I have written before of the importance of good motion capture (MoCap for short) for live performances by Vocaloid characters (or 'avatars' as some call them), and here is an example of just how well this can work. Even at home, with relatively modest kit, nowadays very accurate and smooth motion can be recorded from a human performer and translated to a Vocaloid or other 3-D model.
I feel there is really no excuse for failing to achieve this standard at all times, even at fan-based events (as distinct from the much more generously resourced official events) as it is, I understand, far from being difficult to do nowadays, whereas just a few years ago it was nowhere near as simple for anyone smaller than a video production company..
Notice, for example, that although the human performer casts no significant shadow behind her – there is a faint shadow to the scene's right of her – Teto has her own shadow on the same scene (which was obviously recorded as a 'blank' five-minute chunk, i.e. with nothing happening, or perhaps as a single frame reproduced as many times as needed).
For a concert performance, of course, the performer would be indoors and in front of a strongly single-coloured backdrop that would be mapped to 'transparent' in much the same way as the long-established chroma-key (or 'colour separation overlay' in the BBC's terminology) – often known as 'green screen' or 'blue screen' in popular parlance – has done for decades: I first saw a demo of this at the BBC in 1967.
In this video, the song is Heart Beats and the human performer is known as Apricot* (yes, the asterisk is part of her nickname!). Note that the performances aren't quite identical, as Apricot*'s actions here are separate from those recorded with the special pads attached – which explains why her necktie in particular behaves somewhat differently in the outdoors breeze from how it was originally recorded and faithfully reproduced on Teto...