FMP - week 2 - Dragon creation

Week 2 was mainly about creating a base dragon character we can use for prototyping, and as an exercise to check if the artstyle I would want to do would be feasible for me. In this post I'll walk you through my process and how far I got this week.

Orientation

To get started, I started sketching dragon options that I could use for 3D modelling. I ended up picking the following one.

Before I dove into modelling this dragon I did some research to ensure my model would be able to work in game. The main thing is that the dragon will need to be animated so I looked into dragon rigs to check how other artists tackled this issue. I came across the YouTube channel of "Obsidian Dragon Creations", which is a channel centred around dragon related art, including 3D dragons. I checked his modelling timelapses to see how he rigged his dragon meshes (Obsidian Dragon Creations, 2020). It was quite a straightforward rig which was basically a wolf/cat rig with added wing bones.

I also checked the Tiny Dragons Unity Asset Pack that I had. The rig on that dragon model was very different and basically consisted out of a bunch of adjustment points rather than proper "bones". After checking the weight paints on each bone it became clear that it is actually a pretty straightforward rig that is cleverly set up to ensure the wings could be well animated. Since the Tiny Dragon Pack also comes with a bunch of animations it would be ideal if I could reuse the rig and remap it to the new dragon model I would be creating. I therefore set out to use this rig as my "goal" rig.

Another key part I needed to take into account was customizability of the mesh. I sketched out a couple of customization options for the tail and for the head. I figured I could probably use shape keys to make these changes possible without needing complete different meshes. That means I need to keep enough vertices in these areas that can be used for modification. Other customization options can be done via different textures and will not affect the mesh so I don't need to consider that just yet.

3D modelling

Having done the preliminary research I had the feeling I knew enough to get started. I added my sketch as a reference image in Blender and got to work. When I had a low poly version ready I decided to do a quick rig test with the tiny dragon rig, to see if I could remap it correctly. Although there was still a lot of clean up to do in terms of weightpainting, the rig seemed to work and I could move the wings properly. This gave me enough confidence (for now) to continue the character creation process to a high poly sculpt. After modelling as much as I could I made a copy of the mesh and applied a multires modifier to it. This modifier will help me in making a high poly sculpt that I can bake onto the low poly model that will be used in game. A high poly model would of course be way too compute intensive.

I chose to go the multires route as it was the most performant on my system, it allowed me pretty much full flexibility (much more than remeshing), and it doesn't create the triangulation artefacts that dyntopo does.

The scultping process went very smoothly and was pretty quick. But After scultping I realized I had forgotten to UV unwrap the base model which would be necessary to bake the high poly sculpt onto it. I had a really hard time with the UV. Even though I had run clean up jobs to take care of loose vertices etc. there were still issues with the base mesh whcih made UV unwrapping very difficult. However, after finding all the rogue vertices and edges and cleaning them up, I had my nice UV islands.

I moved on to the bake settings to try and bake the normals of my high poly sculpt to my low poly base mesh. Once I went to the settings I learned that Blender actually has added an option to bake from multires which wasn't there when I last baked normals! Unfortunately I had only UV unwrapped my low poly model and UV unwrapping the sculpted version would take me too long. So I had to stick with old fashioned baking from one object to the other.

Baking normals is always a challenge I find. Even though I'm starting to understand the bake settings better and I have a better understanding for how it works, I still run into normal map artefacts. I adjusted my base mesh to better fit the sculpt to get better normal maps and I played with the settings, but there were some artefacts I couldn't manage to get rid of. I tried to clean them up in Photoshop but unfortunately something seemed wrong and seams could clearly be seen when I attached it to a shader. At first glance it looked like some of the normals might be flipped, but I double checked that and it wasn't the issue.

At the moment I have not yet found a solution for this issue. If I can't manage to bake a proper normal map I might UV unwrap the sculpted version after all and try the baking from multires option. When I made a quick test on a simpler model this baking option seemed to give far better results than the baking from another object.

Research question

Other then working on the dragon model I also worked on my research question. Rob and I had come up with a potential joined question after meeting this week. Essentially we want to research how to target an audience that is defined by their interest in dragons, rather than their interests in a specific gameplay genre. This means that our audience is probably very diverse, with differing levels of gaming sikills and gaming knowledge. It is quite an interesting audience and it will probably need to be reached in different ways than we have learned in 740. I brainstormed about possible ways of formulating the question so that it leaves enough room for exploration and for ending up with a pitch ready dragon game artefact, while not being too broad and becoming too much to tackle.

So far I am happiest with:

"How do we address/cater to a target audience defined by their love of dragons, as opposed to their love of a specific gameplay genre, through a PC game?"

But I will need to align with Rob still on this as we are planning to have a joined question rather than an individual question, although we will each approach the question from a different perspective.

User research

As part of our user research Rob and I are planning to send out a questionnaire to answer some unknowns about our target audience. For the survey to be useful I figured it would be important to identify the dragon enthusiasts quickly and sieve out the non-dragon-enthousiasts. I worked on some questions we could use for that. I also worked on follow up questions to gain more insights into:

- why they buy dragon related content

- what kind of dragon related content they're interested in (i.e. in terms of art style or narrative direction)

- ....

-----WIP------