In my last post I assembled my first 2D vtuber. Inspired by how well it turned out, I decided to immediately try again with a couple upgrades:
- Eyes. This rig will have pupils that look around.
- Hair. My last rig was essentially bald. This one will have flowing hair that I will configure with hair physics.
- Mouth. The mouth will be broken up into different parts so that we can have more sophisticated speech movements.
Also, I simplified the stack from last time, mainly just eliminating GIMP from the process. I read that Procreate can sometimes format .psd files in a way that Live2D doesn't like, but I have not experienced this yet (I think).
Software
- Procreate for iPad ($10)
- Live2D Cubmism (Free 42 day Trial, $15/mo after)
- Vtube Studio Pro on Steam, iPhone (Free Trial Version, Pro Version $20)
Drawing in Procreate
My goal was to make a vtuber that would be easy to rig down the line, so I prioritized the design for functionality rather than beauty. Instead of wasting time trying to figure out how to draw anime eyes and perfect hair, I drew the simplest versions I could. Now that I had a stronger understanding of what each part needed to look like in order for me to animate it properly,I could separate all the layers in a way that would save time in Live2D later.
This time around, I made sure to separate individual parts as much as possible: left and right eyeballs kept separate, irises kept separate, front and back hair pieces kept separate. I learned that the more parts you have isolated, the easier it is to manipulate them in Live2D.
youtube.com
Rigging in Live2D
After the vtuber assets were drawn, it was time to rig them in Live2D. There are three main steps:
- Assign parts to the right folders.
- Create meshes.
- Set parameters for how each part will move.
This is a major oversimplification, but this is how my brain is understanding the process so far. In my next couple vtubers, I imagine my learned rigging process will get more particular.
Unfortunately, certain layers on my vtuber (cheeks, eye shine) went missing during rigging. After some quick Googling, I determined that I lose those parts because 1) I didn't mesh them properly and/or 2) they were lost when I made the texture atlas. I sincerely don't know what happened, but next time I am going to try and manually mesh parts to see if it makes any kind of difference.
I am learning that the rigging process is definitely the most tedious bit. There is a lot of assigning layers to body parts, then carefully tweaking their movements with "parameters" in Live2D, which are essentially keyframes.
This time, I put a little more elbow grease into designing the mouth. The mouth was divided up into lips, teeth, and tongue, with each one moving independently for maximum control. I googled "Mouth Rigging for Vtubers" and found this helpful resource from vtuber Kira Omori.
Using this as a guide, I marked the parameters for about half of these mouth forms. I could have gotten more in-depth and intricate, but again it was incredibly tedious and I think I will have more patience next time.
Hair physics was also a new addition, and its implementation was relatively simple. I just chose the layers that I wanted to have their own physics, meaning they would bounce or sway when the character moved. I used the tutorial below to figure out my hair physics.
Key Takeaways
I now understand the 2D vtuber creation process better, which will allow me to plan out my time more effectively in the creation of more vtubers. I now have a couple characters in procreate that I'm drawing - my next creations will hopefully look a little bit cooler than vtuber 002, but now I know what goes into properly separating layers and what they should ideally do in Live2D cubism.
There are many 2D vtubers that move so well that they look like they are 3D models. I'm going to watch more tutorials to see how to get this desired effect. I am most impressed by the work of Kira Omori, who I referenced above. Their vtuber creations look very clean, professional, and 3D (despite them being 2D in actuality. Achieving this level of refinement will definitely take practice and patience.
Here is a demo of the final working rig:
Top comments (0)