Logbook 1: Gathering research for Master’s project!

So this is the first log book on my exploration in Procedural content generation in AR! This is what my masters will be about and I wish to have somewhere that I note down this thing!

I will write down my meeting notes in bold and the work I’ve done towards them after I have done the work.

Meeting 1 — Main Outcomes:

• Continuing broad research for the next 2 weeks to hone in on what should be project focus.

I have done this mainly by downloading and playing games (mentioned below), seeing where the common issues are technically with AR games on my mobile device (Samsung S9) and also noting down where I think PCG could be used to generate new levels. What I need to do: start writing down PCG methods and exploring implementing them myself!

• Explore current AR games, see how PCG can be used in them, develop test environment to explore different PCG concepts and evaluate.

That is explored below in “AR GAMES EXPLORATION”

• Try focus on one core concept and try not to overshoot expectation of development.

I am still trying to discover what I should focus on, I think I want to definitely explore the design part of PCG and having visually diverse and interesting creations as well as playability, rather than just creating game levels that work. I really want to do PCG creation based on scene recognition so will try and see how feasible that is. Ponder over notes mentioned in “ISSUES I SAW PLAYING AR GAMES” when developing.

• Training on AR development through Unity, Coursera

Still need to view these courses and see if they are good for me. WILL DO THAT TODAY!

UPDATE: 7/02/09: Have looked at the courses, they require money in which I am not willing to part with. So instead, have been looking around the tube and have found this guy who mentioned this discord room where there are a lot of people working on XR development! With information, FOR FREE yaaay! I have also saved some tube videos on my playlist for help with tutorials and ideas. I also have a microsoft edge collections folder where other interesting bits that can help development are kept so yay!

• Read pcgbooks for more insight in basics on PCG

Need to do this today!!

UPDATE: 7/02/09: Have read some of the chapters for brief understanding. Seems like I am leaning towards developing an adaptive (scene recognition) game-level generator, experimenting with different types of PCG algorithms (stochastic, deterministic, generate and test, Neural network evolution?). But this seems like too big and idea so I might need to reduce level creation to terrain or weapon creation. Then create simple mechanics to test out objects created?

• Build yourself test environment and try PCG concept —e.g terrain

This can probably be pushed back until I get a clear idea on the type of PCG I will be focusing on. Must decide on whether I should develop a small game (probably won’t), or is I should develop one level with PCG constraints to generate different levels/scenes. WILL LOOK BACK TO THIS. Ponder over notes mentioned in “ISSUES I SAW PLAYING AR GAMES” when developing.

• Apply different type of PCG and and document it and write your exposure

Same as one above: This is concerning the implementation of PCG in chosen field. WILL LOOK BACK TO THIS.

• Goals

> Try have development of MVP before exam period so main issues can be flushed out and tested (how will I analyse my work?).

> Try thinking about where to get data needed for project

> Try thinking about the steps until the end for project

AR GAMES EXPLORATION

Link to my visuals on actual game play: Research on AR Games on Behance

Ingress — Geolocation based, nice graphics, nice story, what is really interesting about this is that it is not typical AR. I realised there is no real requirement that AR has to involve the camera! As discussed on this reddit thread, although Ingress does not use a camera for augmented reality, by AR definition, it doesn’t need to! As long as it fills the main requirements ( in Ingress case, accurate 3D registration and interaction of virtual objects with real objects), then we AR good to go (haha). It places portals on real world maps! So this is altering perception of reality! Possible uses of PCG: location of portals and behaviours around portals, Jarvis like interaction can have generated speech (which could link to certain portals etc), visual animation around locations of portals.

CubeAR — AR Function relies on QR physical cube you print (very similar to what was discussed in Meeting 1, maybe look into AR marker games? something like AR the gathering?), seems simple enough, rotating cube with maze on it, pick ups and danger zones implemented. Possible uses of PCG: Level generation, cube texture skins.

Egg, Inc — The most fun to play, spend 2+ hours on this game on mobile, constant tapping again NO AR for android as it seems to use ARKit (apple AR toolkit). Possible uses of PCG: Create different types of eggs (texture PCG), creating different types of chickens/drones with different behaviours.

Knightfall AR: Season 2 — Great, you find a flat surface and the game imposes itself there. Beautiful AR scene created is depth dependant meaning one can use their phone to physically move in space to get closer (rather than pinch zoom). Makes the game sooo much more interesting. Click on screen to collect coins and attack. Love how the animation of the characters and detail when you physically move into the view is still great! Really enjoy a full experience without the need to leave physical space or wear VR headset. Possible uses of PCG: Landscape creation, weapon creation, level generation.

Beer Pong AR — Great overlays image of cups on surface available. Swipe up to throw balls in cup. Possible uses of PCG: Cup orientation (for difficulty?).

The Walking Dead: Our World — No AR (I assume it might use ARKit). But I liked the concept of shooting zombies real time through phone. Possible uses of PCG: Zombie type/behaviours.

AR War —Geolocation AR game but with a twist! Builds on AR and brings physical gun accessory that can be used as with phone in FPS game! Concept is really interesting! Similar to walking Dead! Possible uses of PCG: Not sure, probably the same as walking dead.

Monster park AR — AR Dinosaurs, can interact with dinosaur by tapping on them. Change size. Different dinosaurs. Possible use of PCG: behaviour of dinosaurs with items in room, flight patterns of flying dinosaurs.

Jurassic World Alive —Geolocation based AR game. AR working similar to monster park: you get Avatars of dinosaurs you can feed (through pressing a button) or give toy to play. There doesn’t seem to be any sort of hand to AR character interaction. Possible use of PCG: Creating tasks/missions (rewards, what to do for reward), spawning of dinosaurs on location, creating of new species of dinosaur, objects to purchase.

Minecraft Earth — Loading is a bit dodgy on my phone but when it works it is really fun! a mini would of dungeons that work by placing dungeon in user area then traversing to fight enemies and collect materials to build stuff. Minecraft uses PCG already in creating maps with a nice level of detail described here. Possible use of PCG: Map. dungeon creation, weapon creation.

Balanced Tower AR — Seems like it doesn’t need to detect a plane to place tower onto surface, button just places it at centre of screen, which places object in physical 3D space and allows you to rotate around the object in real time. Possible use of PCG: Levels of difficulty (can generate levels which have towers with pieces removed depending on difficulty).

ISSUES I SAW WHEN PLAYING AR GAMES:

• Repeatability for geolocation games came very quick.

This made game fatigue so much higher. Most of the geolocation games I played were veryyy similar and seeing some diversity in game play and look would be amazing and increasing my likelihood to want to play. Also, although there were location based games, there doesn’t seem to be any room specific or even landscape or backdrop location base specific generation. Obviously games like pokemon go make an effort to spawn water type animals near known water sources (as explained here), but other games that don’t have such a extensive environment premade for the game already on placement logistics do not utilises location based or room based generation.

• Hard to detect black surfaces? Camera doesn’t use autofocus to detect based definition to take scan?

Grouped these things together because it all seemed to be AR framework with S9 camera dependant. Will need to take this into account during development. Most games asked for a scan of the floor or area in order to detect plane (often wrong for some reason), so again look into this when developing.