OmniFaceRig Fully Automatic Inner-Mouth-Aware Face Rigging
Across Diverse 3D Character Topologies

Chao Wang*†‡, Guangyao Ma*, John Doublestein, Junming Chen, Yiming Lin, Zhaoen Su, Xiaomin Luo, Shiyang Cheng, Jie Shen, Doug Roble, Dilin Wang, Yilei Li, Rakesh Ranjan
Reality Labs, Meta
*Equal contribution   Project lead   Corresponding author
OmniFaceRig teaser (Figure 1)

OmniFaceRig turns a static, surface-only 3D character — with no pre-modeled mouth interior — into an animation-ready, inner-mouth-aware FACS rig with procedurally fitted teeth, gums, and tongue. One automatic pipeline handles humans, humanoids, and both long- and short-muzzled animals, with no manual landmarks, no user templates, and no per-asset setup.

Facial rigging—creating FACS-based blendshapes together with inner-mouth geometry (teeth, gums, and tongue)—remains a major bottleneck in 3D character production. Existing pipelines still require substantial designer effort, especially for manual landmark annotation, per-character template adjustment, and inner-mouth placement.

We present OmniFaceRig, a fully automatic end-to-end pipeline that converts a static surface-only 3D character mesh, with no pre-modeled oral cavity, into an inner-mouth-aware FACS rig with up to 155 blendshapes, procedurally fitted teeth, gums, and tongue, and re-packed UV/texture. OmniFaceRig supports diverse topologies—humans, humanoids, long-muzzled animals (e.g., dogs, wolves, foxes), and short-muzzled animals (e.g., cats, bears, rabbits, tigers)—with no manual landmarks, no user-provided templates, and no per-asset setup.

The pipeline combines hybrid VLM+CV riggability checking, multi-model face parsing, dense keypoint-driven template registration, procedural inner-mouth construction, and collision-aware blendshape transfer. For non-human characters, OmniFaceRig selects topology-specific face and inner-mouth templates and uses collision-aware inner-mouth fitting to reduce teeth-face intersections without exposing users to category-specific tuning.

We also publicly release Omni-Bench, a freely available benchmark dataset of 1,000 biped 3D characters with FACS facial blendshapes and inner-mouth geometry, spanning humans, humanoids, cats, dogs, and other animals. Experiments show high final rigging success on screened Omni-Bench inputs, nearly complete face detection recall from the segmentation ensemble, reliable inner-mouth placement with low penetration, and 20–30 s end-to-end processing time per asset on a single A100 GPU, including data I/O. Together, OmniFaceRig provides an automatic path from static generated characters to animation-ready facial rigs across both human and non-human topologies.

155
FACS blendshapes per rig, including procedurally generated teeth, gums & tongue
1,000
rigged biped characters released in the open Omni-Bench benchmark
20–30s
end-to-end per asset on a single A100 GPU, including data I/O
4
topology categories: human, humanoid, long- & short-muzzled animals

Explore rigged characters in 3D

Each card below shows three columns: the input body image, the original surface-only GLB, and the rigged GLB with FACS animation. Click the ▶ Load button on a card to activate both 3D viewers at once. Inside each viewer, drag the splitter to toggle texture (left half) vs. solid grey (right half); camera and split position are shared across both viewers (left-drag rotate, scroll zoom, shift+drag pan). After 2.5 s of idle time the model slowly auto-rotates.

Human

12 assets
older_surfer_dude_cleaned
Human
Foreboard Image
Original Asset
Rigged + FACS
100
Human
Foreboard Image
Original Asset
Rigged + FACS
253
Human
Foreboard Image
Original Asset
Rigged + FACS
013_cleaned
Human
Foreboard Image
Original Asset
Rigged + FACS
007
Human
Foreboard Image
Original Asset
Rigged + FACS
010
Human
Foreboard Image
Original Asset
Rigged + FACS
019
Human
Foreboard Image
Original Asset
Rigged + FACS
045
Human
Foreboard Image
Original Asset
Rigged + FACS
079
Human
Foreboard Image
Original Asset
Rigged + FACS
074
Human
Foreboard Image
Original Asset
Rigged + FACS
075
Human
Foreboard Image
Original Asset
Rigged + FACS
076
Human
Foreboard Image
Original Asset
Rigged + FACS
older_surfer_dude_cleaned
100
253
013_cleaned
007
010
019
045
079
074
075
076

Humanoid

12 assets
moss_covered_dryad_android
Humanoid
Foreboard Image
Original Asset
Rigged + FACS
009
Humanoid
Foreboard Image
Original Asset
Rigged + FACS
137
Humanoid
Foreboard Image
Original Asset
Rigged + FACS
202
Humanoid
Foreboard Image
Original Asset
Rigged + FACS
008
Humanoid
Foreboard Image
Original Asset
Rigged + FACS
015
Humanoid
Foreboard Image
Original Asset
Rigged + FACS
017
Humanoid
Foreboard Image
Original Asset
Rigged + FACS
160
Humanoid
Foreboard Image
Original Asset
Rigged + FACS
200
Humanoid
Foreboard Image
Original Asset
Rigged + FACS
204
Humanoid
Foreboard Image
Original Asset
Rigged + FACS
205
Humanoid
Foreboard Image
Original Asset
Rigged + FACS
213
Humanoid
Foreboard Image
Original Asset
Rigged + FACS
moss_covered_dryad_android
009
137
202
008
015
017
160
200
204
205
213

Long-muzzled

13 assets
196
Long-muzzled
Foreboard Image
Original Asset
Rigged + FACS
180
Long-muzzled
Foreboard Image
Original Asset
Rigged + FACS
185
Long-muzzled
Foreboard Image
Original Asset
Rigged + FACS
111_dog
Long-muzzled
Foreboard Image
Original Asset
Rigged + FACS
000_dog
Long-muzzled
Foreboard Image
Original Asset
Rigged + FACS
057_dog
Long-muzzled
Foreboard Image
Original Asset
Rigged + FACS
065
Long-muzzled
Foreboard Image
Original Asset
Rigged + FACS
059
Long-muzzled
Foreboard Image
Original Asset
Rigged + FACS
063
Long-muzzled
Foreboard Image
Original Asset
Rigged + FACS
179
Long-muzzled
Foreboard Image
Original Asset
Rigged + FACS
262
Long-muzzled
Foreboard Image
Original Asset
Rigged + FACS
279
Long-muzzled
Foreboard Image
Original Asset
Rigged + FACS
267
Long-muzzled
Foreboard Image
Original Asset
Rigged + FACS
196
180
185
111_dog
000_dog
057_dog
065
059
063
179
262
279
267

Short-muzzled

13 assets
bear_01
Short-muzzled
Foreboard Image
Original Asset
Rigged + FACS
053
Short-muzzled
Foreboard Image
Original Asset
Rigged + FACS
264
Short-muzzled
Foreboard Image
Original Asset
Rigged + FACS
190
Short-muzzled
Foreboard Image
Original Asset
Rigged + FACS
099
Short-muzzled
Foreboard Image
Original Asset
Rigged + FACS
122_cat
Short-muzzled
Foreboard Image
Original Asset
Rigged + FACS
159_cat
Short-muzzled
Foreboard Image
Original Asset
Rigged + FACS
193_cat
Short-muzzled
Foreboard Image
Original Asset
Rigged + FACS
136
Short-muzzled
Foreboard Image
Original Asset
Rigged + FACS
061
Short-muzzled
Foreboard Image
Original Asset
Rigged + FACS
161
Short-muzzled
Foreboard Image
Original Asset
Rigged + FACS
176
Short-muzzled
Foreboard Image
Original Asset
Rigged + FACS
272
Short-muzzled
Foreboard Image
Original Asset
Rigged + FACS
bear_01
053
264
190
099
122_cat
159_cat
193_cat
136
061
161
176
272

Qualitative results across diverse topologies

99.0%
Final rigging success (human + humanoid)
0.85 mm
Mean vertex error (MAE)
0.05%
Inner-mouth penetration (3.4× better)
~100%
Face detection recall (ensemble)
87.3%
Segmentation mIoU (vs 62.1%)

Two-stage architecture

OmniFaceRig pipeline architecture

Stage 1 · Face Template Fitting

Riggability assessment, multi-model face detection & segmentation, landmark extraction, and rigid + non-rigid template registration produce a fitted face mesh.

Stage 2 · Blendshape Rig Construction

Face mesh fusion, teeth registration & texture baking, and FACS blendshape transfer produce the final inner-mouth-aware FACS rig.

A new benchmark for face auto-rigging

Omni-Bench dataset overview: diverse rigged 3D characters spanning humans, humanoids, cats, dogs, and other animals.
Omni-Bench overview — 1,000 biped 3D characters with FACS facial blendshapes and inner-mouth geometry, spanning realistic humans, humanoid characters, cats, dogs, and a long tail of other common animals.

To facilitate future research in automatic facial rigging, we release Omni-Bench, to our knowledge the first open-source benchmark of biped 3D characters with FACS facial blendshapes and inner-mouth geometry (teeth, gums, and tongue). Omni-Bench contains 1,000 rigged biped 3D characters, split into 500 human and humanoid characters and 500 animals (150 cats across 10 breeds, 150 dogs across 10 breeds, and 200 other common animals including bears, tigers, lions, foxes, wolves, rabbits, and deer). All assets are released in T-pose with up to 12 appearance variations and full generation-pipeline metadata (text prompt, intermediate 2D reference image, final 3D mesh).

Omni-Bench stands out in three critical ways. First, it uniquely supports both human and animal characters within a unified rigging framework. Second, every model is equipped with a complete set of auto-generated FACS blendshapes (up to 155 shapes) that explicitly include teeth, gums, and tongue — a feature absent in nearly all existing large-scale datasets. Third, each asset includes the complete generation pipeline data (text + reference image + 3D mesh), making it a valuable resource for text-to-3D generation and multimodal research.

Comparison with existing 3D facial / character datasets — Omni-Bench is the only dataset providing FACS blendshapes with inner-mouth geometry for both humans and animals.
Dataset Year Total Models Species FACS Blendshapes Inner Mouth Full Character Text + 2D Image
BFM2009200Human
CoMA2018144Human
VOCASET201912 (4D)Human
FaceScape202016,940Human
ICT FaceKit2020ParametricHuman
Multiface202213Human
RaBit20231,500Human & Cartoon
Anymate2025230,000Human & Animal
Omni-Bench (Ours)20261,000Human & Animal

If you find this work useful, please cite our paper.

@article{wang2026omnifacerig,
  title={OmniFaceRig: Fully Automatic Inner-Mouth-Aware Face Rigging Across Diverse 3D Character Topologies},
  author={Wang, Chao and Ma, Guangyao and Doublestein, John and Chen, Junming and Lin, Yiming and Su, Zhaoen and Luo, Xiaomin and Cheng, Shiyang and Shen, Jie and Roble, Doug and Wang, Dilin and Li, Yilei and Ranjan, Rakesh},
  journal={arXiv preprint arXiv:2606.08043},
  year={2026}
}