AR Self-Guided Tour - Sprint 2
My name is Kyle Ramser, and I am currently working as a 3D Artist on an AR self-guided tour being developed for The Abbey of Our Lady of New Clairvaux. We just finished our second sprint, and as a 3D Artist, the process has not been very different from my usual experiences in game development, apart from a few key differences. However, the few differences that there are have made the experience a bit more difficult than making video game assets.
As for the work that I completed during sprint two, I was able to complete the models and UVs for the sacred stone and the refurbished, or “restored” stones. This included high-poly models and normal stamps, mostly to be used on the sacred stone to show damage and stone mason marks. I feel as though I could have for sure done more this sprint, but with the spring break being dead in the center of the sprint, and with some unfortunate personal events occurring during the break, I’m not entirely upset with the work that I finished. The sacred stone was one of the models I was dreading getting finished, and I’m glad I decided to get it out of the way first so that I can move on to slightly simpler models and get a higher number of them done in a sprint.
This brings me to the key differences in modeling for an AR application like this, compared to modeling for a video game. The major difference here that I noticed is the accuracy required in the models I am making. When creating game assets, I mainly need to focus on references and inspiration, but I am otherwise building the models from scratch. While this sounds as though it would be easier when trying to copy an already existing object, it has honestly made it much more difficult for me to try to replicate the exact damage marks, misshapen bits, and imperfect curves of models like the sacred stone. The stone being asymmetrical because of its damage and age makes it even more difficult to model as well, since I can’t scale points and move each side of the stone together.
The only other difference that I noticed and struggled with a bit was trying to find the perfect balance between accuracy and simplicity for the models. For most game assets, simple props like crates and walls can be very basic, low-poly models with texturing carrying most of the weight. With these models, though, they are being run through the web, so they need to be fairly low-poly, but also need to be detailed recreations of the actual objects that they are based on. This has been a difficult balance to find and land on, which is why I am going to try to lean into high-poly baking in Substance Painter to make up for some of the missed detail. With this change, I am worried that the texturing process will end up being a lot more difficult and will take significantly more time, but I believe this is necessary for our program to run smoothly on all devices.
So, overall, this sprint was a bit of a weird one with the break right in the center of it, but it has been a good learning experience. I have a strong grasp on how I will be doing the models in the future, and I will now be able to start working on the textures and figuring out what will work best for these. I am hoping to get more cards finished next sprint than I did this sprint, and get all of the modeling and UVing done, as well as a bit of texturing, so that I can focus fully on textures after this next sprint.
Comments
Post a Comment