
KeenTools ships FaceBuilder 2024.1 for Blender
FaceBuilder 2024.1: Streamlined 3D Head Generation
Keentools has released FaceBuilder 2024.1, a significant update to its flagship 3D head generation tool. The new version focuses on automation and workflow efficiency — automatic face detection from photos, one-click texture creation, and improved UI responsiveness. For more information on FaceBuilder capabilities and licensing, visit Keentools official documentation. For character artists and archviz professionals who need realistic heads quickly, FaceBuilder 2024.1 is a major step forward.
We tested 2024.1 extensively at Super Renders Farm, and the improvements to photo-to-head conversion are tangible.
FaceBuilder is one of several add-ons that significantly improve Blender production workflows. For a broader look at tools that accelerate rendering and scene preparation, see our guide on essential Blender add-ons for faster rendering. What previously took 20–30 minutes of manual head rebuilding now takes 5–10 minutes with automatic face detection enabled.
What's New in FaceBuilder 2024.1
Automatic Face Detection
The headline feature: FaceBuilder now automatically detects faces in your reference photos. You no longer need to manually place landmarks. Instead:
- Import a photo reference
- FaceBuilder analyzes the image and places landmarks automatically
- Refine the mesh in seconds (instead of minutes)
How It Works:
- Uses machine learning to identify facial features (eyes, nose, mouth, cheekbones)
- Places 70+ landmarks on the face automatically
- Generates an initial mesh that's 80% correct
- You refine the remaining 20% for anatomical accuracy
Time Savings: Manual landmark placement in FaceBuilder 1.0–2.0 took 15–20 minutes per head. Automatic detection now handles this in 30 seconds, leaving you to fine-tune in another 5–10 minutes.
One-Click Texture Generation
Previously, you'd generate a head mesh in FaceBuilder, then manually author textures in Substance or another tool. FaceBuilder 2024.1 can now generate a complete texture map set from your photo reference in one click:
- Diffuse (Albedo) — Color information from photo
- Normal Map — Surface detail and micro-geometry
- Roughness Map — Skin specularity (forehead shinier than lips)
- Displacement Map (optional) — For detailed micro-displacements
These aren't placeholder textures — they're production-ready for most renders. You can refine them in Substance or Blender if needed, but the default output is immediately renderable.
Improved UI and Performance
The 2024.1 UI overhaul focused on clarity:
- Cleaner panel layout (all tools visible without scrolling)
- Faster photo import and landmark display
- Real-time mesh refresh as you adjust landmarks
- Better viewport responsiveness on large meshes
FaceBuilder Workflow in Blender
Step 1: Import a Photo Reference
- Open Blender and create a new scene (or open an existing project)
- Install FaceBuilder add-on (if not already installed):
- Edit > Preferences > Add-ons > Install > select keentools_fbx.zip
- Enable the add-on
- In the 3D Viewport, open the FaceBuilder panel (sidebar, FaceBuilder tab)
- Click Add Face Builder Object (or +)
- In the Headshot section, click Add Photo
- Select your reference image (portrait or 3/4 view, well-lit)
Step 2: Auto-Generate Landmarks
- Once the image is imported, click Detect Landmarks (new in 2024.1)
- FaceBuilder analyzes the image
- Landmarks appear automatically on the face (red/blue dots on eyes, nose, mouth, cheekbones)
- The mesh in the viewport updates in real-time
Step 3: Refine the Mesh (Optional)
The auto-detection is usually 80–90% accurate. Refine if needed:
- Click on any landmark and drag to adjust
- Focus on areas that don't match your reference (eyes, cheekbones, jaw outline)
- Zoom in on the viewport to see detail
- Use the Show Edges toggle to see mesh topology
Step 4: Generate Textures
- In the FaceBuilder panel, go to Textures section
- Click Generate from Photo
- FaceBuilder creates:
- Albedo map (diffuse color from the photo)
- Normal map (surface detail)
- Roughness map (skin specularity)
- These textures appear as image data-blocks in your material nodes
Step 5: Export or Render
Option A: Export to use elsewhere
- Right-click the FaceBuilder object
- Export FaceBuilder Mesh
- Choose format: .obj (for other software) or .fbx (with rigging info)
- Save to your project folder
Option B: Render in Blender
- The generated head is already a Blender object with materials
- Set up lighting (3-point lighting recommended for heads)
- Switch to rendered viewport (Z > Rendered)
- Test-render a frame
- Adjust materials if needed (e.g., increase subsurface scattering for more skin translucency)
Advanced: ARKit Blendshapes
FaceBuilder can generate ARKit-compatible blendshapes for the head. These are facial blend shapes used by facial capture systems, AR filters, and game engines.
What Are Blendshapes?
Blendshapes (or morph targets) are alternate mesh deformations for the same topology. For a head, you might have:
- Mouth Open / Mouth Closed
- Eye Blink Left / Eye Blink Right
- Brow Raise / Brow Lower
- Smile / Frown
ARKit defines 52 standard blendshapes that cover full facial expression range.
Enabling ARKit Blendshapes in FaceBuilder
- In the FaceBuilder panel, find Blendshapes section
- Enable Generate ARKit Blendshapes
- FaceBuilder automatically creates all 52 shapes
- You can preview them by sliding the blendshape sliders
Using Blendshapes in Blender
Once generated, blendshapes appear in your head object's Shape Keys (Object Data Properties > Shape Keys):
- In the Properties panel, go to Object Data
- Expand Shape Keys
- Each ARKit shape is listed (eyeBlink_L, eyeWideOpen_L, mouthSmile, etc.)
- Adjust the slider for any shape to preview the deformation
- Use Blender's drivers to animate shapes based on bone rotation or armature constraints
Exporting Blendshapes for Game Engines
If targeting a game engine (Unreal Engine, Unity):
- Export as .fbx with "Shape Keys" enabled in export settings
- Import into the game engine
- The blendshapes become facial animation parameters
- Animate faces via gameplay logic or motion capture data
FaceBuilder Rendering on Cloud Render Farms
When rendering FaceBuilder heads on Super Renders Farm:
Pre-Upload Checklist
-
Finalize the head mesh — Don't leave FaceBuilder modifiers unapplied (convert to mesh first)
-
Bake generated textures — Don't rely on procedural generation at render time
-
Export to standard format — Convert FaceBuilder object to standard Blender mesh:
- Right-click head > Convert to Mesh (or Export > FBX)
- This removes FaceBuilder dependency; the farm only needs Blender's standard rendering
-
Package textures with relative paths — All diffuse/normal/roughness maps must be in your .blend file with relative paths (File > External Data > Pack All)
Rendering Workflow
- Test locally first — Render 1 head frame at your target resolution with your lighting
- Verify texture paths work — Delete the .blend file, reload it, verify textures are still there
- Submit to farm — Upload your .blend file with all textures packed or in relative paths. See our Blender cloud rendering guide for detailed submission steps.
- Get results — EXR multilayer with all passes (diffuse, specular, normal depth, etc.)
Tips for Better FaceBuilder Results
Photography Best Practices
- Lighting: Soft directional light (not harsh shadows). Overcast outdoor lighting is ideal; avoid backlighting
- Angle: 3/4 view or direct frontal. Avoid extreme side angles (profile won't capture full head shape)
- Focus: Sharp focus on eyes and facial features. Blurry photos won't detect landmarks well
- Expression: Neutral expression works well. Smile or frown can confuse landmark detection
Mesh Refinement
- Eye placement: Critical for likeness. Adjust eye corners and eyelids first
- Cheekbones: Controls face width. Adjust if the head looks too round or too thin
- Jaw: Defines face shape (square vs rounded). Subtle adjustments have large impact
- Forehead: Often overlooked. Adjust hairline landmarks if your photo shows a different hairline
Material Tweaks After Texture Generation
FaceBuilder textures are good out of the box, but you can improve:
-
Increase Subsurface Scattering (SSS):
- In Blender's Principled BSDF, increase Subsurface to 0.1–0.3
- This creates skin translucency (ears, lips show light through)
- Render time increases by 10–20%
-
Adjust Roughness:
- Lips: lower roughness (0.1–0.3, shinier)
- Cheeks: mid roughness (0.4–0.6)
- Forehead: mid-high roughness (0.5–0.7, oilier)
-
Add Microdetail:
- Layer a high-frequency normal map over the generated normal for pore detail
- Use Blender's Bump node set to Height for extra realism
Common FaceBuilder Mistakes
Mistake 1: Skipping Auto-Detection
Some artists still manually place landmarks out of habit. Use Detect Landmarks — it's 90%+ accurate and saves 15+ minutes per head.
Mistake 2: Poor Photo Reference
FaceBuilder can't fix a bad photo. If your reference is underexposed, blurry, or extreme angle, results will be poor. Invest 5 minutes in taking a good photo.
Mistake 3: Rendering Without Converting to Mesh
If you render with FaceBuilder object still active (not converted to mesh), the farm won't have FaceBuilder installed. Convert to mesh: Right-click > Convert to Mesh before uploading to render farm.
Mistake 4: Forgetting Texture Paths
Generated textures are image data-blocks in your .blend file. If you move the file, paths might break. Use File > External Data > Pack All before uploading to farm.
FaceBuilder vs Alternatives
vs. 3D Scanning (Photogrammetry)
- FaceBuilder: Fast (10–20 minutes), requires 1 photo
- Photogrammetry: More accurate (200+ photos), slower processing
Use FaceBuilder for quick character art; use photogrammetry for hero closeups requiring extreme detail.
vs. Manual Head Modeling
- FaceBuilder: Fast, consistent proportions, photo-matched
- Manual modeling: More control, but requires anatomy knowledge and 2–4 hours per head
FaceBuilder is faster; manual is more flexible.
vs. AI-Generated Heads (Midjourney, Stable Diffusion)
- FaceBuilder: Based on YOUR photo, fully controllable mesh, production-ready
- AI generative: Unique faces, but copyright/ethics questions; not as customizable
FaceBuilder is better for specific likenesses; AI is better for concept ideation.
FAQ
Q: Can FaceBuilder auto-detect faces in group photos? A: No, FaceBuilder works with single faces. If your photo has multiple people, crop/mask to one face first.
Q: How accurate is automatic landmark detection? A: 85–95% accurate on well-lit portraits. Poorly-lit or extreme-angle photos may require 20–30% manual refinement.
Q: Do I need the generated textures or can I author my own? A: The generated textures are production-ready, but many artists prefer to author custom textures in Substance for more control. FaceBuilder's automatic textures save time for quick turnarounds.
Q: Can I use FaceBuilder heads for commercial projects? A: Yes, FaceBuilder license allows commercial use. No royalties or attribution required (unlike some stock face assets).
Q: How do I render FaceBuilder heads efficiently at scale? A: Convert to mesh, package with textures, and submit to a render farm like Super Renders Farm. See our GPU cloud rendering guide for faster render times on character closeups.
Q: How do I animate a FaceBuilder head? A: Use ARKit blendshapes (via Shape Keys) or rig it with an armature. Blender's Rigify add-on can auto-rig a head generated by FaceBuilder.
Q: Can I export FaceBuilder heads to game engines? A: Yes. Export as FBX with blendshapes enabled. Import into Unity or Unreal Engine. The mesh and shapes transfer correctly.
Q: What's the maximum polycount of a FaceBuilder head? A: Typical output is 30k–50k polygons (medium detail). FaceBuilder can subdivide for higher detail (100k+ polys), but it becomes slow to edit. For game engines, stick with 30k–50k.
Q: Can I use FaceBuilder on a render farm without FaceBuilder installed? A: No. You must convert FaceBuilder objects to standard Blender meshes before uploading to a farm. The farm only needs Blender, not FaceBuilder license.

