Been working a couple of evenings on this little Super Mario Run-clone for my kids to play with. It’s obviously quite early in development, but they love it anyway!
Godot has a pretty decent physics system, but as usual I prefer to do the physics myself. Starting to feel pretty good.
Been a while since I worked on this one. But I have some very intriguing ideas, so I wanted to pick this up again. New features since last time:
More to come…
Been talking with my kids for quite some time that it would be fun to make a game together, Now Edda has started actively talking about it, so during the holidays I went ahead and made one.
Funny thing is that Sigrid is the one who made the graphics so far. It’s placeholder though, as she wants her graphics for another game (not supposed to be a jumping crab!!!)
Starting to refactor systems. Split out the enemy wave manager into a ship builder class (build and configure enemies based on type), spawn position solver, and a waveManager (that decides WHEN to spawn WHAT ship). Turns out I need to put more effort in the last part, while the two former worked pretty well.
Did a new enemy type as well. A re-make of the shotgun turret from pewpew5 (that never made it into the game).
It is supposed to be a bullet hell after all =D
Finally got some time for this old prototype again!
Added one more enemy (you can glimpse him in the gif), strolling around and planting mines to hinder your movement!
Turns out that having enemy bullets (and mines) be solid color really messed up your sense of depth. And that is super important in this game! So I changed it to simple lit. Sense of depth is improved, but it looks really ugly now! I need an artist to fix this… =(
There is a known problem that occurs when you try to have a large number of objects moving towards the same exact point… they will meld together and in the end seem like a single object.
In Ripple I avoided this problem by making sure we did not have enough flying/following enemies on screen at the same time. However, I have a case now where I want a ton of theese… a swarm of enemies.
Funny thing is that the math for moving an object towards a point is dead simple. Does some kind of scatter function have to be incredible complex? Or can we do it in an almost as simple way? I had a theory I wanted to try, here’s the psuedo code.
for each enemy in enemies
if enemy == me: continue
if distance D is less than minAllowedRadius R
move me away from enemy in opposite direction by distance D
Result 1
Looks glitchy. Figured out pretty quick that we push by distance D, which means that if minAllowedRadius is 25 and you touch the outer rim of it, distance will be 24, and you will be pushed 24 units instead of 1 unit. And the opposite, if your are 1 unit away you will only be pushed one unit. We obviously need to push it 1 unit in the first example and 24 units in the second.
Result 2
Whoa! That looks amazing! I did not expect it to be that fluid!!!
Here is the final psuedo code
foreach enemy in enemies
velocity = (playerPos - enemy.pos).normalize * speed
lookAt = velocity.angle() // update lookAngle before calculating scatter
pushback = new Vector2(0,0)
var newPos = enemy.pos + velocity
foreach checkScatter in enemies
if (checkScatter = enemy) continue
otherTowardsMe = newPos - checkScatter.pos
if (otherTowardsMe.squaredMagnitude < minAllowedRadiusSquared)
// Too close to someone!
otherTowardsMeMagnitude = otherTowardsMe.magnitude
// need to adjust amountToPushBack, or we get gif number 2
amountToPushBack = minAllowedRadius - otherTowardsMeMagnitude
pushback += otherTowardsMe / otherTowardsMeMagnitue * amountToPushBack
enemy.pos = newPos + pushback
As you might (or might not) notice is that I use the squaredDistance and squaredMagnitude in the first check, where we see if we are in contact. This is to avoid using squared root in the inner loop (that would be a lot of square root calculations).
However, if we do have a hit we calculate the actual magnitude, so we can adjust the amount of pushback. I store it in a variable as I need it twice.
There might be even more optimized ways to do this, but doing a sqrt for N^(N-1) seemed a bit too expensive.
Oh… and this was also my first attempt at using Godot engine. Is was actually really pleasant! I might actually start using it instead of Unity for experiments and personal prototypes… at least for a while, as a test!
Took a first step towards implementing shadowcasting, the ability to show a tile as lit/unlit/hidden.
Very first implementation is just distance based. Next step: recursive shadowcasting! =D
To be totally honest, this setup of the generated dungeon doesn’t feel very interesting to explore. But that’s not my main concern right now. I’m just trying to get the very basic systems to work before moving on with the gameplay parts.
Started on a strategy for spawning enemies today. I don’t want you to die becaues they appear from behind our out of your vision, so focus is ahead! Trying to find a good min/max distance forward and curving outward. Wanted to keep it wider to the sides and narrow it down a bit in height to allow for an easier feel when looking around you.
First try:
Second try is to distribute the enemies more evenly. Try to find large patches of empty spots and avoid where it is already cramped.
This turned out quite well. I can tell you that it feels a bit intimidating watching all those Trump heads stare at you in VR X-D
I also made a new enemy type:
And here is a very short video of todays progress. Let me tell you that without too much time put into this one, it is already FUN AS HECK!
All right, in order to start doing something cool I need a to navigate in. RougeBasin to the rescue! They have a really cool tutorial on a BSP dungeon (Binary Space Partitioning, simply put: a node tree where every node has two child nodes). Link here
First step: Create a child node of the entire area. Then randomly split it vertically or horisonatlly (offset the split by a random amount) and add the two inner segments as childnodes. Then repeat for each child node.
Second step: Take all the leaf nodes, shrink them a bit and move them around by some random amount.
Third step: Draw a corridor between every sibling node. Then move up to the parent node and draw a corridor between those siblings. This picture only has corridors on it’s outmost leafs.
Fourth step (cheat): Turns out that connecting the parent nodes are a real hassle! In order to keep any kind of speed I did a cheat and drew corridors through the center of each sibling node. As long as I don’t offset any room beyond the center coordinate the corridor will pass through a room.
This was fun, but getting step 4 up and running took way too much time. You don’t want to lose too much momentum when dealing with hobby projects like these, and this is good enough for me to continue with the next exciting step!
Got a scrolling camera to work in tonights short session, showing a part of a bigger underlying map. Next up:
Really quick dev today. Added a fourth shiptype that bursts out a ton of bullets. I have a good initial setup now! Time to start on spawning them in sequence.
Today, by chance, I stumbled upon the free tileset section on Itch.io (https://itch.io/game-assets/free/tag-pixel-art), and I instantly fell in love with so many of them! I’m usually not a big fan of using random peoples stuff, but I think this made me change my mind!
I downloaded a few and started playing around with one of them, making ireggular semi-topdown floor tiles. I could easily integrate them since this is not a console at all, but a regular window I draw bitmap fonts into. Easy peasy!
But the very reason I started with this in ascii was to NOT having to bother with graphics… so even though it looks realy nice, it will have to wait for another project.
Some more progress, this little VR prototype is going in the right direction indeed!
I needed a way to distinguish the different enemy types, so I just pasted different Trump faces on the cube. Consider it placeholder graphics (even if it is kind of funny) =D
The controls did not “feel” right. Has been working on improving them for a couple of evenings now. Great improvement even if there are still some work to be done.
Bought myself a new hobby gamedev laptop. It’s really small, and really neat. And kind of weak… So I needed a project that it could handle.
After trying haxe it was clear that compiling haxeflixel was painfully slow. If I did it directly in c++ instead it worked like a charm.
So c++ project it is! Using SDL and the base from my Ludum Dare code base. Got a dude moving and placing blocks. Good first progress =)