Being able to imagine what we want DUNJUNGUY to look like is one thing, but the pursuit of achieving that vision has taken a modest amount of trial and a disproportionate amount of error. In this second dev diary, I hope to reveal some of the things I learned working on the environment art for the game, as well as poke some light-hearted fun at the mistakes along the way. The process has been enriching, though, and I feel far more confident with the tools associated with the workflow. I’ll start by talking about the space the problem exists in, then discuss the different solutions we sought to employ to address it. Finally, I’ll look ahead toward the work that remains to be done, and how I’ll change my practices for more future-proof environment art.

When development on DUNJUNGUY began (back when we were still calling it “Project Everest”), I made a promise to the team to produce the various art assets the game would need. I figured that I had enough practice with digital art to get started, had a clear enough vision of the aesthetic, and was dedicated enough to follow through with my promises. I overlooked one key detail though - I am a massive, overconfident, naive dumbfuck. We knew that in order to have some good replay value for players, the DUNJUN needed to have a modular tileset, such that we could produce infinitely many rooms with a finite list of assets. This isn’t a new problem in games; I had a trove of examples from games thirty or more years old to teach me. They needed to flow together without annoying repetitions or “seams” where the tiles misalign; they also needed to not have too much contrast or noise, which would distract players’ eyes from the more important information on screen.

“I know,” I said to the team, with all three of my braincells overheating from the stress of the halfbaked idea they were producing, “I’ll draw the tiles on graph paper in pencil, scan them, and slap them into Unity. EZPZ!” It was a well-intentioned plan, with some basis in how I knew tesselating squares were supposed to work. I thought I was clever, and that I’d be adding something new to the wide variety of visual styles seen in the games industry. I found a guide on 9-slices in gameart and got started drawing them, used our college campus’ scanner, and was really underwhelmed with the results. I wish I had taken a moment to ask, “Why doesn’t anyone else do this?” The answer was simple: it looks awful.

We needed another solution, and I started over with an all-digital approach. Using a pirated copy of Adobe Flash CS3, I created a square (but not a perfect-square) and made sloppy measurements before drawing some basic wall and floor tiles. These seemed to work okay, but didn’t look great. I knew eventually that I’d have to replace them with something more sophisticated, but they enabled us to test our level generation algorithms and focus on other tasks. It was the first look of the DUNJUN, and it was awfully purple. But it was a start.

Eventually, we got to the point where we could work on the Tavern scene, like we promised our classmates we would. By then, I had borrowed a Windows Surface Pro 3 from the computer science department, and had been exploring Krita. We had identified a couple key features the tiles needed to satisfy to closer approximate industry practices, like having perfect-square dimensions and have different variants so they aren’t too repetitive. It was around this time that I had been drawing the initial Cave tileset, which was meant to look more organic and less constructed than the DUNJUN. One big mistake I made in the Tavern was that I tried to squeeze as much detail as I could into the textures, and the result was a very noisy and messy tileset. But like the initial DUNJUN set, it was progress, and it could function in the project for its purpose.

Time passed, we graduated from Purdue, and it was time to get serious about making the game stand on its own legs. I had taken an art minor to enrich myself, had a subscription to Adobe Creative Cloud, learned the newer versions of the software, and invested in a better drawing tablet (specifically the XP-Pen Artist 21 - it’s a great product that I’d recommend to anyone seeking an alternative to Wacom’s Cintiq line). We had to replace both the existing Tavern and DUNJUN tilesets. I had gotten a lot more comfortable with the workflow of creating tilesets, and had found a process that worked well for me. By now, I had decided on a final color palette, knew to create assets in perfect-square dimensions, drew tiles on a canvas in context with each other, and simplified the textures to not be so noisy. I sketched the forms, cleaned up the line art, and then painted flat colors underneath. I rendered the colors with some shading (mostly using Photoshop’s multiply and divide Layer Styles), and finished with a bonus step of creating normal maps in SpriteIlluminator (more on that some other time). Finally, after like two years of practice and failure, I had created some functional environment art that I was satisfied with.

For anyone working on independent projects, I can’t speak the merits of an iterative, revisionary philosophy enough. You’ll refine your skills a little more with each attempt, and eventually you’ll land on something that satisfies its requirements. Unfortunately, as you grow, so too will your standards. What terrifies me is the knowledge that I’ll create more environments for new regions of DUNJUNGUY and no longer be satisfied with the existing Tavern or DUNJUN sets. It all adds up to a higher quality game, but I need to learn to do things right the first time, and to know when to move on and put my efforts to rest. This, I think, will be a lifelong discipline.

Thanks for reading our dev diary! I hope this reaches someone working on a personal or professional project, and they gain something of substance from our mistakes and successes alike. See you next time.

Here's what the grid-paper/pencil attempt looked like in our scan.

Join our Patreon for more content and videos with this post!

Aga Pito Studios is creating games/software | Patreon