Efficient 3D Gaussian Head Avatars for Edge Devices

Umar Farooq; Jean-Yves Guillemaut; Adrian Hilton; Marco Volino

cs.CVOct 7, 2026

Efficient 3D Gaussian Head Avatars for Edge Devices

Authors: Umar Farooq, Jean-Yves Guillemaut, Adrian Hilton, Marco Volino

Organizations: CVSSP, University of Surrey

Abstract

Generative 3D Gaussian head avatars provide high-quality, efficient rendering, but synthesising the Gaussian representation remains computationally expensive, limiting deployment on resource-constrained and edge devices. We introduce an efficient generator architecture for unconditional 3D Gaussian head synthesis, based on a parameter-efficient synthesis block and depth-wise separable convolutions while retaining style-based conditioning. Our architecture reduces generator complexity without requiring model compression or quantisation. Compared with the baseline model, our approach reduces FLOPs by 94%, parameter count by 70%, and model size by 81%, while maintaining competitive generation quality. We further demonstrate practical CPU inference and browser-based execution on mobile devices using ONNX Runtime, enabling 3D Gaussian avatar synthesis without dedicated GPU hardware or application-specific software. In addition to conventional image-quality metrics, we evaluate multi-view consistency, training cost, and deployment performance. Code, trained models, and evaluation tools will be released publicly.

Figures & tables

Figure 1 : Overview of our efficient 3D Gaussian head-avatar synthesis pipeline and downstream applications. Our model can generate a novel 3D Gaussian avatar from a random latent code and it also supports subsequent animation and latent manipulation.

Figure 2 : Overview of our 3D Gaussian head-avatar generation pipeline.

Figure 3 : Proposed architecture of the modulated convolution block with depth-wise separable convolutions. The modulation is applied to the instance normalised input feature map and the demodulation to the output of the point-wise convolution.

	Generation Quality and Compute Cost					Generation		Animation
Method	FID $\downarrow$	FLOP (G) $\downarrow$	Params (M) $\downarrow$	Train $\downarrow$	Size $\downarrow$	CPU $\downarrow$	Pixel 10 $\downarrow$	Pixel 10 $\downarrow$
				(hrs)	(MB)	(ms)	(ms)	(ms)
AGORA ( $512^{2}$ ) [ 13 ]	3.17	212	33	50	241	323	2381	32
Ours ( $512^{2}$ )	4.15	13	10	29	46	148	453	32

Table 1: Comparison of generation quality at

512^{2}

rasterisation resolution, together with computational cost, generation latency, and animation latency at 25M training images. FID is computed against

512^{2}

reference statistics. The generator backbone,

256^{2}

UV attribute map, deformation plane, and Gaussian count are unchanged; consequently, the remaining generator and deployment measurements are the same as for

256^{2}

rasterisation.

Figure 4 : A pixel level comparison of our proposed model’s inference in PyTorch against an onnx-runtime implementation. Please note the pixel differences are enhanced by $10\times$ for better visibility.

Figure 5 : Our animation model can drive avatar expression based on FLAME parameters. In each row the expression is interpolated from the outermost images.

Method	FLOPs	FID	$\text{FID}_{\text{3D}}$ $\downarrow$	$\text{PSNR}_{\text{MV}}$ $\uparrow$	$\text{SSIM}_{\text{MV}}$ $\uparrow$	Train Time $\downarrow$	VRAM $\downarrow$
GGHEAD	66 G	9.6	11.57	32.60	0.868	10.51 hrs.	21.7 GB
Ours-Static	2.7 G	12.0	16.40	34.01	0.880	11.59 hrs.	16.0 GB

Table 2 : We show 3D consistency, multi-view quality and efficiency metrics for our method on a smaller static model at 5M images and

256^{2}

resolution.

Configuration	FID $\downarrow$	Params $\downarrow$	FLOP $\downarrow$	CPU $\downarrow$
		(M)	(G)	(ms)
(a) Depth-wise separable convolutions
GGHead [ 25 ]	9.6	28.0	–	–
+ separable synthesis convs	9.6	8.0	–	–
+ separable output convs (C1)	9.8	8.0	–	354
(b) Channel reduction applied to C1

Table 3 : Ablations shown at 5M images and

256^{2}

resolution. (a) shows that depth-wise convolutions are capable of modelling the data distribution and do not lose significant quality. Dashes mark configurations we did not profile.

Appendix figures & tables5 assets

Supplementary material from the paper’s appendix.

Appendix

Component	Params (M)	GFLOP	Size (MB)	Time (ms)
Synthesis	7.74	7.42	31.51	134.8
Deformation	2.13	5.32	14.63	129.4
Full	10.67	12.74	46.12	246.8

Appendix

Table 4 : Cost breakdown of the generator used for

512^{2}

rendering. The branches are profiled individually while the full model is timed as a single end-to-end run of the complete pipeline, so the branch latencies do not add up directly to the full-model latency. All reported times are medians on a single CPU core; the full model takes

171.3

ms when four threads are used. Parameter, FLOP and size figures are identical to those of the

256^{2}

model, since the UV attribute backbone and the deformation plane both remain at

256^{2}

and only the rasterisation resolution changes.

Method	FID $\downarrow$
AGORA [ 13 ]	3.8
Ours	4.3

Appendix

Table 5 : Generation quality at

256^{2}

resolution. FID is computed on the PyTorch checkpoints against reference statistics at the matching resolution, following the protocol of Section A.1 .

Figure 7 : Qualitative results from our proposed model and the baseline GGHead model at $256^{2}$ resolution. Please note these are trained for 5M images as outlined in Section 4 .

Figure 8 : Latent space manipulation. Traversing individual directions in the intermediate latent space of our generator produces smooth, largely disentangled changes to a single factor of variation while the remaining identity is preserved, showing that the depth-wise separable synthesis blocks retain the editable style space of the baseline architecture.

Figure 9 : Resolution sensitivity of each attribute group. Each curve low-passes one group to the resolution on the horizontal axis and bilinearly restores it, on frozen weights, leaving the other five groups untouched; the dotted line is the unmodified model.

Explore similar work

AGORA: Adversarial Generation Of Real-time Animatable 3D Gaussian Head Avatars
Dec 6, 2025Ramazan Fazylov, Sergey Zagoruyko, Aleksandr Parkin +2Animatable 3D Human Avatars Avatars
Mohamed bin Zayed University of Artificial Intelligence, Abu Dhabi, UAE · Polynome AI, Dubai, UAE · MTS AI, Moscow, Russia
Multi-view Consistent 3D Gaussian Head Avatars 'without' Multi-view Generation
May 24, 2026Aviral Chharia, Fernando De la TorreAnimatable 3D Human Avatars 3D Gaussian
Carnegie Mellon University
One-Shot Feed-Forward 360 $^{\circ}$ Animatable Avatar via Inpainted UV-Space Gaussian Modeling
Jan 19, 2026Shuling Zhao, Dan XuAnimatable 3D Human Avatars Avatars
The Hong Kong University of Science and Technology · Zeekr Automobile R&D Co., Ltd.

Dec 6, 2025, cs.CVJ/K move · Enter open · S save

AGORA: Adversarial Generation Of Real-time Animatable 3D Gaussian Head Avatars

Ramazan Fazylov, Sergey Zagoruyko, Aleksandr Parkin, Stamatis Lefkimmiatis+1

Mohamed bin Zayed University of Artificial Intelligence, Abu Dhabi, UAE · Polynome AI, Dubai, UAE · MTS AI, Moscow, Russia

Animatable 3D Human Avatars Avatars Neural Rendering 3D Generation Generative Adversarial Network Adversarial Examples

Open paper AskPDF arXiv

The generation of high-fidelity, animatable 3D human avatars remains a core challenge in computer graphics and vision, with applications in VR, telepresence, and entertainment. Existing approaches based on implicit representations like NeRFs suffer from slow rendering and dynamic inconsistencies, while 3D Gaussian Splatting (3DGS) methods are typically limited to static head generation, lacking dynamic control. We bridge this gap by introducing AGORA, a novel framework that extends 3DGS within a generative adversarial network to produce animatable avatars. Our key contribution is a lightweight, FLAME-conditioned deformation branch that predicts per-Gaussian residuals, enabling identity-preserving, fine-grained expression control while allowing real-time inference. Identity is further preserved through spatial shape conditioning of the identity branch, and expression fidelity is enforced via a dual-discriminator training scheme leveraging synthetic renderings of the parametric mesh. AGORA generates avatars that are not only visually realistic but also precisely controllable. Quantitatively, we outperform state-of-the-art NeRF-based methods on expression accuracy while rendering at 250 FPS on a single GPU and, notably, at

\sim

9 FPS under CPU-only inference -- to our knowledge the first demonstration of CPU-only animatable 3DGS avatar synthesis. This work represents a significant step toward practical, high-performance digital humans. Project website: https://ramazan793.github.io/AGORA/

Abstract

Explore similar work

AGORA: Adversarial Generation Of Real-time Animatable 3D Gaussian Head Avatars

Multi-view Consistent 3D Gaussian Head Avatars 'without' Multi-view Generation

One-Shot Feed-Forward 360∘^{\circ}∘ Animatable Avatar via Inpainted UV-Space Gaussian Modeling

One-Shot Feed-Forward 360 $^{\circ}$ Animatable Avatar via Inpainted UV-Space Gaussian Modeling