Stylogenyin the style of VTuber Avatar and Stream Graphics

Digital native · 2016–present · East Asia, Global / networked

VTuber Avatar and Stream Graphics

Also known as Virtual YouTuber, VTuber, バーチャルYouTuber, Live2D streamer avatars, PNGTuber (sprite variant)

The graphic kit of virtual YouTubers, streamers who perform behind an animated anime-style character instead of their own face. An illustrator (the 'mama') designs a persona whose costume and colours tell its story and draws it in separable layers; a rigger turns it into a Live2D or 3D model that blinks, turns and lip-syncs with the performer's tracked face. On stream the half-length figure sits over game footage or a themed backdrop among chat and alert panels, framed by a name logo, overlays and thumbnails in its colours.

Make something in this styleGet advice in this styleMix it with another style

A sample layout drawn from the style’s design choices, until its AI image is made.

Use it with care

VTuber characters are owned designs, by agencies or by the performers who commissioned them, and early contracts often gave the character's rights to the company, so performers who left had to start again as someone new: draw an original persona rather than a look-alike of Kizuna AI, Hololive or Nijisanji characters, and credit the illustrator ('mama') and the rigger ('papa'). The performer's anonymity is part of the form, so never attach a real person's face, name or photograph to an avatar or speculate about who is 'inside'. Do not default to a sexualised cute-girl type: scholars argue that avatar idols perpetuate female stereotypes and that the virtual idol industry commercialises sexism (Lu et al., CHI 2021).

01

How it works

  • Start from a persona and let the drawing state it: an anime-style figure whose costume, accessories and hair colour spell out one identity (a shark hat for a girl from Atlantis, a reaper's motifs for the Grim Reaper's apprentice, cat ears, a devil, an AI), simple enough to grasp at a glance; the character does not age, and new outfits can be added later. By default the performer's own face and photographs stay out of every graphic, so the avatar carries the identity and the community builds it out with fan art under the character's hashtags (since about 2021 some virtual singers have also appeared in person).
  • Render the figure in anime-style line art over flat colour with its own shadow colours, blended with the airbrush where needed: a gradient across the bangs and the shadow they cast on the face, with the light in the eyes often drawn on a layer of its own. Hair can carry a dyed gradient, as in Kizuna AI's long brown hair with its pink gradient dye, and the character's standard colour becomes the key colour of its logo and kit.
  • Keep line and colour for every moving part on separate layers (lashes, whites, irises, mouth, bangs, side and back hair, clothing) so one illustration can be rigged in Live2D to turn its head, blink, lip-sync and sway with hair physics, without strands swinging over the eyes; build a 3D model for full-body motion. Early VTubers such as Kizuna AI were 3D, Nijisanji made the drawn half-length Live2D model the norm from 2018, and 3D returns for '3D debut' streams of song and dance from a motion-capture studio and for concerts; a PNGTuber swaps a few static sprite frames instead.
  • Show the character half-length and cut out, in a corner over game footage or centred before a designed backdrop, among chat, alert and title panels, so the screen reads on several levels at once; put several avatars side by side on one stage for collaborations.
  • Make thumbnails that read at list size: a big cut-out of the avatar, its expression set and eyes retouched, filling a third to a half of the frame and looking toward the title; a few short words in a heavy, simple face, outlined in white or black with a drop shadow so they do not sink into the background, each character tilted or resized on its own, in warm colours, red or gold, and clear of the video's time stamp.
  • Carry the character's standard colour, limited to two or three theme colours, and its motifs into a name logo, overlays, intermission screens and thumbnails so the whole channel works as one kit; the logo takes sparkles, stars, hearts or a detail from the costume (cat paws for a cat) and an outline that keeps it off the background.

Japan. Kizuna AI, made by the production company Activ8 with character design by En Morikura, posted her first video on 29 November 2016 and was the first to call herself a 'virtual YouTuber'; earlier animated hosts include Nitroplus's mascot Super Sonico (videos from 12 February 2010), Ami Yamato (first video 13 June 2011) and Weathernews's Weatheroid Type A Airi (2012). Cover Corporation's Hololive began with Tokino Sora (debut 7 September 2017). Ichikara (Anycolor from May 2021) launched Nijisanji in early 2018 around an iPhone X face-tracking app for Live2D models, promoting Live2D livestreams in contrast to the edited 3D clips Kizuna AI had made popular. Hololive opened an English branch in September 2020, agencies opened branches in China, South Korea, Indonesia and India, and by January 2020 there were over 10,000 VTubers

02

Its design choices

From Stylogeny’s vocabulary of design choices: defining choices make the style, characteristic ones are usual, occasional ones turn up now and then.

Layout & composition
Space & hierarchy
Typography
Colour
Form & shape
Surface & texture
Ornament
Imagery
Medium & production
Tone & intent
Avoids
Straight photography, Monochrome, Typography only, Photocopy & distress, Generous white space, Earth & mineral, Rational & objective, Authoritative & institutional

Palette

  • #ffffff
  • #ff7eb6
  • #5bc8f5
  • #b9a6ff
  • #ffd84d
  • #1e1b2e

Typefaces

  • Heavy, simple bold lettering for thumbnails

    a few short words outlined in white or black with a drop shadow so they do not merge with the background (CastCraft; Vtuberアナライズ), sized to read in the YouTube list, each character tilted or resized on its own and kept clear of the time display (Tsukino Mito's method, MoguLive, 28 January 2021)

  • A custom name logo drawn for each character

    one of the four parts of a VTuber's visual kit with the character design, the stream overlays and the thumbnails (VTuberデザイン大全, KADOKAWA, 2024), set in the character's standard colour and outlined or given a sticker-like edge so it does not blend into the background (Megohime, Clip Studio Tips)

  • Bilingual name lockups, the English and Japanese forms of the name set in different fonts, with smooth, plump, slightly thick letters (or all lowercase) for a cute character, as in Megohime's tutorial 'Simple Tips for Creating a Vtuber LOGO' (Clip Studio Tips)

Known for

  • Kizuna AI (Activ8), 'A.I.Channel', first video 29 November 2016

    character design by En Morikura, 3D model by Tomitake with Tda supervising; long brown hair with a pink gradient dye, a pink heart-shaped hairband and a white sleeveless top marked 'A', in a pink-and-white palette; the official 3D model was released free in December 2016 for MikuMikuDance, Unity and Unreal Engine (on hiatus from February 2022, back on 26 February 2025)

  • Kaguya Luna (VIC Inc.), channel opened 4 December 2017, original character design by the illustrator Mika Pikazo

    a mainly black outfit with Japanese-style elements and red, blue and yellow accents (no posts since October 2020)

  • Tokino Sora, Hololive Production's first virtual idol (debut stream 7 September 2017), illustrated by Ordan

  • Nijisanji's first Virtual Livers (Tsukino Mito, Shibuya Hajime, Higuchi Kaede, Shizuka Rin, Moira and Elu), recruited by Ichikara in January 2018 around its iPhone X face-tracking app for Live2D models, the start of Live2D livestreaming as the agency format

  • Hololive English, first generation

    Mori Calliope (debut 12 September 2020, illustrated by Yukisame), the Grim Reaper's first apprentice, and Gawr Gura (debut 13 September 2020, illustrated by Amashiro Natsuki), a descendant of Atlantis in a shark hat, who overtook Kizuna AI as the most-subscribed VTuber on YouTube on 30 June 2021 and graduated on 1 May 2025 with a short 3D concert

  • Tsukino Mito (Nijisanji), video on how she makes YouTube thumbnails (27 January 2021)

    text thick enough to read in the list, clear of the time display, each character tilted or resized, warm colours, and a cut-out of her own 3D model, its expression set and eyes retouched, edited in MediBang Paint

  • VTube Studio (DenchiSoft, on Steam from 12 March 2021), webcam and phone face tracking for Live2D models, with hotkey expressions and PNG items pinned to the model, used alongside OBS or Streamlabs

03

Make something in this style

Say what you need, and Stylogeny’s image studio draws it from VTuber Avatar and Stream Graphics’s design choices and checks it against them. It takes about a minute, and you get a link to share.

Rule out a choice, read the prompt, or blend in another style: take this brief to Mix styles.

04

Where it comes from, and what it’s close to

Before it

The style it grew out of, and the styles it borrowed from.

This styleVTuber Avatar and Stream Graphics2016–present · East Asia, Global / networked

After it

The styles that grew out of it.

No later style in the atlas grew out of it yet.

Alike in its design choices

Styles that make many of the same choices, related or not, besides those above.

05

Sources

  1. Wikipedia, 'VTuber': https://en.wikipedia.org/wiki/VTuber (Kizuna AI, Activ8 and the first use of 'virtual YouTuber', late 2016; Super Sonico from 12 February 2010, Ami Yamato from 13 June 2011, Weatheroid Type A Airi 2012; FaceRig; Nijisanji early 2018 and Live2D livestreaming instead of the prior focus on 3D models; Hololive with over 90 performers; English branch September 2020; VShojo November 2020; over 10,000 VTubers by January 2020; Gawr Gura passing Kizuna AI on 30 June 2021; Live2D textures and VRM 3D models; PNGTuber sprites; commissions up to US$2,000; platforms including Niconico, Twitch and Bilibili; avatars drawing on anime and manga aesthetics and moe anthropomorphism, designed by online artists)
  2. Wikipedia, 'Kizuna AI': https://en.wikipedia.org/wiki/Kizuna_AI (channel created 18 October 2016, first video 29 November 2016; character design by En Morikura; 3D model by Tomitake, supervised by Tda; long brown hair with a gradient hair dye of pink, heart-shaped hairband, pink-and-white palette; free 3D model December 2016; indefinite hiatus after the February 2022 tour, activities resumed 26 February 2025; Bilibili channel from 2017)
  3. Wikipedia (Japanese), 'バーチャルYouTuber': https://ja.wikipedia.org/wiki/バーチャルYouTuber ('mama' for the illustrator and 'papa' for the modeller; the model as 'gawa' (shell) and the performer as 'naka no hito'; Live2D physics for hair and clothing; non-human avatars; characters that do not age and can be redesigned; early contracts transferring character rights to companies; OBS Studio for compositing; '3D debut' (3Dお披露目) streams showing a new 3D model with song and dance from a motion-capture studio; since about 2021, especially among virtual singers, activity in both virtual and real form)
  4. Wikipedia (Japanese), '輝夜月': https://ja.wikipedia.org/wiki/輝夜月 (channel opened 4 December 2017; original character design by Mika Pikazo; VIC Inc.; black outfit with Japanese-style elements and red, blue and yellow accents; no X or YouTube posts after 30 October 2020, end of the fan club announced on 28 September 2021)
  5. Wikipedia, 'Nijisanji': https://en.wikipedia.org/wiki/Nijisanji (recruitment announced 11 January 2018 around Ichikara's iPhone X face-tracking app for Live2D; first generation; Live2D and livestreaming in contrast to Kizuna AI's 3D clips; renamed Anycolor 17 May 2021)
  6. Wikipedia, 'Live2D': https://en.wikipedia.org/wiki/Live2D (rigged models, usually anime-style, from 2D textures split into layers and warped into '2.5D' motion; Cybernoids founded 2006, renamed Live2D Ltd. 2014; face tracking with FaceRig and VTube Studio; in 2021, 70% of Cubism Pro users were or made VTubers)
  7. Live2D Inc., 'About Live2D': https://www.live2d.com/en/about/ ('dynamic expressions from a single illustration'; physics and lip-sync; used in VTuber streaming) and Live2D Cubism Editor manual, 'About Material Separation': https://docs.live2d.com/en/cubism-editor-manual/divide-the-material (cutting the illustration into parts such as eyelashes and eyeballs before modelling)
  8. Zhicong Lu, Chenxinran Shen, Jiannan Li, Hong Shen and Daniel Wigdor, 'More Kawaii than a Real-Person Live Streamer: Understanding How the Otaku Community Engages with and Perceives Virtual YouTubers', CHI '21 (Yokohama, 2021), https://doi.org/10.1145/3411764.3445660, author PDF https://www.dgp.toronto.edu/~jiannanli/resources/2021-chi-more-kawaii.pdf (half-body 2D avatars made with Live2D driven by facial capture; 3D avatars with full-body motion capture; only eyebrows, eyes and mouth animatable; personas such as vampire, devil, princess, fairy, alien or futuristic with matching appearance; avatars on one virtual stage in co-streams; Super Chat and channel badges; scholars arguing that avatar-based idols perpetuate female stereotypes, and sexism in otaku culture and the virtual idol industry)
  9. hololive production, talent profiles: https://hololive.hololivepro.com/en/talents/tokino-sora/ (first virtual idol, debut 7 September 2017, illustrator Ordan), https://hololive.hololivepro.com/en/talents/mori-calliope/ (debut 12 September 2020, illustrator Yukisame, hashtags #calliolive, #カリオライブ, #callillust, #カリイラスト) and https://hololive.hololivepro.com/en/talents/gawr-gura/ (debut 13 September 2020, illustrator Amashiro Natsuki, shark hat; now listed as an alumna)
  10. Wikipedia, 'Gawr Gura': https://en.wikipedia.org/wiki/Gawr_Gura (graduation announced 15 April 2025; final stream, a short 3D concert, on 1 May 2025)
  11. MoguLive editorial staff, 'YouTubeサムネイル画像制作のコツは？ 月ノ美兎が自分流の方法を紹介', MoguLive (28 January 2021): https://www.moguravr.com/?p=139615 (thumbnails checked as they look in the video list; text clear of the time display and thick enough to read; warm colours; single characters tilted or resized; cutting out her 3D model, setting its expression and retouching the eyes in MediBang Paint)
  12. Sae Oguri (supervisor), VTuberデザイン大全 あなたの魅力を引き出すアイデア集 (Tokyo: KADOKAWA, 22 May 2024), ISBN 9784046068040, publisher's page: https://www.kadokawa.co.jp/product/322312000227 (character design, logos, overlays and thumbnails, with work by about 20 creators)
  13. VTube Studio, Steam store page: https://store.steampowered.com/app/1325860/VTube_Studio/ (DenchiSoft, released 12 March 2021; Live2D Cubism models only; webcam tracking via OpenSeeFace and iOS/Android tracking app; hotkey expressions; PNG items pinned to the model)
  14. ClipStudioOfficial, with commentary by MUGENUP Inc., 'Let's animate illustrations with Live2D! Dividing character parts in CLIP STUDIO PAINT', Clip Studio Tips: https://tips.clip-studio.com/en-us/articles/2684 (Japanese: https://tips.clip-studio.com/ja-jp/articles/2684) (line art and colour layers kept apart for moving parts; eyes divided into lashes, whites, pupils and irises; the upper body usually on screen; hair must not swing over the eyes; a gradient across the bangs blended with the airbrush; the cast shadow of the bangs painted in its shadow colour)
  15. renoa, 'Modeling 101: layering the character', Live2D Creators Forum (February 2016): https://community.live2d.com/discussion/comment/731 (eye layers including an optional 'eye gleam (the light in the eyes)'; hair as bangs, side hair and back hair)
  16. Megohime, 'Simple Tips for Creating a Vtuber LOGO', Clip Studio Tips: https://tips.clip-studio.com/en-us/articles/10235 (an outline so the logo does not blend into the background, or a sticker-like feel; the character's standard colour, or its most dominant colour; a smooth, plump, slightly thicker font for a cute character, and all lowercase; English and Japanese in different fonts; sparkles, stars and hearts, or motifs from the design such as cat paws for a cat)
  17. CastCraft (Cascord Inc.), 'Vtuber向けサムネイルの作り方', updated 7 June 2026: https://castcraft.live/blog/310/ (white or black outlines and drop shadows so text does not merge with the background; red or gold; bold, simple fonts; short, large text; the character at a third to a half of the frame; gaze and pointing to lead the eye to the text)
  18. Vtuberアナライズ, 'Vtuberの初配信サムネで新規ファンをつかむ7つのポイント' (11 January 2026): https://coro-vtuber.com/vtuber-first-broadcast-thumbnail/ (bold, readable fonts with outlines or shadows; a large bust-up character with the face or gaze turned toward the title; theme colours limited to two or three; collaboration thumbnails with characters facing each other or in a circle)

How Stylogeny researches and checks a style: where the facts come from.