CAGD 373: Final Presentation

ePortfolio

Game Scene Final Project Post

Publication - 12/17/25

The Goal

            The goal for this game scene was to recreate the Comet Observatory from Super Mario Galaxy. Recreating this complex scene tested our entire team's ability to model, color ID, texture, and implement it into a game engine.

The Role

            As lead and producer in this class, I have experienced a very eventful time in this class managing my team by organizing workloads with cards and ensuring everyone is up-to-date on work. I learned many lessons with regard to how to manage a team better and also how to manage a Trello board better to ensure that each member is on the same page, so work is done efficiently and only once without iterations being necessary. There were many chaotic moments for me, mostly due to my allowing of work to be done up until the moment we present for a sprint review. This often led me to frantically organize the Trello board up until the moment we presented. I've learned from this, and I will not allow it anymore. It creates unnecessary headaches for me for a small gain overall to include in the sprint review.

The Game Scene

The Original

            

            Another name for the Comet Observatory is Rosalina's Observatory. Regardless, this scene is crucial for Super Mario Galaxy, as it serves as Mario's main hub between levels, or galaxies, as they're called in this game. Mario accesses these galaxies through six different domes located around the Observatory. The Observatory serves as a stark contrast to the cold depths of space, one might think, as the environment is very warm and comforting. The audio is an orchestrated waltz that feels alive and cozy; many Lumas and Toads are running around the place to make it seem less lonely. When Nintendo designed this location, they really made it feel like home in space; they even named each dome after a conventional room in a home, such as the Kitchen and Bedroom.

Our Version

            This is the final product of our version of the Comet Observatory. I believe we nailed the outward projection with them looking very similar with minimal differences. Most assets in ours are very realistic compared to the real versions. The whole scene makes me proud, but the part that sounds out the most to me is my texturer's work on the spire, as it really ties the whole scene together. We made sure to include music in hopes of creating a similar atmosphere.

            When we started this project, we aimed to get every last detail of the scene done and implemented. This went partially to plan; we were able to get most models done and created. However, due to a late start on texturing, we unfortunately had to scrap some assets, or they just got lost in the mess of FBX files.

            Many of the interactables that are present in the real scene were recreated by my programmer; these include 1-UPs, Power Stars, Functional Pipes, and a power-up that makes the character fly. We had a few more, but unfortunately, we had to scrape some in the scramble to get everything ready for submission.

My Work

            The first section I really tackled in the Observatory was the Terrace Section. This was pretty tedious for me, as I ended up doing multiple iterations I was never happy with my outcomes. Eventually, I landed a model that I was happy with. Learning from building the terrace, I applied my new methods to the library path, which required more precision as it would be one single model instead of multiple. The biggest issue I had doing these assets was learning the Boolean tool. Later on, I added pipelines that connected the domes to the "power supply" of the Observatory; this took a very long time to do.

            These are the same materials after I textured them. I believe that these came out accurately and pretty well done. I created materials using Adobe Substance Sampler and screenshots of the materials in-game. These ended up looking better than I had expected. The pipes also came out pretty nice.

            More adventures regarding the Boolean tool appeared when I created these intricate stairs. Took many revisions until I was satisfied. I spent fixing Ngons, but this seemed the best way to make those pipe cutouts.

            I ended up texturing this whole section, including those steps. This texture set looks very accurate to the real version, as I used Sampler again with screenshots.

            Probably the most difficult section I needed to model was the library. The shape was so odd and exact that I had to spend longer than I would've liked to make it accurate to the real one. To think that model all started from a singular cube is something. Now it's a bunch of meshes combined.

            I ended up texturing this material set as well, I think the final product is good; the books, however, for the bookshelves they could've been a bit better. Sampler wasn't so kind with those screenshots. Yet the finished product was pretty accurate, so I'm not disappointed.

            This is the bedroom floor I modeled; it's missing faces, as we would duplicate an asset in Unreal to save texture density. This model also came out pretty accurately.

            For the tower's assignment, I created a launch star. At the time, I wasn't happy with it, so this time I took a new approach by overlaying an image on it to help outline the shape as I went. This worked tremendously as it looks way better than my older version, and I am finally happy I learned how to model star-shaped objects.

            I also worked on this red star power-up that makes Mario fly. This was created using similar methods, and I'm really happy with how this one came out.

Problems/Solutions

            During the development of the game scene, while the team was still in Maya, everything seemed to be going strong and on course. And it was on course for a bit until we kept modeling and modeling, but never started texturing. We started texturing at the very end of Sprint 5, meaning we had to texture the entire scene within a few days so it could be submitted. This was a cramming session for everyone on the team. All our organization went out the window as it was models being passed around for texturing and exporting. In the midst of the chaos, we ended up losing some models that should've ended up in the final scene. These were small casualties, however, due to the constant issues we were having exporting these textures to Unreal and having them work. Many had to do double time on textures just to get them all done in time. My programmer had to abandon some interactibles to help finish the textures, leaving many of them unfinished, unfortunately. On the last weekend we had for the project, I ended up staying up for 23 hours to get all the textures working in Unreal for my programmer so he could assemble the scene while I napped. We cut this project so close, with us having nothing to show at 3pm on submission day. On the last crucial day, my programmer and I were the only ones doing work towards the project, which didn't help the progress go smoothly. Instances of poorly modelled assets also halted our progress during crunch time as we had to constantly revise them on the fly.

            Going forward, to make sure these steps never get repeated again, I will make sure that certain steps are taken to prevent dragging feet. We should have stopped modeling around sprint 3 with textures starting around sprint 4's start, which would have given us an extra couple of days to work out the issues. I was too lax as lead/producer as I didn't enforce deadlines or roadmaps I planned, next time I will do this as it backfired horribly on me. Dividing the workload better in the later steps would also be beneficial as the programmer and I were suffering the last day. Being more thorough in the verification section would have prevented the halts in progress we had during the crunch weekend. 

The Result

                The big takeaway from this was to prioritize certain aspects of work based on the timeline, as not doing so costs the team heavily. Additionally, dividing the workload towards the end more equally would have helped too. Many things were learned through this project, and all these I will use to become better in later production classes. For now, I am just happy we have a scene to present to the class. Having it feel similar to the real one was really the icing on the cake.