Dev Log: Texturing Levels in Warden

Hey folks,

Lets delve into our method for creating and texturing the levels in our upcoming game, Warden. Every level in Warden is built using Blender 3D. The base mesh of a level is created as a single, contiguous piece, with smaller details added in-engine later.

Textures are created or re-used based upon what and where the level is, and many variations upon them are used for blending purposes. The entire level is then UV mapped inside Blender. We apply textures to the faces of the level and export the mesh as an FBX. Doing so will cause Unity to create materials for each texture and assuming the texture is identically named within the project, this will automatically assign the correct texture to the material.

We do not embed our textures inside the level FBXs that would blow out the size of the project and cause double-ups of the textures in the project, which is just bloat.

When texturing in Blender the process is simple. Textures come in three types. Tiling (On both U and V), tiling only on U (Horizontally) and tiling only on V (Vertically). Often walls of the level will need to tile horizontally while the floors will need to tile in all directions. Only very thin and tall objects will only need to tile vertically.

Generally speaking we will block out segments of our level with UV seams to divide the textures into easily managed islands. After laying down the base textures we can add extra seams to apply transitional textures to blend between texture sets.

Tips and tricks: Texturing levels can be time consuming so ensuring solid mesh flow is key to not getting bogged down unwrapping. To avoid obvious seams you need to be careful about tiling textures and colours. Often large dark vertical shapes in the texture can be used to cover up a seam. making it much less obvious when playing the game.

Strengths of this technique:

• Works well in unison with low-poly level geometry.
• Easy to keep a consistent look and feel to areas.
• Faster than tweaking unique textures for each level geometry.

Weaknesses of the technique:

• Potentially uses a lot of textures (Texture atlases can help mitigate this).
• Loses effectiveness with noisier textures (Takes much longer to make them line up right).

Happy texturing! Tim

Dev Log: Warden's Mecanim animations

Hey folks,Let's look at Unity 3D’s Mecanim animation system and how we have implemented it in Warden for our playable characters.

Every animation the character plays transitions from it’s “Default” state. Default is a series of nested blend trees to blend between Idle, Walk and Run animations, as well as directional movement such as sidestepping and leaning into a turn. Almost all the other states in the tree return to this point with a trigger called “Default”. From the Default state we branch in many directions for all the different kinds of movement and actions the player can do. For example jumping, crawling and attacking.

To keep the state machine tidy we organize states into key categories such as Damage, Jumping, and Default Tasks. The Jump state tree is quite simple. There are three animations, one for take off, one for falling and one for landing. When the player hits the jump key, we hit the “Jump trigger” to push the animation into the beginning of the jump. From here based on the “exit time” of the animation the character naturally reaches the apex of their jump and begins falling, while automatically transitioning into the fall animation. Finally, we detect if the character is grounded again and when they are the “Land trigger” is hit and the animation for landing is played.

We also have a catch all for when the player falls off ledges. This allows the character to transition into the falling state without jumping.

Jumping!

The attack tree was initially quite a simple and tame beast. After many attempts we solved the issue of having multiple weapons and styles with an integer to determine weapon type. Each attack can combo. We count where the player is up to in their attack combo to determine which attack animation should be played. Then we check what their current weapon’s “type” is using the aforementioned integer and play the relevant animation.

For example, the player has a sword. They press attack. From default the animation transitions into attack combo 01 which is a blend tree containing all of our weapon types first combo animation. The sword is weapon type 0 and thus the combo blend tree is set to “0” and that animation is played. If the player was using a spear, their weapon type would be 1 and a different animation would have played.

We found Mecanim's triggers to be a godsend. Initially, when we started work on Warden, triggers didn't exist yet and we had to do a lot of nonsense with bools to get it all to work. All of the Warden playable characters share the same animations, re-targeted across multiple rigs using Mecanim's humanoid re-target system. This is a real time saver as our animator Rob doesn't need to create unique animations for every single Warden we want in the game, allowing more time to be spent making the animations better.

Hope this was insightful, if you have any questions about using Mecanim, comment below or drop us an e-mail! Peace out. Tim

Dev Blog: Fixing Texture Banding in Unity

Hey Devaroos, Just a quick post from me today, but I had fun optimizing our current project, mainly looking at textures and materials. Whilst plumming the depths of the Unity project and fiddling with the import settings of our assets I discovered a surprising and quite frankly beautiful act of compression.

The default "automatic compression" in Unity doesn't like colour gradients. If one was to change the compression from automatic to DXT5 compression we not only have no more banding, which is awesome, but it also halves the size of the texture.

Our programmer loved this, and the game looks better!

So if you are having trouble with a texture banding or having its colours blend horrifically, try changing the Texture Type setting to advanced, and the compression to DXT5. I encourage you to try out all the compression that Unity offers, as some can reduce the size immensely while maintaining or even improving perceived quality.

Thanks for your time, I hope this helps out.

Tim.

Dev Blog: Quick and Easy Particle Sprites

Hey Devs,

Today we're going to have a look at how to create particle sprites. There are a great many different kinds of sprites that one can paint however I’ll go through my basic technique here. It is really fast, and easy to get a lot of variation out of.

Starting in Photoshop, I create a new document of 256x256. This is a huge sprite to make; most sprites we use get down-scaled to 64x64 or maybe 128x128. However I like to author my image large enough for the resolution to never become a problem.

With our blank image I use the gradient tool, set to radial from white to black. With the ruler tool I can start the gradient in the very center and drag it out to just before the edge of the document. This is the base that we paint our sprite details onto.

Smoke:

For a smokey particle, I create a new layer and use the Cloud filter. With adjustments from the levels tool I can blend this cloud layer with multiply to make the white base we started with become much more wispy.

The sprite however is now too circular, this can become quite noticeable in particle systems with few particles, to clean this up we can paint black into the base layer, this has the effect of essentially erasing the particle.

Fire:

For a flame effect, you need to know what kind of fire it will be and how your particle effect is going to function. Often you can get away with the same particles you use for your smoke, but if you you want to go the extra mile utilize an eraser with a hard falloff and use the chalk brush preset to give some harder edges to the flame sprite. For added effect you can apply the liquify tool and drag out tendrils of the flame, however this can quickly become messy and look nothing like fire.

Tips:

• You can generate an alpha by pasting the final image into the alpha channel, or if you are working in Unity, you can simply tag the "import Alpha from greyscale" in the texture import options.
• Setting the opacity of the sprite in Photoshop to 60 - 90% will give it a softer feel.
• You could generate a series of these to create a sprite sheet, tweaking them via the warp tool, liquify, etc.
• Be wary of dragging out long tendrils, these will make the sprite much more recognizable in engine.

Applying these techniques can quickly get you a variety of particles and often sprites designed for one thing work surprisingly well for another. Don’t be afraid to use the same sprite again and again but be careful how big and obvious those sprites become as players will pick up on the repetitive nature of them. Really, they are so easy and quick to make that unless you need to cull textures from the project, you may as well make another.

 Cheers guys, I hope this helps out for some quick particles. Timothy

Dev Blog: Inheritance in Unity 3D

This blog post is technical and aimed at programmers new to Unity.

Howdy readers,

This is going to be an introduction to using the Object Oriented programming principle of Inheritance in our game engine of choice, Unity 3D and programming language of choice, C#. Inheritance is one of the primary OO concepts and deals with creating classes that are parents and children of each other, allowing them to share information and functionality.

I thought for years that classical inheritance was incompatible with Unity’s Component-Object model but not only is this not the case, it’s implementation in Unity is particularly potent, and compatible with Unity’s inspector. The main advantages of using inheritance are twofold; firstly, Subtype Polymorphism allows parents and children to co-exist as members of a single array or list, as well as be substituted for each other as parameters for functions. Secondly, it means you can set up basic behaviour only once for any given object type, such as a bullet or enemy, and have that identical functionality shared across all subtypes.

Hopefully you’re already a little familiar with using inheritance, but how does it work in Unity? Let’s find out!

Step 1 - Creating a base (parent) class to extend from

Creating the parent class is no different to creating any regular script file in Unity, with one exception; if you want a function to be over-ridable (adding extra functionality but also retaining base functionality) in a child class, that function needs to be made “virtual”. The following screenshot shows a virtual start and update function. Virtual functions are, confusingly, still real functions and can be called as such, but now they can also be extended by a child class. (Unity will helpfully let you know these functions also need to be public or protected, a keyword we’ll cover in step 2.)

Step 2 - Creating a child, how to use parent variables, and the keyword protected

You can start creating a child class again like any regular class, but there are a couple of changes to be made. Firstly, that bit of the class header where we say “ClassName : MonoBehaviour”? The class after the colon, in this case MonoBehaviour, is the parent we are inheriting from. We want to change that in our child class to be the new parent. Don’t worry, because the parent still inherits from MonoBehaviour we still have all of MonoBehaviour's functionality! (like transform and gameObject)

Now that we have a parent and child set up, let’s have our child access and use some of the variables it inherits from being a subtype of the parent. You should be familiar with the keywords public and private. Just like public variables are visible to the rest of your program, they’re visible to children too. And just like private variables are hidden to the rest of your program, they’re hidden to children too! It doesn’t take a genius to realize neither of these are ideal. There could easily be variables you want a subtype object to access that aren’t available to your whole program. This is where the keyword protected comes in. Protected variables (and functions) are halfway in between private and public: visible to children but not the rest of your project.

To test this theory you can create one public, one private, and one protected variable in your parent class. Try accessing them from your child. Then see which are available in Unity’s inspector if you attach that script (or perhaps the child script!) to an object.

Step 3 - Parent methods; overriding and newing

The last important, and probably most mind-bending, part of inheritance is how to use parent methods (aka member functions) in a child class. The protected keyword is useful here too but we also need to learn a few extra ones. As we learnt in step 1, parent functions need to be marked as virtual if you want them to be “overridden” by a child, so the first step is to ensure your function is virtual.

Next step is to recreate the same function in the child. Don’t make it virtual this time! You can add some new, subtype-specific functionality to this one. You can access and modify both parent and child variables in here to your heart’s content. Okay. Try running that and see what Unity says. You should get the below warning, because there’s a few more things we have to do. One option is to add the new keyword to our child’s function. But if you do this, it will act like the parent’s function of the same name doesn’t even exist. Not exactly what we want.

So, we also need to specify a new keyword to our child’s function to tell the program “This function is adding extra commands to an existing parent function”. We do this by specifying it as an “override” function. See the below image.

Almost there! Your code should compile without errors now, but your child function will be ignoring the parent’s. The last step is to actually call the parent function from the child one. You do this with the line “base.FunctionName();”. Base is another new keyword, it refers to “my parent class”. If you’ve used Java or ActionScript 3 before you might recognize it as the equivalent of super.

Our class relationship is complete! Any time the child function -  be it Start, Update, or something we define ourselves - is called, it will do it’s subtype-specific functionality alongside that of it’s parent. (Before or after depending on where you placed the base line.)

To use this setup in your games, attach subtype script files to your objects, such as SmallBullet.cs and LargeBullet.cs to their appropriate prefabs, and place all the functionality relevant to all of them in the parent Bullet.cs class. You don’t ever have to actually attach the parent script to anything! It will happily do it’s job from the shadows. With inheritance in place I can now create a list or array of Bullets and add both Small and Large bullets to it without a problem, they will peacefully coexist.

If you would like to download the project I created for this demonstration you can do so here, it’s compatible with Unity 4 and up: InheritanceProject.zip (130 KB)

---

Phew! Well done to anyone still reading, you should now have a greater understanding of how to utilize inheritance inside Unity 3D. Hopefully this aides you in future projects, I rely on it heavily to create expandable, readable code. In my next dev blog we’ll be looking at Unity’s player prefs library, a subject easier on the technical side but still aimed at programmers.

Until then, keep making games! Calum