CG / VFX

3D Characters

From reference photos to a sculpt-ready base mesh – in minutes.

Free up sculpting time. Get a 70–80% accurate base mesh with matched cameras.

A photo-to-3D head engine for character artists – turning a set of reference photos into an aligned base mesh and solved cameras, ready to take into ZBrush, Maya, or Mudbox to finish off with a hyper-realistic character.

Try in Playground Read the docs

Loading preview...

What It Does

Feed in reference photos – KeenTools API automatically estimates focal length per image, aligns everything to a shared 3D space, and solves camera positions. The output is a base mesh with accurate likeness and proportions, plus the matched cameras – ready to export into your sculpting software of choice.

The API / Application

Available as a web application and via API for studio pipeline integration.

input:  reference photos (any camera, any source, mixed sets supported)
output: base mesh + solved cameras + texture
time:   minutes

What You Get

Base Mesh: Likeness-accurate head geometry derived from your reference photos – proportions, primary forms, and overall structure in place.
Automatic Focal Length Estimation: No need to know or match camera specs by hand. Mix reference photos from different cameras, different sessions, even photos sourced from the internet – KeenTools estimates the focal length for each image and aligns them into a consistent 3D space regardless of where they came from.
Solved Cameras: Camera positions are calculated and exported alongside the mesh. Bring both into Maya, ZBrush, or Mudbox – no manual camera matching, no eyeballing alignment.
Texture: Base texture projected from your reference photos, giving you a working starting point for further refinement in Substance Painter or your texturing tool of choice.
Export formats: FBX (including cameras), OBJ, GLB – built for direct import into industry-standard DCC tools.
MetaHuman-ready, when you want it.: If MetaHuman is your destination, output in MetaHuman topology – compatible with Conform from Template Mesh and the broader Unreal pipeline.

Applications

Client Concepts and Approval

Win the pitch faster. Generate a solid likeness base mesh in minutes instead of days, and put a real, approvable face concept in front of the client before committing to detailed sculpting work. A/B different likeness directions quickly, get sign-off on the right match before moving into the high-cost phase of the pipeline.

Likeness from Incomplete or Historical Reference

Reconstruct a likeness from whatever reference actually exists – archival photos, press stills, fan shots, or any arbitrary set of images. Recreate a likeness for a person who is no longer available to be photographed or scanned, working from whatever images can be gathered.

Concept-to-Likeness for Original Characters

For characters that don't exist as real people, generate a reference set with AI image tools, then run it through KeenTools the same way you would real photos. Get a sculptable base mesh for an entirely imagined character, without a physical actor or model.

Remote, Hardware-Free Likeness Capture

Capture an actor's likeness without a scanning rig and without requiring the actor to travel. Photos can be taken anywhere – on set, at a personal appearance, or sent in by the actor or their team – and turned into a base mesh remotely. Remove the hardware dependency entirely.

Crowd and Background Character Generation

For projects needing a large number of distinct, plausible faces – crowds, background NPCs, secondary cast – generate base meshes at volume from photo or AI-generated reference sets.

How it works

Feed in reference photos – the API automatically estimates focal length per image, aligns everything to a shared 3D space, and solves camera positions. You get a likeness-accurate base mesh plus matched cameras, ready to export into ZBrush, Maya, or Mudbox to finish off with a hyper-realistic character.

# 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.zip

Get your base mesh in minutes

Try in Playground Talk to us about your pipeline

No hardware. No subscription required to start. Questions about Maya or ZBrush integration, camera export, or studio deployment? Reach the team directly.