Enterprise Avatars
Photoreal 3D heads from photos – for digital humans and conversational agents at scale.
Deploy hyper-realistic digital human agents that look like real people – without scanning, CG artists and the wait.
A cloud REST API that turns a small set of standard photos into a high-fidelity, animatable 3D head – ready to power your conversational AI agents, virtual assistants, and customer-facing digital humans.
What It Does
POST 3–7 photos to the API. Get back a clean, high-fidelity, fully textured 3D head – rigged, blend-shape ready, and compatible with the animation and rendering pipeline your digital human platform already uses.
The head is generated from what is visible in the photos: overall head shape, facial structure, proportions, skin tone. The result follows the surface of the skin – not a generic template, not a stylized approximation – accurate enough to be recognisable, clean enough to animate.
The API
REST API, cloud-hosted. Output delivered to your destination URL or storage.
Two data modes:
POST /v1/head input: 3–7 photos (URLs or upload) output: 3D head mesh + photoreal texture + blend shapes time: 1–3 min
- Ephemeral – photos pulled from your URLs, head model pushed to your destination, zero retention on KeenTools servers. GDPR-compliant by design.
- Managed – model generation and storage on KeenTools infrastructure.
What You Get
- Geometry
- Clean, watertight 3D head mesh in MetaHuman topology – face, overall head shape, neck, shoulders. Uniform topology, identical vertex layout across every output. No retopology, no cleanup, no surprises downstream.
- Texture
- Photorealistic, projected and blended from the input photos. Accurate skin tone and facial features.
- 3 photos recommended minimum (front + 45° angles)
- Up to ~7 photos for full coverage
- Any camera: smartphone, mirrorless, DSLR, studio photography setup
- No depth sensor, no proprietary hardware, no calibration
- Accepts both real photographs and AI-generated face images – design a persona entirely in AI, then generate the 3D head from it
- Export formats: OBJ / FBX / GLB / STL / PLY
Applications
Customer-Facing Conversational Agents
Deploy a named, recognisable digital human as the face of your customer service, sales, or support operation. The agent can be modelled on a real brand ambassador, a designed persona, or a specific person – generated from photos in minutes, not weeks.
Pair the KeenTools head with your LLM, TTS, and animation layer (NVIDIA Audio2Face, UneeQ, or your own pipeline) for a complete, photoreal conversational agent.
Virtual Brand Ambassadors
Give your brand a consistent, photoreal human face – across web, kiosk, mobile, and XR. KeenTools generates the head; your platform handles the conversation. The result is a digital human that looks like a specific person, not an off-the-shelf avatar.
Internal Agents – HR, IT, Training
Use KeenTools to generate heads for internal-facing agents – onboarding assistants, IT helpdesk avatars, compliance training facilitators – from photos of real employees or designed personas.
Immersive Training & Roleplay Simulations
High-stakes conversation training – customer objections, difficult HR conversations, clinical patient interaction – is more effective when the simulated person looks real.
Multilingual & Localised Agent Variants
Generate multiple agent heads from different reference sets to create a regionally appropriate, culturally matched face for each market – at API scale, without separate CG production runs per region.
Privacy & Compliance
Enterprise deployments have strict requirements around biometric data. KeenTools is designed to meet them.
Ephemeral mode means photos never persist on KeenTools servers beyond the processing session – pulled from your URLs, processed, model returned, everything deleted. Your organisation controls where the data lives, before and after. No training on client data. GDPR-compliant by design.
On-premise deployment is available for organisations with data residency requirements or air-gapped environments.
How it works
POST a small set of photos and receive a clean, high-fidelity, fully textured 3D head – rigged and blend-shape ready. The head follows the surface of the skin from what's visible in the photos, so it's recognisable enough to represent a real person and clean enough to animate in the pipeline you already use.
# 1. Open a session — returns an avatar id and photo upload URLs
curl -X POST https://api.cloud.keentools.io/v1/avatar/init \
-H "Authorization: Bearer $API_KEY" \
-H "Content-Type: application/json" \
-d '{ "image_count": 5 }'
# 2. Upload each photo to its pre-signed URL
curl -X PUT "$UPLOAD_URL" \
-H "Content-Type: image/jpeg" \
--data-binary @photo-1.jpg
# 3. Start reconstruction on the session
curl -X POST https://api.cloud.keentools.io/v1/avatar/$AVATAR_ID/process \
-H "Authorization: Bearer $API_KEY" \
-H "Content-Type: application/json" \
-d '{ "focal_length_type": { "focal_length_type": "estimate_common" } }'
# 4. Download the result — OBJ + texture, delivered as a ZIP
curl -L -X GET "https://api.cloud.keentools.io/v1/avatar/$AVATAR_ID/get-3d-model?mesh_format=obj&texture=jpg" \
-H "Authorization: Bearer $API_KEY" \
-o head.zipGenerate your first digital human head
No hardware. No subscription. Start with a free credit allocation. Questions about blend shape compatibility, animation pipeline integration, or data handling? Reach the team directly.