Overview

Avatars lives under Design in the console sidebar (/avatars). It's Intempt's library of AI-generated brand characters: people you can reuse across marketing images with a consistent face, body, and voice instead of generating a new person every time.

The page's own hero banner puts it plainly: "Bring fully custom, real-time intelligent avatars directly into your websites, products, apps, and services." Every avatar you see is either a built-in preset or a custom character your team created, and both live side by side in the same library.

How it works

Browsing your avatar library

The list combines built-in presets and your custom characters into one view. You can:

• Search by name or category.
• Filter by source (my characters vs. built-in presets), category, gender, and ethnicity.
• Sort by "My characters first," "Presets first," "Recently updated," or "A → Z."
• Switch views between Feature (card grid), List (rows), and Table.

If your project has no avatars yet, the page shows "No avatars yet" with a "Create your first avatar" button. If a search or filter matches nothing, it shows "No avatars match" instead.

📘 Media pending

Screenshot of the Avatars list page hasn't been captured yet.

Creating an avatar

There's no blank form for building an avatar from scratch. Every creation path opens Blu, Intempt's AI chat panel, scoped to the avatar skill:

• The "New" button offers "Generate with AI" (opens chat with no pre-filled text) or "Upload file" (accepts .md, .txt, or .json files describing the character).
• In Feature and List views, a "Custom avatar" tile or row opens chat with the prompt already filled in: "Create a new avatar from scratch. I'll describe them and upload reference photos."

Either way, you describe the character in chat and Blu generates the character record, including a main reference image, once the conversation produces one.

The character detail page

Opening any avatar splits the page into two panels. The left panel holds images and generated metadata; the right panel toggles between an editable Settings view and a read-only AVATAR.md view showing the character's full markdown brief, with a button to copy it.

On the left panel:

• Main reference image: the character's primary portrait. For custom avatars, a button lets you regenerate it. Because a new main image implies a new face, regenerating it re-renders every other image for that character using the new main as the identity anchor.
• Reference images: up to 3 fixed slots: Three-quarter, Full body, and Product in hand.
• Emotions picker (custom avatars only): 10 fixed cinematic presets (Explosive rage, Manic laughter, Villainous glee, Total shock, Extreme fear, Joyful defiance, Pure anger, Raw grief, Cold intimidation, Extreme shame) plus a free-text custom render option. Every avatar is capped at 9 renders total.
• Renders gallery: every emotion or custom render generated for the character so far.
• Demographics: gender, age range, ethnicity, and build.
• Expressions, verticals, posing, and mood range: shown as tag lists.
• Voice: a text description plus four sliders (Stability, Similarity, Style, Speed) and a play/stop preview button. The preview button is disabled if the character has no voice assigned yet; if playback fails (for example, the voice provider is unavailable), it shows an inline "Preview failed" message.
• Appearance, wardrobe, and photography notes: free-text fields for anything else that shapes how the character gets rendered.

For custom avatars, an Edit button switches both panels into inline edit mode with a Save/Discard footer at the bottom of the page.

📘 Media pending

Screenshot of the character detail page hasn't been captured yet.

Presets vs. custom avatars

Built-in presets are read-only: you can view every field, but editing and deleting are reserved for custom avatars. To make a preset your own, use Remix. It copies the preset's metadata into a new custom character, but doesn't copy any images. You generate the new character's images from chat afterward, the same way you would for any custom avatar.

Use cases

1. Standardizing a spokesperson. Reuse the same AI character across every marketing image instead of generating a new person each time.
2. Starting from a preset instead of a blank brief. Remix a built-in character into your own editable copy rather than describing demographics and wardrobe from scratch.
3. Building a full emotion set for a campaign. Generate several of the 10 fixed emotion renders so a character has coverage for different ad placements.
4. Giving a character a consistent voice. Set a voice description and tune the stability/similarity/style/speed sliders so scripts sound the same everywhere the character speaks.
5. Bringing in an existing character brief. Upload a .md, .txt, or .json file describing the avatar instead of typing it into chat.
6. Sharing a character's brief with a teammate or agency. Copy the AVATAR.md markdown from the detail page instead of re-describing the character by hand.
7. Refining a character's identity. Update photography notes or wardrobe, then regenerate the main image and renders together so every image reflects the same updated identity.
8. Auditing what's built-in vs. custom. Filter by source to separate your own characters from presets before deciding what to remix or delete.

Where to go next

• Brand Kit – manage your brand's broader identity, styles, and voice.
• Managing Avatar Emotions – a full walkthrough of the emotions picker and the 9-render cap.
• Creating a Custom Avatar – step-by-step avatar creation, from chat prompt to a finished character.