Brightwall - Architecture reference

There was one issue with the concept I have chosen. The village buildings that can be seen on it don't really hold enough details to make them look interesting enough. They are very simple houses with just a few intriguing details, like the curved and exaggerated roofs.

Because of this, I went ahead and looked up what fantasy/medieval type of building concepts I could find that match the style I'm aiming for.
Interestingly enough, what I was looking for did not come from CG concept artists, rather real life model makers.

Here are the references:

3D blockout:

The final picture shows all of the modular elements that can be re-used to create different, unique houses.
I have also added the necessary details to achieve a more "Warcraft look". The aim was to remove straight lines from wherever it was relevant. Silhouettes need to be exaggerated a bit to make the building look different, interesting.

The next step involves a lot of technical elements, like unwrapping each of the modular pieces for texturing and importing them into the UDK engine.

Brightwall - Importing the first assets into UDK

I've set up a basic lighting environment inside UDK, so I can have a good representation on how my models and terrain feel like.

There were some issues with the Height Map, but eventually things fell into place and this is what I've ended up with:

In order for the play not to fall through the terrain model, the model has to have collision. If I would have imported the land from 3DSMax, I would have had to set up a collision mesh as well, which is extra work. The "Landscape" tool creates amazing collision when you generate a terrain through it via a HeightMap. As seen in the video, the player doesn't bump into invisible walls nor does he/she fall through the model.
The only collision issues are with the placeholder houses, but that's all right for now.

Everything seems to be in order for me to move on with the modeling stage and start looking at some architectural reference.

Brightwall - Blocking in the terrain

I've spent several days examining all the details of the terrain withing the walls of the village. As a first step, I began drawing over it, looking at how the curves behave and what direction they are going in. I've did this a couple of times, redrawing and getting familiar until I could imagine it in my head. It was important because I only had this concept image, which is quite hard to read. This is because the sizes and the ratios change as things are further from the viewer. Distortion had to be taken into consideration.

After a couple of failed attempts, I've began producing terrains that resembled the one in the concept.

Here is the final blockout, along with placeholders for the buildings:

The terrain was taken into ZBrush to go over it by hand and adjust the details according to what felt right. After it was polished up, I had to go into a very technical phase of the work, which involved creating a HeightMap. Why is this needed and what is in the first place?
The answer lies in how the UDK game engine handles terrains: UDK has an in-built "Landscape" tool, which generates landscapes. These landscapes are extremely well optimized and have a very important characteristic. As the player moves further from the terrain, the distant parts begin to use less polygons, less resource. The program takes away detail from the distant areas, but since the player is so far away, it will be completely unnoticeable. This saves a large amount of resource.

The "Landscape" tool also has sculpting features, similar to that of ZBrush, so if I place an asset on the terrain, but the ground is not flat enough to sustain it, then I just go in and sculpt away. This, again saves a huge amount of time, so I don't have to jump back to ZBrush all the time.

That said, I had to create a Height Map which is a 2D image using only grayscale values. These values contain height information. Black pixels are the lowest points of a terrain, whereas the white ones are the highest. The inbetween values generate the desired landscape.

The next step is to take all this into UDK to see whether it's going in the direction I want it to.

Final SS2 project - Referencing

I've come to the point where I feel that the gathered knowledge from the past few months allows me to embark on a bigger, more complex piece of work.

The aim with my final project is to learn as much as possible about what it takes to create a working environmental level within a game engine. This piece of work will serve as the leading piece of my portfolio.

As a first step, I've looked at hundreds of reference images to get a slight idea about what kind of level I want to create. I need to be very ambitious about this project as it will stretch out until the middle of May, meaning I will have to concentrate entirely on it for two and a half months. It would be a very dangerous situation if I should end up disliking or unsure about the concept I've chosen, so this is probably the most important phase: what am I making?

Here are some of the concepts that made it into the final few I need to choose from:

What I like about most of these is the unique feel of the environments they portray. I look at them and instantly wish I could walk around that place for a bit, feel its atmosphere first hand.. or at least in a game.
Also, technically they are acceptable to create, even though some of them are bigger than the others.

After days of staring at all of them and examining what each would require artistically and technically as well, I've decided to go ahead and attempt a concept which was originally created for the game Fable 2.

What is especially good about this one is the fact that it definitely needs to be created in a modular way. This method is crucial to be mastered today. The level and its assets (especially the architecture) have to be created in a "re-usable" way. For instance: I need to examine all the buildings and decide what fundamental elements, what pieces they are all composed of. After this is done, I need to create a library of walls, support wood, roofs, stairs etc. and then re-use them all over the place. I don't have to create each building separately and uniquely, I need to create the lego pieces that put all of them together.
To eliminate the obvious repetition, I have to apply different textures and the job is done. This is going to be a modular level.

Tileable texture

Another very important matter I need to address is creating tileable textures. They are absolutely basic when creating game art... or any 3D art for that matter. 

When put next to each other horizontally or vertically, these textures provide the illusion of one continuous pattern. Very good tiling textures are those that do not appear to be multiplied over a surface. One texture by itself provides so much randomness that even when tiled multiple times, it still looks as if it one, big, unique texture pattern.

I will be attempting to learn how to create these properly and make them look very appealing at the same time.

I've chosen 3DMotive's Tileable Textures with ZBrush tutorial package to help me understand the very basics. The good thing about these tutorials is that they not only touch the 3DSMax side of the work, but the very artistic, ZBrush/scultping phase as well.

My goal was to create a tileable brick pattern. This is the blockout I have created + showing it tiled:

Some people would say that this is far too much attention being invested into a task that can be achieved much much faster through Photoshop and some various other tools... and they would be right. Creating tiling textures can be done from pure real life photos in a matter of minutes, but I have a different goal.

What I have noticed is that there is only a handful of artists whose work really stand out. There is an "elite" group of people who make their art look a bit more than what is expected. They are regarded as very talented people and this can be true to a certain degree, but I believe their is more to it. I see these artists invest crazy amount of time in tasks that other resolve through "shortcuts". 

This extra attention and detail you add to the very building blocks of your scenes in the end makes your work stand out from the rest, plus it might even make you finish faster, because you won't have to be doing a lot of polishing and clean up work, because you've built everything perfectly from the get go.

This is the mentality I've chosen to follow and believe in, hence I do not regret spending 1-2 days on a tiling texture as in the end, I might be saving up headaches and time.. plus.. its fun!

When the bricks were set up in a random manner and also to tile, I've imported them into ZBrush for sculpting + basic paint.

The goal here was not to spend days on perfecting colors and other significant details to texturing, but to understand the workflow and the important aspects of tileable texture creation.

All this detail eventually ends up as a 2D texture, which will be applied to a simple plane. Once the color and the normal map is applied as well, the plane will look like a bunch of bricks that actually take up 3D space.

Here is the texture tiled, the normal map:

Color Map

Normal Map

The texture applied to a plane in Marmoset game engine: