back to tell the hive which then decides if it's worth taking a better look based on which is most likely to have the best resources out of the array.

Monday, 5 November 2012

Scripting


Why use a scripting language instead of hard coding it in to the engine?
Well most game logic can be scripted keeping the game code and the engine code separate. This way its easier to find stuff and modify plus you can re-use the engine code for another game. There are other pluses too. In the industry it can take quite a while to see your changes in the game because you would have to recompile the entire game each time (heavy rain had 4 million lines of code). But if you use scripting you can see the effects in game immediately. Even more efficient is the possibility to teach non programming roles how to modify bits of code so they can tweak it themselves without having to involve a programmer.

Making realistic actors
In a film or book you have no control over the content and so there is no agent. You watch the characters make decisions and stick to their roles which gives the illusion of personality. However in a game you have more control due to the interactive nature of the gameplay. This means your the agent, you make all the decisions which is why, even though the characters history might be known, you do not have any attachment to the character your playing. In the Sims a mother is playing with her baby. You tell the mother to feed the baby. The mother then puts the baby on the floor (to end the last animation) and then picks the baby up to feed him (beginning the next animation). This is very unbelievable!

So how can we make believable actors if they don't make any decisions?
The key to this is believability. Although it must be understood that there is a difference between realism and belief. Bugs Bunny is a believable character but he certainly isn't realistic. The idea is if you can imagen what that character could be doing now then it must have a believable personality. However if you can only think of the character as a empty husk then the believability has obviously failed. Again if you have made all the decisions for the character than you have nothing to latch onto about the characters personality. You have to be able to create characters that feel like they have a responsibility over their own actions and yet allow us to also be responsible.
One way to fix this is by using procedural movement and behavior. Combine multiple movements by overlapping the animations to produce a more believable action. A good example would be to walking and reaching for something in a cupboard rather r than walking to the cupboard, stopping, then reaching for the door. You can also use pseudo-random movement to create a continuous presence. Such as shifting weight or blinking. The point is use more non linear animations.

Monday, 15 October 2012

Finite State Machines


I read AI for Game Development, the chapter on finite state machines (a book in the recommended reading list). In which it explains a simple model for a state machine depicting the ghosts from Pac-Man. I decided to update the example using a situation common to a more modern game. Imagine a guard patrolling a camp. His behaviour can be depicted using this same FSM model. 



There are 3 states and 8 transitions. When all the guards in the camp have this kind of behaviour it can appear quite complex and give the illusion of intelligence. If we assume the patrol state is the default state which will continue while there are no enemies, when a new enemy is seen it will switch to the attack state and remain there while the enemy is close by. Once he has lost too much health he will then call for back up by switch to the final state. If his health has regenerated enough or he has been healed etc,  he will then switch back to the attack state and continue the battle. Once all enemies are disposed of the guard will return to the default state to patrol his camp.

Monday, 1 October 2012

Can we use the Turing Test for Games



Although the Turing Test is known to be a measure of human intelligence it is never referred to as comprehensive anymore. Instead it is thought of as insufficient to cover all aspects of what we now consider human intelligence to be. However games are not required to actually be intelligent rather they demonstrate the illusion of intelligence to the player(s). For this reason any result of an AI test in a game setting should not search for a result that indicates its intelligence but instead should look for overall gameplay improvement. Therefore a Turing style test focusing on believability rather than intelligence could be applicable in the game industry assuming it was part of a more complete test.
Games are currently known for their ‘bad’ AI where situations occur like killing a patrolling guard who dies loudly but doesn’t attract the attention of the guard a few feet away. Or a persistent opponent that attacks your base in the same place each time despite being constantly defeated. These unrealistic conditions can take the player out of the game but implementing behaviourism could help with this. Using neural networks as a method to train agents  or ‘condition’ them may be a solution. AI opponents can then learn from their mistakes and attack bases from other angles to avoid troop loss for example. Behaviourism relies on the theory that only the observable behaviour should be analysed rather than the internal thought process behind the behaviour. This is useful in the industry because all the player sees is the behaviour, we don’t necessarily need to know why the AI reacted a certain way, just that it did. Behaviourism is currently being used to train speech/conversation AI to beat the Turing test and win the Loebner prize and is measured by using lexical semantic analysis. So as academic research progresses so might the AI in games. Although this technique isn’t frequently used in the game industry yet, it is becoming more common.
Theoretically any game can be tested on this kind of test because every game is a form of imitation game. For example can a player can tell the difference between an AI and humans when running around a virtual world. Or can a player distinguish whether it is playing against a human or an AI opponent in a strategy game. But its not as simple as that. The problem with testing for believability is that it’s subjective. People can have altered perceptions based on cultural differences. As discovered by Mac Namee, 2004 who came across the following scenario:
‘In one test, subjects are shown two versions of a virtual bar populated by AI agents. In one version the behaviour of the agents is controlled in a deliberate manner because they try to satisfy long-term goals – buy and drink beer, talk to friends, or go to the toilet. In the second version the agents randomly pick new short-term goals every time they complete an action. One Italian subject felt that having agents return to sit at the same table time after time was unrealistic, whereas the other (Irish) subjects mostly felt that this behaviour was more believable. Sadly, further tests to determine whether this – or other – results are indeed culturally significant have yet to be carried out; the possibility that such differences exist does appear quite likely, however.‘
Similarly Mac Namee also points out that players new to the game can have a different experience of the AI due to lack of exposure of the game environment compared to a veteran of the game. The veterans know exactly what their looking for and have more time to concentrate on the AI  because other game related skills are second nature.
It is possible that the player can be concentrating on the game so much that they don’t notice the AI. Or in some genres they may not get enough interaction with the AI to make a firm judgement. A way to fix this is by making the AI more obvious. Butcher and Griesemer (2002) found that players need to be given exaggerated visual cues in order to even notice the AI. But surely if the AI is exaggerated and for example over emotive it becomes less realistic and less human? On the other hand it is possible to overdo the AI and confuse the judge by over exaggerating a common mistake that humans experience. This is exactly what McGlinchey and Livingstone (2004) found when their Pong playing AI moved the bat in a slow, jerky way (by accident) players often mistakenly thought it was human. Although some people suggest having an observer as a judge to alleviate these issues it should be remembered that the aim is to please the player not the people watching.
Laird and van Lent, 1999 conducted a test in which the observers judged of players of mixed skill play against the Soar Quakebot and found that the opponent was considered ‘more human’ if it had a slower reaction speed, an inaccuracy when shooting a target and tactical reasoning. (However none of them thought the AI was actually human). The results of this test led them to devise 3 simple principles:
·         AI should have human like reaction speeds not super-fast
·         AI should not have ‘superhuman’ abilities like overly accurate aiming
·         AI should have some kind of strategic reasoning so it doesn’t just react to each situation
One of the obvious flaws of the original Turning test was that it gives a very black or white result. It either passed or it hasn’t. But in a creative industry there needs to be more flexibility. There have been attempts that score the AI on a scale of different features.  Using the following parameters formulated by Daniel Livingstone by analysing game AI for common ‘bad AI’ situations it could be possible to devise a series of tests of believability to cover different aspects.
·         Plan
o   demonstrate some degree of strategic/tactical planning
o   be able to coordinate actions with player/other AI
o   not repeatedly attempt a previous, failed, plan or action
·         Act
o   act with human-like reaction times and abilities
·         React
o   react to players’ presence and actions appropriately
o   react to changes in their local environment
o   react to presence of foes and allies
It seems to me that a Turing style test has many flaws and many researchers seem to advocate it so much that they are just desperate for it to work. Would it be accurate to say the only reason this kind of test is such mainstream knowledge is simply that there is no alternative to replace it? Maybe researchers should stop trying to find a modification that works and start again in the search for a suitable comprehensive AI test. In conclusion I feel that a Turing style test is inappropriate for use in a game sense, even as a part of a more complete test, because of its problematic testing parameters and ambiguity of results.

BUTCHER, C. AND GRIESEMER, J. 2002. The illusion of intelligence: The integration of AI and level design in Halo. Presented at the Game Developers Conference (San Jose, CA, March 21-23, 2002).
LAIRD, J. E.  AND  VAN  LENT, M.  1999. Developing an artificial intelligence engine. In Proceedings of the Game Developers Conference  (San Jose, CA, Nov. 3-5, 1999).
LIVINGSTONE, D. 2006. Turing’s Test and Believable AI in Games. (PDF) Last accessed: 26/09/12. Available online at: http://www.acso.uneb.br/marcosimoes/Arquivos/IA/games_3.pdf
MAC NAMEE, B. 2004. Proactive persistent agents: Using situational intelligence to create support characters in character-centric computer games. Ph.D. dissertation, Dept. of Computer Science. University of Dublin, Dublin, Ireland
MCGLINCHEY, S. AND LIVINGSTONE, D. 2004. What believability testing can tell us. In CGAIDE, Proceedings of the Conference on Game AI, Design and Education  (Microsoft Campus, Redding, WA).
TURING, A. M. 1950. Computing machinery and intelligence. Mind LIX, 236  (1950), 433-460.
WETZEL, B. 2004. Step one: Document the problem. In Proceedings of the AAAI Workshop on Challenges in Game AI (San Jose, CA, July 25-26, 2004).