Hi,
I’m Marcelo Vianna a Level Designer at Hermit Crab Studio and an indie developer recently relocated to Vancouver!
I’ve been making small games and building levels for almost a decade now… from Warcraft custom maps and RPG Maker 2000 to indie games and level editors such as Trenchbroom and Hammer.
Eventually, I stumbled upon Unreal and fell in love.
I quickly migrated to UE4 than 5, started making levels in Fortnite, and even became a true evangelist for Hermit Crab Studio to enter the Fortnite landscape.
But a s a Level Designer, I felt very limited by templates. I could make levels and tweak templates here and there, but not as much as I wanted to actually make a full game. I’m pretty confident in my skills to create levels in Unreal but I have almost no idea how to actually make a game in the Engine.
That’s why I applied to the Fellowship!
About Me and my Goals
The Fellowship is a huge opportunity to learn everything about Unreal and develop a really cool game that will be played by a bunch of people on an actual arcade machine.
So my approach was to make a tiny list of what I wanted to learn in these 4 weeks.
My main goals are:
- Learn as much about AI and Behaviour system as I can;
- Make an arcade game meant to be played on an arcade machine (a once-in-a-lifetime opportunity!);
- Make a scalable game that I could keep developing after the Fellowship;
With these goals in mind, the idea of making a Pac-Man style game with Metal Gear guards came to life! Somewhere along the way, I realized that the character wouldn’t have any gear, and “Naked Sneak” felt like a pretty funny pun.
Week 1 | Pre-Production and Tests
My first step was actually in Week 0. I saw the week-by-week roadmap and used it to create a roadmap of my own. Then, I iterated on that roadmap on Day 1 with Kaitlin Perry.
With a rough roadmap in hand, I now have deadlines and goals for each week.
Game Design Document
Right after setting the roadmap, my first task was to make a GDD, so I downloaded the template from the Fellowship and at first, I just threw in some images and references.
Then I added bullet points for each topic, elaborated on each one, and BAM! GDD complete just like that!
I have put into paper what was in my head, and by having this idea in a material state I could iterate on it.
Making it more concise
This step is one of the most important… cutting unnecessary elements, removing redundancies, and identifying what could make the game better..
One thing I always try to do is get different resources and combine them into one, for example in Naked Sneak in my first GDD iteratation there was, time and smokebombs that would work as life.
But by revisiting the GDD I came into the conclusion that the game sounded cooler if instead of having a health, to use the timer as the health bar, so the only game over condition is the time out.
By combining health and timer, i made Timer a much more interesting mechanic and the game seem much more fun this way, and it came with the bonus of reducing my scope in mechanics and made my Hud clearer!
Skateboard First and Moscow
My approach to development was a Level Design technique from Steve Lee called “Skateboard First”. The idea is that I would make the more barebones version of my game first, make it fun then build on top of it.
For it to work I went to my GDD and did a “Mo.S.Co.W”, basically labelled everything into:
Must Have, Should Have, Could Have or Won’t Have.
My plan was to face my Must Haves and make them first, I plugged into every lab I could and section with the things that I needed in mind, so everything I was learned I already had a BP making the version that I would use in my project.
This way I was much more intentional with my learning process, and was very hands-on. Every exercise and talk I was already exercising how that new piece of knowledge would fit my design.
Gym Level
As a Level Designer, once I had my template a bunch of questions about how this game played came into my mind. So to start I made a new Level, a gym level to test the parameters and mechanics.
First Mechanics
Almost everything that I made in this first week was either learned from one of the talks, a freebie that I’ve been hoarding from Fab for a while or something I tweaked in the Template (in my case the TopDownTemplate).
Considerations
The week started very frantic and I was kind of scared, but by developing a roadmap and with the support of Kaitlin, I scoped and managed my projects and not only it improved Naked Sneak and helped me absorb much more from each talk and apply what I was learning into a real context.
Week 2 – Keeping Up the Momentum
After a strong start last week, I didn’t want to lose any steam.
I kicked things off by listing everything I wanted to implement in my game and scoring each item based on importance vs. time investment. That’s when I hit a little roadblock.
Priorities & Roadmap
Scoring my ideas helped me realize a tough truth: the essentials take time, but not always the kind of time I want to invest my precious fellowship time on.
Some tasks, like sound, widgets, and sequencer, weren’t areas I was eager to explore but were quick to implement. Meanwhile, modeling and texturing were massive undertakings that I wasn’t willing to divert my Unreal learning time into.
So, I found a simple solution: ask for help.
Assets & References
Before reaching out to anyone, I made a detailed asset list, marking what could be done with basic blockout modeling and what needed real 3D work.
To streamline the process, I organized everything in PureRef, a moodboard with references for every asset and texture.
With everything mapped out, I reached out to my talented friend Fernanda Corrêa, and she came in to the rescue!
With my 3D needs covered, it was time to tackle my main goal for the week…
AI aka Process of Learning in the Fellowship
As I mentioned in Week 1, Behavior Trees were one of the areas I was most excited to learn about in the Fellowship.
Taking advantage that one of this week’s courses focused on AI and the Navigation system, I dove headfirst into it.
Learning AI
The class, mentored by Ed Bennett, was the perfect starting point for someone like me, coming in with zero AI experience. I learned about AI Controllers, Trees, and Blueprint Tasks, and during the Q&A, I even got extra insights into decorators based on my example.
Live Labs – AI Bootcamp
My AI learning didn’t stop with the class. During the Live Labs, I camped out in the same room for the entire session, bombarding Trent Cornwell with every AI-related question I had. He guided me through setting up AI Perception, working with stimuli, and integrating them into my Behavior Tree.
The Rabbit Hole
When the Fellowship sessions wrapped up for the day, I went deeper. I found Ryan Laley’s AI Playlist, which had everything I needed and then and after that I even started his older and larger UE4 playlist!
I followed along, implementing patrolling, investigation, and search sequences, and it was starting to look promising!
More Live Labs
By the time the next Live Lab rolled around, I already had a “working” AI system. I spent the session troubleshooting AI combat behavior, getting the guards to shoot at the player (which, ironically, was way harder than I expected), and picking up some extra AI tips from Trent.
Starting the Blockout
Beyond AI, I tackled smaller tasks like reworking my Switch Door System and experimenting with Niagara Systems.
But in the middle time I started blocking out the level!
With functional mechanics and AI, level design became way more fun. I sketched out a layout inspired by Groznyj Grad (MGS3) and Shadow Moses (MGS1), but I added verticality inspired by PSX-era platformers like Miss Pacman and Sly Cooper. The level is still in its early stages, but I can already playtest it and find the fun parts of the game.
Final Thoughts
This was AI Week, no doubt about it. I’m super happy with the progress I’ve made with Behavior Trees, and the game is looking more promising each day.
But with only two weeks left, I’m starting to feel the crunch. Even if I tackle all the big tasks, the sheer amount of small details might be my next challenge… but I also have a poretty big task ahead regarding the character movementation.
But while I keep on managing my weeks everything will be fine, and if I ever feel like I couldn’t deliver I can scrap some mechanics to reduce scope thanks to the Skateboard approach.
But well that 's Week 2!
2 weeks gone, 2 to go!
