Tag Archives: Game Design

Jetboard Joust Devlog #100 – Die and Try Again!

This blog should really have been Configuring Things Out pt. 2 but I thought ‘Die and Try Again’ was more appropriate really.

Now all the enemies and weapons are finally done I’ve been going back over the game worlds, playtesting and (re)adjusting all the various character/weapon stats and levelling rates. The only way of balancing the weapons is to simply play the game a lot, and if I feel I’m either avoiding particular weapon because it’s too weak, or that I’m constantly gravitating towards a particular weapon because it’s too overpowered, I tweak the stats as appropriate. It’s been a very time consuming process and I have to say I’ve been at something of a low ebb whilst doing it.

Unlike fixing a bug, you don’t ever get to the point where you think something is ‘done’, and I know I’m going to have to go through this whole process again at least one more time to refine things further before the game is complete. It’s really getting to me now.

The thing that’s made this phase so much tougher than the first configuration ‘pass’ is that I’ve been through every single boss and mini-boss fight in the first four worlds and attempted to adjust the difficulty for each. There are over fifty of these and, by nature, they are supposed to be tough, so this entails a *lot* of failure both in terms of dying in-game and in terms of setting the difficulty level either too easy or too hard.

Most of these challenges are mini-boss battles featuring the Guardian enemy and a new weapon unlock or upgrade. To maintain a certain amount of narrative ‘flow’ (and to stop me being lazy) I’m having the enemies guarding the unlocks armed with the same weapon they are protecting. The strength of the enemy is heavily dependent on the strength of the weapon, therefore a tweak to the stats of said weapon (because I think it’s too weak or too strong in-game) has a knock-on effect in every single battle that features it. It’s a cyclical process and one that I’m starting to think could go on forever, like painting the Forth Bridge. On a good day though I see it more as a kind of gradual ‘whittling down’ with every pass getting me nearer to the ideal balance.

It also makes a difference what weapons are available to the player during these battles. I was choosing these weapons completely randomly (but based on a consistent seed) though towards the end of the process I realised this wasn’t working and I needed a better balance of weapon choices. Consequently I’ve divided the weapons into four categories – long range, short range, wide range and explosive, and I make sure a balance of weapons from each category is chosen. I think I may ditch the consistent seed as well so each time you enter a level the weapon choice is a lottery, albeit a balanced one.

There’s also been a myriad of bugs and other gameplay tweaks I’ve made along the way though at this stage I’m trying to restrict myself to fixing bugs/tweaks that actually affect the balance of gameplay and noting everything else down on Trello cards for later attention. There’s currently around fifty of the Trello cards (groan) and there’s at least another ten things I haven’t noted down yet!

I’ve also been back through the first four major boss fights and made a proper attempt to balance the difficulty of these with the (rough) stats the player should have achieved by that point in-game. If you’ve been following this blog you’ll remember how much time I’ve spent on these boss fights already, so it’s been doubly demotivating having to go back to them, debug them again, run through the fight procedure again and again AND again whilst discovering yet more bugs and issues that need fixing. On a positive note though, it’s been nice to revisit them in different colour palettes.

Lastly, and I know this is going to sound like archetypal first-world moaning, I’ve been surprised how physically exhausting this process has been. Basically playing (often ludicrously difficult) boss fights for eight hours a day for eight or nine days straight takes its toll. I try to be as careful as I can be about posture but my shoulders ache, my wrists and elbow joints ache, and my back is fucked. Yes, I know it’s not exactly working down a mine, but being physically in pain doesn’t do wonders for one’s motivation. I can’t wait to finish this game and take a break from this shit.

Oh yeah – one thing that’s kept me going through this process is the fact that I can just turn the game sound off and listen to music. I’ve finally succumbed to subscribing to Spotify and have been listening to a lot of Tangerine Dream. Strange coincidence – I finally got round to watching Netflix’s enjoyable Bandersnatch interactive movie at the same time and what does that feature? A ZX Spectrum game developer slowly going insane trying to finish a game whilst listening to lots of Tangerine Dream! Weird…

Right, on to those Trello cards…

Dev Time: 8.5 days
Total Dev Time: approx 268 days

previous

mockup_3x
The ‘Spinner’ Boss In Colour

mockup_3x
The ‘Squirmer’ Boss In Colour

R.P.G. Guardians. Die and Try Again – and Again, and Again…

Jetboard Joust Devlog #97 – Creepy Crawlies!

No prizes for guessing the classic arcade game that’s the inspiration for this latest enemy – yup, it’s another of Atari’s masterpieces – Centipede! Working title for this enemy is the ‘Scuttler’ (I already have a ‘Crawler‘ and a ‘Squirmer‘)!

The mechanics of this enemy are pretty simple, I thought the hardest thing to get right would be the algorithm that makes the segments ‘follow’ the head (I’ve had to right similar code in the past and got myself into a right mess) but the code I came up with, unbelievably, worked pretty much right of the bat!

There’s probably a better way of doing it but my basic approach here is to ‘remember’ the direction each segment is travelling and to continue moving in that direction by default each frame. If the total horizontal and vertical distance between one segment and the next is less than the desired segment spacing no movement occurs. If the segment aligns horizontally or vertically with the segment in front we switch orientation (i.e. from horizontal to vertical or vice versa). This seems to work well enough for my purposes but if anyone has any better ways of doing this I’d be interested to hear them as it’s a gamedev problem I seem to run into quite a bit.

Unlike the Atari Centipede I don’t have any mushrooms to run into to initiate a change of direction so I had to improvise a bit here. Changing direction when it hits buildings was an obvious one, but I also have it switch direction when it hits the edge of the screen (i.e. camera area) and, with a certain amount of leeway, when it aligns with the player on the opposing axis. This approach seems to maintain an authentic ‘Centipede’ feel whilst working within the confines of the Jetboard Joust gameplay.

I also added a slight ‘sway’ to the segments as they move as a fixed horizontal or vertical movement just seemed too ‘static’ in context even though it would have been truer to the original game. I want to tip my hat to these classics rather than slavishly replicate them.

Of course I also had to have the centipede splitting into two when it’s health is reduced which means things can get pretty manic (in a good way, though I’ve toned it down a bit since this video as things were getting too out of hand too quickly).

I’ve also been working on a Centipede style retro arcade palette but have been running into a few issues trying to get this to look good across all sprites. The red outline you can see is used on some of the sprites in the original arcade game. I like the way it looks here as I designed the sprite around it but it looks terrible on many of the sprites I’ve already designed so I think I’m going to have to use a more generic approach. If I ever make another game I’m going to make sure I treat my outline colour as a completely separate part of the palette – lesson learned!

Dev Time: 2 days
Total Dev Time: approx 253.5 days

previous

mockup_3x
Close Up Of The Scuttler – Centipede Tribute Palette!

Jetboard Joust Devlog #95 – Configuring Things Out pt. 1

This is one of those devlog entries that seems kind of dull because there’s been a lot of ‘under the hood’ type goings on and not a lot of eye-candy to show for it.

But, it’s taken time and it’s really important to the game’s development so I’m going to blog it anyway, boring or not!

What I’ve been focussing on is the overall structure of the game world and how difficulty progresses throughout the game. At this stage it has been very much a first pass at this, as much about providing the tools to allow me to tweak gameplay efficiently as it has been about balancing the gameplay itself.

I’ve broken down what I’ve been doing into three key tasks:

1. Code Refactoring
As with any game of significant scope, there are multiple interconnected parameters that affect gameplay difficulty in Jetboard Joust and it’s a hell of a lot easier to tweak things if the code that manages these parameters is in one place rather than split across a myriad of individual class files. Consequently I’ve set up a static ‘Config’ class that contains all the algorithms and configuration parameters for pretty much anything to do with rewards and difficulty throughout the game. This has involved a lot of tedious cut and paste but I know it’ll be worth the effort in the long run (it already has really). I’ve set up generic parameters here for the amount things like enemy health and weapon damage/difficulty scale throughout the game so at least I have a baseline to work with and can tweak individual scaling from there if necessary.

2. Mapping
I’ve created a template in InDesign for mapping out levels in each of the game worlds and have been through this with a first pass attempt at placing weapon unlocks and new enemy ‘reveals’ in each. There seems to be enough content to fill four level ‘pyramids’ of around twenty rows each with a reveal rate of a new enemy or weapon every couple of rows. I need a couple of different ‘non-jetboarding’ enemy types but was expecting that anyway, I added an additional jetboarding enemy to span the difficulty gap between the ‘minion’ and ‘master minion’ which was fairly simple to do. I may make the fifth and final world smaller, probably ten rows, with the final boss right at the end.

3. Auto-Levelling
In order to be able to arbitrarily test game difficulty I need to be able to jump to a particular level and have an idea how the player might have ‘levelled up’ at that point. What weapons will they have unlocked and how powerful will they be? What will their base health be? It’s not straightforward to figure this stuff out so I ended up writing an algorithm that takes a destination point within the game pyramid, figures out the location of each treasure chamber before that point, then does a mock play though of the game to unlock each treasure item. On the way I collect the cash that would be awarded for defeating enemies, rescuing babies, and completing ‘sectors’ (rows of the pyramid). Once cash is earned it is spent on the most expensive upgrade available.

It took quite a while to test this code and get it working but it’s going to be invaluable for testing as I can now start the game at any point and have the player ‘levelled up’ as appropriate. Using this code I can also monitor things like how long it will take to level up a particular weapon, how much a player might earn for completing a level at the point weapons are unlocked (hence what a sensible starting price for updates might be) and all sorts of other stuff I haven’t even thought of yet!

4. Basic Gameplay Testing
Using the code above I began going through the game to check and tweak parameter scaling at key points (i.e. the ‘reveal’ level for each weapon and enemy) to make sure things seemed sensibly balanced. Unsurprisingly they were way off to start with but after much fiddling I’ve reached the point where relationships at least seem workable, haven’t done the treasure chamber guardians and bosses at all yet though.

One thing that became apparent was that weapons that are unlocked earlier in the game need to have a greater number of upgrade levels than ones that come later on, otherwise they either ‘max out’ too quickly and end up becoming useless on high level enemies or they cost far to much to upgrade in the early stages. This was particularly apparent with the default weapon (the pistol) so I ended up adding a new default weapon (the .45 magnum, essentially a pistol on steroids) which is unlocked about halfway through the game and has enough grunt to take the player through to the end of the game.

I also ran into a shedload of bugs, it’s been ages since I looked at many of these enemies/weapons so there’s a ton of small issues created by various changes I’ve made. Fixed a bunch of them as I went along (partly why this phase took so long) but I’ve still got a very long TODO list in Trello!

Dev Time: 6 days
Total Dev Time: approx 247.5 days

previous

mockup_3x
Mapping Out The Levels

mockup_3x
Debug Output From The Auto-Levelling Code

mockup_3x
A New Type Of Minion Enemy

Jetboard Joust Devlog #94 – The Armed and the Dangerous

So far this year I’ve been focussing on weapons and the weapon unlock/upgrade mechanic in preparation for doing the wider gameplay and difficulty balancing. I’ve broken this down into three key areas…

1. Ammo Drops
It became clear whilst testing the bosses that the way I was calculating ammo drops was flawed and I needed a better method. The method I eventually came up with is simpler than its predecessor, works far more effectively and should ‘scale’ automatically as weapons are upgraded and the player faces enemies that soak up more ammo. For each weapon I now work out the maximum amount of damage that can be done to any enemy from a clip’s worth of ammo (the amount contained in a single ammo drop). I then scale this amount based on the accuracy of the weapon in question (weapons that have a lower accuracy scale down more as one must assume that not every shot will hit its target). Once the player has dealt out damage to any combination of enemies that exceeds the resultant ammo refresh rate a new ammo drop is awarded. It’s important to record the damage dealt as the amount of damage that would be dealt if the enemy had infinite health, otherwise enemies that are destroyed by the attack score too little and this can really skew the system.

To test this I set up a ‘sponge’ enemy that does nothing but takes loads of damage and tried out all the different weapons on it in turn, tweaking the accuracy scaling and checking the method I was using to calculate the max damage per clip was correct on each one. This was easy for weapons that simply fire bullet-style projectiles but more complex for weapons like the flamethrower. For ‘area of effect’ style weapons like the grenade launcher, RPG and sonic boom I can only really approximate an idea of maximum damage.

Whilst in the process of the above I got pretty distracted re-working the shotgun blast effect as it still didn’t seem to give an accurate indication of the blast’s area of effect. This is the third time I have re-worked this(!)

2. Weapon Switching
To date the player has only been allowed to carry one weapon at a time. If the currently armed weapon runs out of ammo they were automatically switched to the default weapon (pistol) which has infinite ammo. If they wanted to arm a more powerful weapon again (pretty much guaranteed) they would have to pick one up from a weapon crate AND find an ammo drop to recharge it should it have run out.

I decided this mechanic was no fun and therefore, according to the Scott Rogers principle, had to go. Now I am allowing the player to carry two weapons at once – the default weapon with infinite ammo and a (generally) more powerful secondary weapon. If the secondary weapon runs out of ammo the player is switched automatically to the pistol as before but this time all they need to do to recharge it is collect an ammo drop. The new mechanic seems to feel much more natural and fun to me, though I’m a little worried it might give the player the opportunity to over-exploit powerful weapons but we shall see…

As an adjunct to the above I also implemented a key to switch weapons so that the player can switch to the pistol if they want to save ammo on powerful but understocked weapons such as the RPG.

3. Weapon Unlocks
Previously, in order to unlock a weapon, the player had to catch the jetboard of an enemy that was armed with it. This worked OK, but it was a bit easy and I didn’t really think it made a big enough deal of the weapon unlock process.

I’ve decided instead to have weapon unlocks as a type of treasure. Rather than being guarded by a boss, the treasure chambers that contains these weapon unlocks will be guarded by a fleet of enemies armed with the weapon in question. This enables me to make more of the treasure chamber mechanic, adds another layer to the gameplay, and also allows me to use the big ‘weapon upgrade’ icons (which I was rather pleased with) in-game as pickups.

It didn’t take me long to design these ‘guardian’ enemies but I spent a fair bit of time on implementing some special AI for them. Firstly I enabled them to swoop down and steal the player’s health pickups to heal themselves (I may allow other enemies to do this once the reach a certain level), and secondly I implemented a special ‘wrap attack’ whereby if a bunch of them have been chasing the player in the same direction for some time a few will take advantage of the world wrapping by peeling off and heading in the opposite direction to meet the player head on!

The video demonstrates unlocking the shotgun by defeating a small fleet of enemy guardians. They’re pretty tough opponents – as you can see I had to rely pretty heavily on the jetboard attack here and was pretty lucky managing to take out three of them in one go!

Dev Time: 241.5 days
Total Dev Time: approx 4.5 days

previous

mockup_3x
Using a ‘Bullet Sponge’ to Test Ammo Drops

mockup_3x
Enemy AI Now Enables Them To Steal the Player’s health Pickups

Unlocking The Shotgun – Note The Guardian’s ‘Wrap Attack’ Technique

Jetboard Joust Devlog #55 – The Bubblewrap Effect

Everyone loves popping bubblewrap, yet no-one really knows why. For some reason the combination of the sound and tactile response makes it incredibly satisfying despite being utterly pointless. I had a friend who use to refer to this type of action as being ‘urgey’ – once you’ve done it you have the urge to do it again, and again, and again…

I’ve always thought we should aspire to this ‘bubblewrap effect’ when designing games. Most games, even the supposed AAA ones, comprise a fairly limited set of repetitive actions. If you can make those actions an enjoyable experience in and of themselves, regardless of gameplay, then you are onto a winner because no matter how good the player is at playing your game they will be having fun and come back for more.

Recently I’ve seen this loosely referred to as ‘juice’ or ‘game feel’ but these terms are rather vague and are often used to refer to all sorts of things. I’m talking about something pretty specific here – make all your repetitive actions as ‘urgey’ as popping bubblewrap. Usually this is a combination of both visuals and audio.

Now I’d already spent a lot of time on this stuff in Jetboard Joust but, whilst surfing GDC talks on YouTube, I came across this excellent talk by Jan Willem Nijman of Vlambeer on adding these types of elements to your game. I’d already implemented many of the techniques he talks about (camera shake, gun recoil, enemy and player knockback etc) but he made me rethink some aspects and put a bit more effort in to areas that were somewhat lacking.

So – here’s what I’ve been working on as the result of @jwaaaap‘s talk.

1. Bigger Bullets
To be honest a) this would never have occurred to me and b)I never would have thought it would work if it did. I was using little pixel squares for bullets as 1) they seemed appropriate for the size of gun and 2) this worked in Defender so why fix what ain’t broke? But I thought – ‘what the hell?’ and gave it a go. I started increasing the size of the bullets a little and was amazed how much better this felt, so I increased them what I would have thought was a ridiculous amount and it felt even better! It makes no visual sense whatsoever but the pistol (and particularly) the gatling gun are so much more satisfying to shoot now. I haven’t tried playing with the accuracy yet but should really do that too…

2. Camera Knockback
I already have some pretty hefty recoil on weapons but @jwaaaap suggests also recoiling the camera a certain amount when a weapon is fired. This didn’t make a massive difference in Jetboard Joust, probably because the camera is generally moving pretty fast anyway, but it is noticeable under some circumstances so I left it in.

3. Explosion Delay
Adding a very slight delay when an enemy is destroyed adds to the ‘jolt’ effect and makes destroying enemies much more satisfying. It’s subtle but it works. I’m using a delay of 32ms. I had to be careful here to not implement the delay until the next frame (ensuring the first frame of the explosion is drawn before the delay occurs) and also to clamp the delay time so that destroying a bunch of enemies at the same time didn’t result in a massive delay. I also improved the first ‘flash’ frame of the explosion by adding a ‘threshold invert’ to my collision shader and making the circles that briefly appear larger, brighter and less pixelated. Enemies really look like they’re getting nuked now!

4. Permanence
I had been wanting to do something to make battles seem more ‘permanent’ for some time and @jwaaaap‘s talk was the kick up the arse I needed. I talk about adding smoke in my previous post but that’s still not really permanent so I also added bones that fall from enemies when they’re destroyed and collect on the ground as a permanent record of the carnage that’s ocurred there.

Adding the bones was easy, the trouble started when I decided that they were too static and should react if the player hit the ground near them or crashed into a building that they were resting on. I didn’t want to run collision checking on every bone (there can be tons of them by the end of a level) so worked out a system whereby the world is divided into a series of overlapping ‘bone zones’. When a bone is static it is added to a zone and an entire zone can easily be discarded from the collision detection process in one go. I’ve used this approach before and it works well but I got myself into a bit of a flap with it here, plus it took a long time tweaking the various parameters so that the bones seemed to get disturbed by the correct amount. It still looks a little odd sometimes but its much better than having them totally static.

I’d really like to add some permanent damage to the buildings but I haven’t yet figured out a way to do this that would be a) be cpu/memory efficient and b) not involve creating a load more pixel art. I will continue to give this some thought – it could be that I’m underestimating the memory available on modern devices as a spent so long developing for J2ME feature phones!

So I hope that was all worth it and makes my game feel a little more like popping bubblewrap. I’d like to say these were the last gameplay tweaks before I release the alpha but watching my son play it has led me to implement just a couple more things…

Dev Time: 2 days
Total Dev Time: approx 99.5 days

previous|next

mockup_3x
My… What Big Bullets You Have!

mockup_3x
These Bullets Are Ridiculous – But Somehow They Work!

mockup_3x
Explosion Delay (Exaggerated), Smoke, And Improved ‘Flash Frame’

mockup_3x
A Battle Amidst The Bones Of Fallen Enemies

mockup_3x
Bone Bashing – A Stupid Detail That Caused Me Much Grief!