Facetracking avatar of a succubus
This is a face tracking avatar, that was created with Live2D Cubism software. The drawing has been created within Photoshop and then prepared for rigging. Afterwards the drawing has been rigged for the facetracking software VTube Studio.
The avatar follows the movement of the puppeteer in front of the webcam to create a live 2D character. Content creators use them to give their unique spin to their content.
I made an in-depth Making Of as a video: Making of Eliza. From character design to tracking set up, I did everything on this model.
Considering the character design, I already had some drawings of Eliza before I started with this project, so I only had to update some details for her avatar.
Drawings for face-tracking models have special requirements. However, one can just start with a normal sketch like I did. Planning the drawing symmetrical will save time later on, but isn't necessary.
All separate pieces of the body need to be drawn in fully on their own layers, regardless if they're visible in the drawing or not. So after being content with the sketch I had to create a separate layer for all pieces and drew their outlines on them, which looks quite chaotic at first. After giving each layer their respective color fill, the character looks a little more like a traditional drawing again (but this is because the overlapping parts aren't visible anymore, they still exist!)
Each part gets shaded similar to a "regular" Manga-styled drawing. Each part still needs to be completely separate, including it's shadows and whatnot. Since this results in a huge amount of layers (this drawing has over 600), I tend to organize them in groups and coloring the groups. A good organization is mandatory. After finishing the drawing I prepare the file for rigging in Live2D.
Within Live2D every piece will be organized on a texture atlas. Since this figure has a bikini, naked and uniform version all had to be fitted onto two texture sheets.
Within Live2D movements will be created by using deformers (a deforming net layer, hierarchy shown in the screenshot in the left corner) and setting their values to parameters. (Parameters shown in the lower middle.) The software will interpolate the inbetweens.
Eliza consists of 160 deforming layers and is able to close her eyes, move her body, eyebrows, mouth and head. I also added a "blush" and an "angry" mode. The whole model features physics (flying hair, eyelashes, wings etc.). The finished parameters can be mapped to facetracking software movements (shown on the first videos on this page), or animated traditionally with keyframes (like in the following showcase video).