The following is required, and very helpful when modeling:
Only one of the reference artworks needs to be in color, the others can just be line-art or sketches.
2K textures are prone to pixelation on extreme close-ups, though some tricks can be done to use more texture space for areas that will be closer to the camera (like faces and hands). 4K textures look much sharper for the entire model.
Textures are one of the most crucial parts of visual fidelity. Specifically on character close-ups. Textures store information for model color in lit and shaded areas, and can even define which areas should always be lit or shaded!
17 visemes covers very basic facial expressions, but is ideal for someone who only needs speech and blinking with a few extra facial expressions. Someone who wants full-featured facial tracking (like with iOS ARKit) will want to select the 52 shape option.
Facial expressions can either be driven by shape-keys or bone deform. My recommendation is to use shapekeys. These are compatible and callable in logic in almost every software. They are also blendable. Whereas bone-based poses are more complex to call and control via scripting in most software. Particularly game engines.
This step is crucial for animation. A good rig makes animation much easier, and allows you to be creative without dealing with complex motions manually (such as keeping a hand on a steering wheel while moving the elbow and shoulder naturally). This sort of motion is simplified with Inverse Kinematic control (or I.K. control). Let my rigs do the heavy lifting for you and save tons of time animating!
Add animations to the rig, which can be exported anywhere!
