Meshy Abominations

At the Game Developers Conference this year, we checked out some cool new tech that turns text and pictures into 3D models. We tried it out to see if it could help our 3D artists with our Crimson Knight capstone project. The idea was cool, but the results were a bit mixed.

The tech can create 3D models, but they ended up being way too detailed. Each model we made had way too many tiny parts (polygons), about 300,000 to 400,000, which is a lot more than we needed. We can fix these models to make them simpler without losing too much detail, though.

One problem with these models is that they have lots of tiny errors, like holes and random points that stick out. Fixing these could take a lot of time, which kind of beats the point of using this tech to save time in the first place.

However, the tech did a great job with textures and UV maps, which are the parts that make the models look good with colors and details. It even made a new Pokémon character look awesome by adding cool details to its feet, making them perfect for a dog-like creature.

Another good thing about this tech is that it gets the size of the models just right. We didn’t have to adjust the size at all, whether it was for characters, props, or big parts of the scene.

Even though there are some issues with the tech right now, it’s really good at making textures and getting the size right. It’s promising, and it could be super helpful for making video games in the future.