Lil Miquela’s first Instagram photo went up in April 2016. It looked like a normal lifestyle shot: a young woman in soft afternoon light, slightly off-center, slightly off in a way you could not quite name. By 2026, she has 2.6 million followers, an industry of imitators, and a market around her estimated at $8.3 billion. The question this article tries to answer is the unglamorous one underneath the spectacle: how virtual influencers are actually made. Not just the rendering, but the persona, the voice, the pipeline that takes a sketch and ends with a character your feed believes in.
Step one: the persona, before the polygons
Most teams start with writing, not modeling. Before any 3D software opens, a designer or strategist drafts the character’s bio: where she is from, what she cares about, what she is bad at, what kind of jokes she would or would not tell. Brud’s documents about Miquela described her as a 19-year-old Brazilian-American with a music habit and a tendency to overshare. Aww Inc., the Tokyo studio behind Imma, worked from a list of fashion references and cultural markers before deciding on the bobbed pink hair that made her recognizable.
The persona work matters because it is what the audience eventually bonds to. A character with no inner texture reads as a 3D demo. A character with a consistent voice, taste, and history can hold attention across hundreds of posts. This is the layer that separates a brand mascot from a creator audiences treat as a peer.
Step two: the 3D modeling pipeline
The traditional path from sketch to feed uses 3D animation software. Artists sculpt the character’s geometry in Blender or Autodesk Maya, refine the mesh, paint skin and fabric textures, and rig a skeleton so the character can be posed. The model is then placed into a virtual environment, lit, and rendered into a still image or short clip. For high-end characters like Miquela or Brazil’s Lu do Magalu, this is closer to a film VFX pipeline than to phone-based content production.
Several details get more attention than people expect. Skin shaders typically use subsurface scattering, so light appears to penetrate the skin a fraction of a millimeter the way it does in life. Hair is often simulated strand by strand. Eye geometry includes a small refractive layer on top of the iris. The uncanny valley collapses or holds depending on whether these details work together.
Once the model is set, the team builds a library of expressions and outfits that can be reused. A virtual influencer who posts three times a week needs a workflow that does not start from scratch every time, which is why so much of the production effort goes into reusable assets rather than one-off renders.
Step three: the AI-driven approach
A growing share of virtual influencers skip the 3D pipeline entirely and use generative AI image models instead. Tools like Midjourney, Stable Diffusion, and proprietary diffusion pipelines can produce photographs of a character from a text prompt once the character has been “anchored” with reference images and fine-tuning. Smaller studios and solo creators can now ship a working virtual persona without a single 3D asset.
The trade-off is consistency. A 3D-modeled character looks the same from every angle because she is the same model. A diffusion-generated character has to be re-grounded for every post; if the workflow drifts, the face starts to change, the audience notices, and the persona breaks. Most serious AI-driven studios solve this with character LoRAs (fine-tuned adapter weights that hold the face stable across prompts) and a strict approval queue.
A hybrid approach has become common: build the character once in 3D for a baseline reference, then use generative AI to produce most of the day-to-day posting. This is, for now, the path that scales best for virtual influencer marketing teams, who need a high posting cadence without the cost of full CG production for every image.
Step four: voice, motion, and personality
Image is only one layer. Video posts, livestreams, and short clips need movement, and movement is harder than it looks. Motion capture suits and head-mounted facial cameras are common: a human performer drives the body and face of the character in real time, similar to how VTuber pipelines work. For pre-rendered clips, keyframe animation is still used for fine control, especially around facial micro-expressions.
Voice is its own layer. Some characters use voice actors recording in studio. Others use AI voice cloning trained on a reference set. The most polished operations maintain a small bible of phrases the character does and does not use, so that the written caption, the spoken voice, and the visual style all read as one consistent person.
Personality, finally, is a layer that lives in the writers’ room. Even the highest-fidelity render falls apart if the captions sound like a brand. Teams that take the persona seriously typically have one or two writers who own the voice and review every post.
Step five: the content engine
A finished virtual influencer is, in operational terms, a small content studio. A weekly schedule might include two to three feed posts, daily stories, one short video, and ongoing reply work in comments and DMs. Some of that reply work is human; some of it is increasingly handled by conversational AI tuned to the character.
This is where the line between “virtual influencer” and “AI conversational companion” begins to blur. A character who can hold a real-time conversation with thousands of followers at once is no longer just a face in a feed. Many of the same techniques that go into building an AI conversational companion (persona writing, voice consistency, retrieval-augmented dialogue) are now part of the standard virtual influencer stack, especially for characters whose business model depends on sustained one-to-one engagement rather than passive following.
Why the build matters
A common assumption is that how virtual influencers are made is a purely technical question, separate from why people engage with them. The opposite turns out to be closer to the truth. The choices in the production pipeline directly shape the kind of relationship the audience forms.
A character built with a heavy persona layer, consistent voice, and an active reply workflow tends to attract a smaller, denser audience that feels addressed. A character built primarily as a rendering exercise tends to attract a larger, thinner audience that watches. The audience can tell which one is which, even without language for it. For studios that care about long-term audience retention, the persona and voice work has steadily become the most important part of the pipeline, even more than the visual fidelity.
The top virtual influencers tend to share that pattern: the ones that have held attention across multiple years are not always the most photorealistic. They are the ones whose creators kept the writing honest. The technology is the floor. The relationship the build supports is the ceiling.
FAQ
How are virtual influencers actually made?
Most are built through one of three pipelines: a 3D modeling and animation pipeline (Blender or Maya, with motion capture for movement); a generative AI pipeline (diffusion image models anchored on reference shots and LoRAs); or a hybrid where a 3D base model anchors a faster AI-driven posting workflow. All three pair the visual layer with a persona document, a voice, and a content team.
Do virtual influencers use real people behind them?
Often, yes. Motion capture performers drive movement for video work, voice actors or AI voice clones provide audio, and human writers handle captions and DMs. The character is the public-facing persona; the production is usually a small team.
How long does it take to make a virtual influencer?
A polished 3D character with a full asset library typically takes a small studio several months. An AI-driven character can be live in weeks. Sustaining either at a believable posting cadence is a continuous operation, not a one-time build.
How much does it cost to create a virtual influencer?
Costs vary widely: a solo creator using AI tools can spend a few thousand dollars to launch; a studio-produced 3D character with motion capture and a dedicated writing team can cost six figures upfront and a five-figure monthly run rate. Industry trade publications and Statista’s virtual influencer market overview track the broader spending trend, which is moving toward AI-driven workflows for cost reasons.