Tag: developer log

I really pushed myself to the limit for this Ludum Dare 36 game development competition. Looking back at the length of time for all of my live streams on YouTube, I calculated that I spent over 22 hours of development time over two days for this event. Since there is no voting or rankings this time, I felt more encouraged to do my best work since my game will only be played by those who genuinely want to play the game. The game that I developed is called Ancient Adventure, which has gameplay based on classic adventure style games.

The premise of the game basically came from some of my discussion points on the latest episode of the Knoxville Game Design chat podcast, which I am a frequent contributor. On that podcast, we discussed Anodyne, and I made several points about how they got the classic Zelda formula wrong. The following is a list of some of the points that I made, and how I implemented those in Ancient Adventure.

A “compass” upgrade showing where the key items are located.

I created a compass item, which I call the all seeing eye which displays on the mini-map the location (in red) of all of the collectible artifacts. With the mini-map for this game, I tried to make my map more intuitive, by having all of the rooms grayed out initially, and the rooms that the player has visited in a brighter shade of gray. The player’s current position is highlighted in green.

Keeping with the ancient Egyptian theme, I used the classic eye hieroglyphic. It’s the eye that is a popular tatoo and similar to the mage symbol in World of Warcraft. After some research, I found that this is actually called the “Eye of Horus”, which is a symbol of health and protection. The full myth of Horus can be read on Wikipedia. Ironically, the Eye of Providence, which is depicted on top of a pyramid and on the back of the United States dollar bill, has nothing to do with Egyptian mythology. Using this theme made me wonder why there are relatively few games and movies based on Egyptian mythology, since there are numerous games based on Norse and Greek mythology. The only game that comes to mind is Age of Mythology, which is actually based on a collection of ancient religions. There seems to be a lot of untapped material in Egyptian mythology which could be made into games. In the English language, we named our planets based on Roman gods and the days of the weeks based on Norse gods. The best known reference to Egyptian mythology in our culture is probably the Steve Martin King Tut sketch on Saturday Night Live. I wonder if Egyptian mythology has been written out of our history because the culture was overly oppressive, using slavery to build pyramids to their gods, which was at odds with the Abrahamic religions (Judaism, Christianity, and Islam). I will admit that I would have never made a game based on ancient Egyptian culture, if it had not been for the “Ancient Technology” theme for this Ludum Dare.

The point of gathering collectibles.

I used the Egyptian Ankh as the ancient artifact that must be collected to become the supreme being. It is a quick and simple objective and story, but it is at least something that gives a little exposition of the game. I just had in my mind one of the typical gods from Egyptian mythology, who has the head of a dog-like creature and the body of the human. Since there were many Egyptian gods, collecting these artifacts would turn you into the “supreme being”. It’s not a very noble objective, but I think it’s something players can wrap their heads around, since many ancient myths are about gods fighting each other to become the strongest and most powerful. Again, after some research about the ankh, I learned that it is actually a symbol of life. So if I had it to do over again, I probably would have made the ankh the symbol used for the health meter (i.e. “hearts” in Zelda) and made the collectible something different. In this game, the health meter is represented by a bird creature (at least that’s what I intended it to look like) which was used in many Egyptian hieroglyphs. The important thing was to have a symbol which was symmetric, since I intended to split the icon in half to represent a half unit of health remaining. Unfortunately, I didn’t have enough time to represent half health units, so I just doubled the number of full health units. Under the hood, the health value is just an integer anyway.

I tried to ensure that all of the collectibles were spread evenly across all of the rooms. The right side of the map is heavy on ancient artifacts. This is balanced by having the staff upgrade on the left side of the map. It is possible to gather the artifacts without getting the staff upgrade, but having the upgrade makes completing the game much easier. Also the all seeing eye item is in the first room to the left, making it an item that should be picked up early in the game to assist with finding the ancient artifacts, which is the primary goal of the game. I put the all seeing eye in the room to the left, since players of the classic Zelda game are accustomed to starting out left being a dead end.

How to tell if you have collected all of the required items in a level.

Below the health meter, the number of ancient artifacts (ankhs) the player has collected is displayed. The icons are initially displayed as blacked out, which reinforces to the player that there are eight to collect. As the items are collected, they are filled with the purplish color that I decided to use for the ancient artifact item. For the pickup in the game world, I used the same two rotating spotlight effect again, which looked really nice in my game Kitty’s Adventure.

Weapon attack swinging animation.

In classic Zelda fashion, the player starts with no weapon and is defenseless until the player picks up a weapon. In this game, I decided to make the player’s weapon a staff.

In the early stages of development, I spent more time (about 5 hours) than I had desired on getting the staff swinging, collision detection, and animation working. I had to decide if I wanted to animate the staff swing in Blender or the Unity Mecanim interface. I decided to go with Mecanim, and I found that the benefits to be great, because the Mecanim animation also animates the capsule collider and all of the children objects with it. Since I had a light source added to the end of the staff, the light source moves with the swinging staff, which is a really awesome looking effect! The major issue that I had with Mecanim was moving back to the idle state after the swinging animation was completed. I set a Mecanim boolean value to start the swing, but there was no way to set the boolean back to false after the animation completed. The swing animation would just keep repeating. After looking at some Unity Mecanim tips, I learned that a Mecanim trigger (different from a collider trigger) could be used instead of a boolean value, which would set the animation state back to idle after the attack animation completed, which solved my repeating swing animation problem.

By default, the staff had a mass value, which would cause enemies to be deflected when it. It was a nice effect, but it would also push the player back slightly as well. I made a design decision to set the mass of the staff to zero to eliminate the recoil effect on the player. I think it may also reduce the possibility enemies getting pushed into the wall.

In Blender, I modeled a simple staff with a gem on top. There is one upgrade to the staff, which doubles the attack power. The enemies that normally take two hits to defeat are killed with one strike. The upgraded staff is a simple texture swap and a change in the color of the light source. Originally, I had the standard power staff using a green colored gem and green light, and the upgraded staff as a blue gem with blue light. However, I thought the blue color was too close to green and it didn’t show up very well, so I changed the upgraded staff color to red.

One check that I had to add was to not populate an item that has already been picked up in that room. For the ancient artifacts, I had to keep a list of the rooms that had collected artifacts. If the room number is in the collected artifact list, then an artifact item should not be instantiated in that room. The same goes for the staff pickups. If the player’s staff collected boolean is true, then don’t instantiate a staff pickup. If the staff’s power has been upgraded, then don’t instantiate the staff upgrade.

The purpose killing enemies.

In Ancient Adventure, the player is forced to kill all of the enemies in a room to proceed. Once the enemies are defeated, the doors are lowered which allows the player to proceed. The number of enemies remaining is determined by taking the child count of the EnemyGroup Unity GameObject. When the room is setup, enemies are always assigned to the EnemyGroup GameObject by using the SetParent method on the tranform property of the enemy GameObject. If it is equal to zero, then the method which lowers the doors is called. One problem that I came across is that this led to the door lowering method being called on every frame after the enemies are defeated, which caused problems with the door sound effect being started on every frame. To resolve this, I had to create a boolean value which tracked if the door lowering method had been called, so it is only called once when the enemies are defeated. The door sound effect was created by me flipping the pages of a book over my Blue Yeti microphone, and then lowering the pitch in Audacity. I was impressed with how much it sounded like a huge slab of rock being lowered into the ground.

When the door is lowered, the Exit GameObjects are accessible. These are simple GameObjects with cube trigger colliders. When the player triggers it, then all of the child GameObjects under the Room GameObject are destroyed, and then the GameObjects for the next room are instantiated and parented to the Room GameObject. The player is moved to the opposite side of the room, to give the illusion of transferring to the adjacent side of the next room.

One problem the Exits originally presented was that the enemies could go through them (because it is a trigger instead of a standard collider). Since I wanted to keep all of the enemies inside of the room, I added the doors, which had a regular cube collider, which kept the player and enemies inside. I had to move the exits back one unit outside of the room (determined by the exit row or column), because the player could still trigger the exit when touching door.

Another problem with wall colliders in general was enemies getting stuck in walls. In my code, I had it so that when an enemy collided with a wall, it would make a 180 rotation on the Y world axis, and then start moving the other way. However, the enemies were still getting stuck. After some debugging, I realized that after the enemies did the 180 turn, they were still getting a collide event on the next frame. Therefore, I had to wait until the OnCollisionExit event was called for the enemy on the wall, before I would accept another OnCollisionEnter event. There is still a small bug that occurs sometimes, when the enemy collides with a wall, and then immediately turns, placing them in a direction where they can not exit the wall. The enemy turn behavior is defined in a separate Playmaker FSM, which is independent of the wall collision FSM. Usually, after a few turns the enemy will eventually be put in a direction so that it can exit the wall. Again, the chance of an enemy turning during a wall collision at the same time is fairly small to being with, but it is noticeable when it happens, which would lead to players thinking that the game is buggy. Sometimes it seems like gamers put every tiny glitch (whether it be game breaking or not) under a microscope.

Currency system.

While I was developing the game, I had so many ideas that I eventually had to write the down in a text file, and I ranked them from most important to least important. The currency system was a fairly low priority, but I was able to add a gem display value and have the enemies sometimes drop gems when they are defeated. I used the Blender gem generator to make the gem meshes, which turned out a lot better than I expected. The way the light reflects off of the edges looks almost too perfect. There are two classes of gems which add 1 (green) or 5 (blue) gems to your total. The gem dropped is based on the strength of the enemy defeated. Unfortunately, I didn’t have time to implement a shop or things to buy with your gems, so it is purely there as a score value. The gems dropping when an enemy is defeated, along with the health drops, also give more of a purpose to killing the enemies.

What could be made better?

There is really no elegant way to do a 2.5D Zelda clone. If you do a completely overhead view, then you are only seeing the top of the player’s head. If you do an angled view, then there are issues with things popping in that should be outside the player’s field of view. There have been many articles written about this, with one solution being an overhead view with all of the items rotated at an angle. I decided to use the angled camera, and I just use a camera fade when the character enters another room. With the angled camera, to keep the realism, all of the rooms would need to be instantiated in the player’s field of vision, which would not be very efficient. Also, it would be presenting the player with more information than they need to see. The player should only be focused on the current room. One possible solution may be to create very tall walls around the room, to keep them from seeing outside of the current room, which would keep the realism. Another option would be some sort of fog around the room, which would prevent the other rooms from being visible.

I defined all of the rooms in text files, which are assigned as text assets. Walls are 1’s, doors are 2’s, enemies are lowercase letters, and item upgrades are uppercase letters. Using text files makes creating the levels fairly simple in a text editor. The file contents are parsed using the string Split function and the resulting strings are looped over and read as character arrays. Ideally, these text files should use a better format such as XML, but that would increase the size of the level files and the bulkiness of using an XML parser would probably not be beneficial for load times. Alternatively, the level definitions could be turned into bit strings, since each cell can only have a limited number of values. However, it would make it much less manageable, since it wouldn’t be editable in a text editor. The bit string level definition approach would be good for an online version of the game, if the level data was variable and had to be passed between client systems.

Since the character model was created in Blender and the staff swing animation was created with Unity Mecanim, the arm and staff don’t always line up properly during the swing animation. I could try to resolve this by recreating the character model swing in Mecanim as well, so that the arm properly aligns with the staff. Trying to animate removable components on a character model has been a problem that I’ve had that I’ve never found a good way to solve.

There really aren’t any Easter eggs in the game, but there were some subtle influences from other media. On the Game Over screen, the maniacal laugh was a reference to the game over screen from Zelda II: The Adventure of Link. On the game completed screen, I tried doing my best Val Kilmer Iceman impersonation of his Top Gun “You are still dangerous” line. Some of the room layouts were obviously influenced by the dungeon designs in the original Legend of Zelda for the 8-bit Nintendo Entertainment System.

Overall, I was satisfied with the game that I developed for the Ludum Dare 36 competition. Since I have a solid core engine and gameplay, I plan to develop this game further and release the game on various platforms.

Ludum Dare 35 was my tenth time participating in the full Ludum Dare 48 hour game development competition.  The theme this time was “shapeshift”.  I knew fairly early that I wanted to make a game with controllable shape characters.  My original idea was to have three shapes that you could switch between, and each shape would have a unique ability or power.  I had envisioned having three playable shapes, which were the square, triangle, and circle.  The gameplay was going to be similar to Trine, where the player would need to switch between the three to solve puzzles.

I wanted to take a break from Unity and create a game in some other engine this time.  I had heard a lot about Godot, so I planned to use that engine for this competition.  Godot is a free and open source game engine that is very similar to Unity.  Before the weekend, I worked through some Godot tutorials, so I was able to move a simple sprite around on the screen.  In Godot, instead of GameObjects, it has Nodes which have similar properties such as position, rotation, and scale.  Godot can be used to make either 2D or 3D games.  It has a scripting language that is very similar to Python, where tabs and spaces are significant.  Unfortunately, I wasn’t ever able to fully understand the physics system.  In Unity, physics is fairly simple.  You attach a RigidBody component and one (or more) colliders to your GameObject, and then you are good to go.  In Godot, there are three types of physics objects (KinematicBody, RigidBody, and StaticBody).  Your object has to have a CollisionShape or CollisionPolygon as well, and everything has to be parented correctly under your main object.  It was very confusing, especially with the bounding boxes, and it just never did work correctly for me.

After about four hours into development, I knew I wasn’t going to have the time to successfully make a game in Godot.  Therefore, I fell to my backup plan which was to make a GameMaker game.  I had heard a lot of bad things about how GameMaker handles things like collisions.  Although, I have been impressed with some games that I’ve played recently, such as Crashlands, which was created in GameMaker.

Fortunately, most of my initial development was creating the vector graphics for the game characters, so switching to GameMaker didn’t sacrifice too much work.  I had worked through some GameMaker tutorials before, but this was my first time creating a full original GameMaker game.  The GameMaker development environment is very similar to Stencyl, which I used to create Dream World for Ludum Dare 30.

All of the graphics were created with Inkscape.  I made a simple tree, which was composed of the union of three circles for the leaves, and a simple path for the trunk.  I created other obstacles such as rocks (rounded gray rectangles) and water (curved blue bezier lines) I created a red square, green circle, and blue triangle characters, which all had two frames of walk animation.  I had originally envisioned the heroes to use an additive color model, while the enemies used a subtractive color model (magenta, yellow, and cyan).  I abandoned that idea, since a yellow circle looked much better like an emoji.  I created a purplish trapezoid as the first enemy.  I also had the idea of making the heroes parallelograms with sides of equal length, while the enemies would have sides of unequal length.

When I first started designing the movement of the playable character, I used the default GameMaker actions.  The problem with this method, is that the character is stopped whenever one of the movement keys is released.  So the player could be holding two movement keys (such as up and right arrows), but the player would completely stop whenever either of those keys are released.  For instance, the character would expect to keep walking right if the up key was released.  Also, with the default GameMaker movements, the chacter would move when the key was initial pressed, and would pause for a moment before it would continue moving.  This is similar to typing into a text document, when a key is held down.  First it adds the character, then pauses, then repeatedly adds the character multiple times.

To resolve this, I found a tutorial on making a platformer in GameMaker, so I just modified that method to work with a game in overhead view with no jumping.  This method keeps track of a horizontal a vertical movement speed, and the speed values are modified based on which keys are held down.  Now, the movement controls felt much more accurate and responsive.

Now that I had everything necessary to create a level, there needed to be a goal.  I made a simple cyan diamond as a crystal.  To differentiate it from the other world objects, I made it spin and flash between cyan and white.  Adding custom scripts to GameMaker objects is fairly simple, and the GameMaker API is well documented.

The first enemy that I created was the purple trapezoid.  It simply just moves vertically on the screen until it collides with an object, then it reverses direction.  When the player collides with an enemy, the player is returned to the starting position.  If I had more time, I could have added a health meter so the player wouldn’t have to restart after being hit just once.  I gave the enemies angry faces with two frames of animation, which reminded me of the boxer profiles in Mike Tyson’s Punch-Out!!.

The second enemy that I created was the yellow circle, which was modified from my previous green circle hero.  I just had to remove the arms and legs.  The yellow circle just moves in a random direction for two seconds and then changes to a random direction and moves for another two seconds.  This behavior is very easy to create in GameMaker, and I had remembered implementing the behavior in the “Catch the Clown” tutorial that comes bundled with GameMaker.

The third character I created was the green triangle.  I wanted this to be a water dwelling enemy that shoots projectiles at the player.  This behavior is similar to the Zora in the original Legend of Zelda.  If the player stands in one position for too long, then they will be shot, so the player has to keep moving to stay out of the line of fire.  The third level becomes more difficult with four green triangles shooting at the player at once.

The final enemy that I created was the orange pentagon.  Originally, I had made the pentagon cyan, but I thought that would be too confusing with the cyan crystal goal, so I changed the color.  The orange pentagon shoots in four directions at once and randomly changes location after every shot.  These characters are similar to the Wizzrobes in the original Legend of Zelda, except they shoot in four directions.  Their projectiles can be easily avoided, as long as the player doesn’t stand directly horizontal or vertical from one of the orange pentagons.

For the music, I used GarageBand again on my MacBook Pro.  A few weeks ago, I had read an article on Gamasutra about music theory, and to watch the number of steps and skips, so hopefully my music sounds a little better for this game.  I usually try different instruments until I find something that sounds good.  I use the command + left mouse button to add new notes to the piano roll.  I’ve found that it is much easier to modify the lengths and positions of the notes in Piano Roll editor than the Score editor.  I used the C major key, so I just made sure I just stuck with all of the “white keys”.  Overall, I think it took about an hour to create the music.  Like I’ve done with my previous games, the title screen music is just the game music slowed down with some of the instruments modified or dropped.

I used BFXR for creating the sound effects when the player is hit or when the crystal is collected.  I used Audacity for touchup work.  I also recorded my voice again for the title screen and game complete screen.  I used Audacity to add a bit of an echo sound.  Unfortunately, my voice sounded a little faint against the background music, and I never found a way in GameMaker to make it louder.

There are many ways the game can be improved.  First, I would like to give the player a method for attacking the enemies.  I would also like to have a way to move from room to room, instead of collecting crystals.  I would also like to have various items, money for purchasing new equipment, and many more enemies.

Finally, I played the game MANY times to ensure that all five levels were completable and also not too easy to finish.  I made both Windows desktop (EXE) and HTML5 (web) builds, and uploaded to Itch.io and GameJolt.  I think the Windows build is the better experience, but I know some people only prefer playing the web version.  It just seems like there is more lag and dropped frames in the web version.

Overall, I was happy with the simple game that I was able to make in GameMaker.  I really appreciate everyone who has played it already.  It already has over 60 votes, which is by far the most votes of any of my Ludum Dare games have received.

 

I began development by making a simple cat model in Blender.  Next, I textured mapped my model by UV unwrapping the model, by mostly using the “Project from View” operation while selecting specific faces of my model.  I exported the UV layout, and then I created the texture map in Gimp.  I used the dropper tool get the fur color from a real photo of Kitty.  I ended up adding a little more saturation to the colors, as the texture looked a little faded when mapped onto my model.  I used the reflect modifier, so that I only had to model and texture map the left side of the body.

I created an armature for my model, with bones for the body, arms, legs, head, and tail.  Then I created three animations for my character, which are standing, walking, and jumping.

Next, I created a simple maze in blender by scaling a plane object, and then subdividing it by 20.  Then I selected all of the faces that would be walls, and then extruded those upwards by about 1.5 units.  I downloaded a flower texture from CGTextures and mapped it as the wall texture.

In my Unity project, I imported my cat model and maze model.  I made a capsule game object for the player, and parented the cat model to the Player game object.  I used Playmaker for controlling the movement of Kitty with the Get Axis Vector and Translate actions, using “XZ” as the Map to Plane option.  It is important to add a mesh collider to the maze model, otherwise Kitty will walk through the walls.

I created four scenes for this game.  The “memory” splash screen shows a brief photo of Kitty that fades in and out.  The Camera Fade In and Camera Fade Out actions accomplished that effect.  However, I had to disable all of the GUIText and GUITexture objects after the fade out, otherwise you would see a flicker for a frame before the next scene is loaded.

I added the title screen last.  It just displays the game’s title and waits for the user to press a mouse button to start the game.  I added my Kitty model with the walk animation and I applied a constant rotation action to make Kitty spin.  It doesn’t look very natural, so I may go back and modify it later.

The game scene is where the bulk of the gameplay occurs.  I added entrance and exit objects, which are just cubes with triggers.  The entrance really isn’t used now, but walking into the exit cube with transition the game to the final screen.  It took me a little while to get the perfect height for the maze walls.  If the walls are too tall, then you can’t see Kitty at times.  If the walls are too low, then it makes the maze too easy to solve and looks unnatural (why wouldn’t you just step over the walls?).

I also wanted to use controls similar to a Mario 64 style game.  Playmaker provides a Smooth Camera Follow action, but it makes the camera always stay behind the character, and only works well when using a control scheme where you press up to go forward, down to go backwards, and left and right to rotate.  The control scheme I wanted to use is pressing up to go into the screen, down to go to out of the screen, and left and right to go to the sides of the screen.  The camera would need to be set to the position of Kitty with an added offset and never rotate.  I couldn’t parent the camera to the player game object, because it would have the camera turn wildly whenever the player’s rotation changed.  When the walls were too tall, I would have problems of being able to see through some walls, since the camera was inside of the walls.  Ensuring that the camera was always positioned above the top of the walls resolved this problem.

After Kitty reaches the exit, then it transitions to the final game over screen.  I have a simple congratulations message with Kitty jumping up and down.  At first, the player could still control Kitty on this screen, so I disabled the movement FSMs when the scene is loaded.

At this point, I had a working game, but it was not very complex and it used the same maze every time.  I did some research, and I found that Prim’s algorithm can be used to generate mazes.  I created a new test scene, just for programming the maze layout.  I created a new object and attached a new PrimsMaze script to it.  The script takes a game object (maze wall prefab) and the number of rows and columns to use in the maze.  I found a good video explaining how to make a maze using Prim’s algorithm, so I wrote a C# script using the method described in the video.

The maze generation works pretty well.  I set the exit to the last open space in the last row, although that really isn’t the most elegant solution and could result in problems if there are no open spots in the last row (my Unity editor crashed more than once due to not being able to find an open spot, resulting in losing all work since the last save).  One nice thing about Playmaker is that it will detect and break out of infinite loops, however infinite loops in a script will result in the entire Unity editor hanging.  There used to be a trick where you would attach the MonoDevelop editor to the Unity process and then change the variable causing the loop, but I couldn’t get that to work with the Visual Studio editor.

Now every time someone plays the game, they get a unique maze.  The maze complexity also increases as the player completes the levels.  I created a Playmaker global variable to hold the level number, which is incremented at the end of every stage.  I recently discovered that Playmaker global variables are accessible from any scene, which is really helpful to keep data (like level number) persistent across scenes.  I created a new function in my maze generation script, which has a switch statement that determines how many rows and columns to place in the maze based on the level number.  After level six, it uses the default row and column values.  I need to do some tests to see how many wall objects I can have before the game experiences slowdown.

There are a few more features I want to add to the Kitty’s Adventure game.  Obviously, I need to add music and sound effects.  I’ve also had the idea of adding paw prints on the ground as Kitty walks around the maze, which would be like breadcrumbs showing the areas where you have already been.  There are plenty of obstacles that could be added to the maze.  It would also be nice to have a timer showing how long it has taken to complete each maze.

Developer Commentary