~
In September I finished implementation of all levels necessary for the demo's release. Before August the only complete stage was the pub used in the game's introduction (and that level was completed over two years ago, christ). While that level alone is responsible for around half of the demo's playthrough, the other half is comprised of a wide array of outdoor stages, none of which were as-of-yet implemented. I had actually attempted to jump-start this process last Autumn by sketching out some designs for these levels. However my motivation for this facet of development evaporated shortly after I completed their sketches. The drawings were tossed on a shelf, untouched and unused for the better part of a year.
Now twelves months later, I thought it was about bloody time I cracked on and finished them.
The demo takes place within a quiet district inside the 'old town' of Valentine. As such, I wanted to make use of traditional building styles, namely half-timbered construction:
I think it's fair to say that this style of structure has become a cliché within the fantasy game genre, particularly J-RPGs. That said, they are rarely found in games set in modern eras despite the fact that they are still prevalent to this day in large regions of England, France, and Germany.
If we look at games in particular, implementations of this architecture can be flat and disappointing; oftentimes they are reduced to simple pale boxes painted with a wooden outline. While straightforward designs such as these can give buildings a somewhat serviceable amount of detail without a vast amount of effort, I would argue that they are doing a great disservice to the craft. These veneers often fail to capture some of the most unique and unusual aspects of the design such as jettying:
 |
| not the most obvious example of jettying but it's the only photo I have |
The inclusion of these overhangs in your work can surely serve a greater purpose than academic masturbation. First of all, they create distinct and eye-catching outlines for the homes, halls and hovels that feature them, enhancing their visual appeal and atmosphere: narrow alleys become suffocating when these jetties lurch over the path, robbing it of daylight and dwarfing those caught under their shadow. They can also perform a functional role: on days with heavy rain, pedestrians could take shelter from the weather by seeking refuge underneath some of the more pronounced overhangs.
All of this functionality could be derived from the inclusion of this single element of design.
It should go without saying that buildings don't need to be 100% accurate, far from it-- I'm certainly not pushing for this. That said, I think the value of investigating and researching some of the finer points of architectures is criminally underrated.
It should go without saying that buildings don't need to be 100% accurate, far from it-- I'm certainly not pushing for this. That said, I think the value of investigating and researching some of the finer points of architectures is criminally underrated.
I also think we need to reevaluate how we look at levels in a broader capacity. One of my pet peeves with level design is that we often see player convenience and accommodation of gameplay mechanics take precedence over the creation of a believable and rational world: why is the enemy standing right beside some TNT? Why aren't all NPC villages in the part of the world with no monsters? Why is there a convenient exit at the heart of this dark dungeon?
Yes, I know some of these may sound ridiculous. However I find it worrying that we simply gloss over these issues; as an audience we've grown so accustomed to them that we no longer question their inclusion. Can we not have convenient design and convincing worlds?
This problem can be seen at both micro- and macro- scales. To give a glaring example, take a look at Pokemon. Despite being 'open world', the game (largely) follows a single route that you cannot diverge from in any meaningful way. Despite being one contiguous world, the Pokemon you encounter, the strength of the trainers you meet, the items sold in shops are all conveniently tailored to the given route set out by the designer. Does this make sense within the world? No! No explanation is given in-game, either. It is simply a world made convenient for the player and no one else.
I absolutely detest this feeling of being lured down a single route, lead by the designer rather than by my own agency.
◇◆◇
So how does that rant relate to The Valentine Watch? Well ultimately I aim to minimise as much of these faults as I can with my level design. While there is only so much I can do within a short 15-minute demo there are several key points I have tried to tackle.
When it comes to creating a realistic and believable world, I have tried to design this district of Valentine as a web of streets rather than a series of levels or stages. The urban environment behaves and presents itself much like a normal, real-life neighbourhood: the streets branch and wind, weaving through other streets and alleys. The path you are expected to travel is not glowing or coloured gold-- it is a town of inhabitants, not your playground.
This is not to say that it is entirely directionless. There are places that you are 'meant' to go, but it is ultimately your choice whether you follow these signals or not. You are 'meant' to walk home with your mentor after a drink at the pub but you could just as easily stay and have a few more drinks. But if that's the case, good luck finding your own way home.
From a technical perspective, the body of each building is derived from a tileset composed of 24x24 tiles. Each 'texture' of half-timber (clay, cobble, etc.) has around 90 unique tiles. By making the set highly flexible I was able to make every building within the demo unique, reflecting the one-of-a-kind nature of their real-world counterparts.
The construction of these tilesets took a surprising amount of time, the bulk of which came from testing each tile's ability to align and tessellate with other tiles. This process was constant and gradual-- small, incremental tweaks were made as new inconsistencies were discovered. The sets weren't fully complete until halfway through level implementation.
While tiles were used to build the walls of each structure, additional elements were required for doors, windows, and other miscellaneous objects. About 200 of these assets were required in total for the demo, the majority of which were made from scratch.
A tileset was also created for terrain. It was absolutely imperative that this set was flexible, even more so than half-timber as it has to blend into several ground patterns: cobble, dirt, grass and more. It was important that frequently-used tiles such as cobble had plenty of variation (e.g. different stone orientations, missing stones,...) so that their repeated use is easy on the eyes. Overall, I'm pleased with the result and I think this set provides more variety than your standard 2D-RPG fare:
Plenty of engine adjustments were necessary to speed up the development pipeline and accommodate these new levels. Some of these levels were magnitudes larger than the pub so the engine had to be optimised to accommodate them (up until this point I didn't realise how much I had to care about GC in Java, what an amateur).
You can see the influence these engine changes and asset generation had on the development time for each individual level:
- First Level: 1 Month
- Second Level: 10 Days
- Third Level: 1 Week
- Every Other Level: 1-3 Days
Once the ball started rolling, it rolled.
All in all, it took just under two months to develop the stages. The last few weeks of the process were absolute hell (80 hour weeks are no fun at all) but I'm glad they are done.
With all levels completed, the only significant portion left to complete is a very nasty bit of AI. I'll tell you all about it very soon.
** For those who are curious, I took the photos in Dinan, a delightful town in Northern France.
