Face Overlay Lab — mesh face on the blank bases

Base — targets first

Casting Deck — named characters, castable on any base

A character = every knob + palette + style, portable across the 1,100+ site bases. Chips cast onto the CURRENT base; base-anchored geometry (position/size/ear anchors) stays the base's own. Saved chips live in localStorage only — disk files are never touched.

My Avatars — save the complete look

→ Studio

An avatar chip restores EVERYTHING — base image, cast character, glasses/hats/necklaces with their dragged positions, colors, every gauge. Export downloads a portable Alive Agent Avatar package (.alive.json — self-contained, image embedded, sha-256 sealed); Import restores a package or a legacy avatar .json on any browser. IndexedDB only — disk files are never touched.

🛒 NetShow Alive Marketplace ↗

every published Alive Agent Avatar — browse, wake a card, use one on your site. Opens in a new tab; your work here stays put.

Use your avatar — where it goes

Two files, two jobs. ⬇ Export avatar (above) downloads .alive.json — the sealed, complete Alive Agent Avatar package (image embedded, sha-256 sealed). That's the file you place places. ⬇ Export character (Casting Deck) downloads a character .json — the look recipe only, for re-casting onto other bases here.

Your website — two lines: the alive-element.js script + <netshow-alive> pointed at your .alive.json. the recipe ↗
Universal Stage — tick “use on the Universal Stage” when you save and this avatar fronts the stage. open the stage ↗
Marketplace — list the sealed package on the open NetShow Alive marketplace, at your price or $0. browse it ↗
Embed anywhere — the configurator builds a copy-paste snippet for any page or widget. configure an embed ↗
Your own model or agent framework — .alive.json is plain JSON: load it yourself and drive the face from your agent. developer docs ↗
Email & social — export a short GIF of the living face for places that only take images. recipe 03 ↗

Bring your own AI + voice. $0/min · $0/sec — no meter, you pay only your own AI & voice usage: the model and the voice come from accounts YOU connect, on your own keys (your OpenClaw agent connections ride those same keys). Proven lanes today: OpenAI and Google Gemini adapters, plus a text adapter that fits any framework or self-hosted model — that's how Claude and kin ride. Studio voice lanes: OpenAI · ElevenLabs · Gemini — more named lanes coming; the adapter contract is model-agnostic. the adapter story ↗

What a .alive.json actually is: one honest page ↗ · the full manual: guide ↗

Photo lane — Mrs. NetShow (pre-blanked, live overlays). Upload any portrait through YOUR AVATAR to replace her; for a RAW photo, flip the blank plate ON to erase its real brows/eyes/mouth first.

Live test

idle

starts a billed live call — a real OpenAI Realtime session (your mic in, agent audio out) on the ?business= key; the CURRENT overlay face is driven by the REAL engine frames while connected

Drive

Emotions $0 · the agent's face channel

Save your Alive Agent Avatar

use on the Universal Stage list in the marketplace USD · 0 = free open marketplace ↗

saves the COMPLETE current look — base, character, accessories, every gauge — into MY AVATARS (this browser). “Use on the Universal Stage” makes this the stage's Hybrid Alive Agent Avatar. Marketplace listing publishes the portable .alive.json package to the open NetShow Alive marketplace.

Quick actions — save & test

idle

the two things you do most, right up top. Save lands the COMPLETE current look on the MY AVATARS shelf — the same save as the buttons below. Live test starts a billed real voice session — the same button as under the picture.

Your Avatar — upload a portrait

Three doors, one fit tool. Take a photo — use your camera right here: line up in the halo, shoot, then pick either lane for the shot (the camera turns off the moment your photo is taken; every frame stays in this browser). Upload as-is — $0: your photo keeps its real brows, eyes and mouth, and the animated Alive overlay rides them. Mute upload — billed: a real OpenAI image edit digitally removes the brows, eyes, eyelashes + mouth (your face's contours stay) so the animated face is the only face. Any shape works — portrait, landscape, even full-body: the fit tool auto-frames the head and you can drag/zoom to adjust. JPEG/PNG/WebP, 12MB cap; everything stays in this browser (IndexedDB), never on our disk.

How to shoot your portrait

straight (neutral) face, mouth closed, eyes open
shoulders visible, head like the outline
camera at eye level, even light, no shadows
plain background if you can

Adjust face — the mathematics stay live

Face

Face display

Brows

Brows version Hair amount Brow Y Brow length Brow len back Brow thick Brow curl Brow head Head square top Head square bottom Brow fade Brow real Brow rim Brow indep Brow emotion

brows

Eyes

Eye shape Lashes Lash alpha Lash fan Lash band Eyes Y Eye spacing Eye size Upper lid flat Lower lid flat Back corner bend Mid front bend Iris size Iris texture Pupil size White alpha Limbal ring Eyeball shade Fade lid top Fade lid bottom Catchlight Eyeliner fade Eyeshadow Eye emotion

iris eye outline white eyeball shade eyeliner fade eyeshadow lash upper lash lower

Nose

Nose version Nose ink Nose

nose ink

Mouth

Lips version Furrow depth Lip style Mustache Stache size Stache thick Mouth Y Mouth width Lip mass Lip ratio Lip ratio top Rest part Lips alpha Lip rim Mouth line Seam taper Mouth alpha Mouth version Teeth Teeth lines Teeth shine Gum line Smile teeth

Tongue stage Bottom teeth Tongue Lip shading Lip sheen Lip edge Top lip upper Top lip lower Bottom lip upper Bottom lip lower Dither grain Fade top Fade bottom Lip lobes Lip corners Mouth emotion

lip rim lips lip dither seam taper fade taper bow light mouth inside tongue

Face wrinkles

Wrinkles Smile lines Smile X Smile Y Smile size Forehead lines Forehead X Forehead Y Forehead size Frown lines Frown X Frown Y Frown size Crow's feet Crow's X Crow's Y Crow's size Tear troughs Tear X Tear Y Tear size Marionette lines Marionette X Marionette Y Marionette size Mental crease Mental X Mental Y Mental size Neck lines Neck X Neck Y Neck size Crinkle amount Crinkle X Crinkle Y Crinkle size

Overall

Size Left / Right Aperture Articulation Boil Breathing

Nuances — toggleable extras, all default OFF

Expr. lines Line strength Earrings Earring size Blush Blush amount

earrings blush

Accessories — the collection; drag them on the image, click to select + fine-tune

Flower Flower size Hat Hat size Glasses Glasses size Lens tint Necklace Necklace size

flower flower center hat glasses lens necklace

Colors — ink that reads on skin and on flat fill

brows iris eye outline white eyeshadow lid skin (blink) eyeball shade eyeliner fade fade lid lash upper lash lower nose ink lip rim lips lip dither seam taper fade taper bow light mouth inside tongue teeth gum earrings blush

Export downloads face-tuning-export-<base>.json (top-level zones/pal/ink — the shape tunedBase reads). This lab NEVER writes face-tuning.json or the live zoneTune keys; the download is the only output.

Voice & brain — your models, your keys

Voice

pick a provider, add its key below, then preview.

Brain (LLM)

API keys

Keys stay in THIS browser (localStorage) — never on our servers, never inside saved avatars or exported packages. Voice previews and agent runs dial the provider directly with your key and bill your account. The voice + brain CHOICE (provider, voice, model — never keys) saves with the avatar and travels in its package. ElevenLabs: create a custom voice in their Voice Lab, then add its ID here.