Publications

Selected work in 3D computer vision and 3D synthesis. Click a paper for its official abstract and links.


Papers

GaussianTalker teaserACM Multimedia 2024GaussianTalker: Real-Time High-Fidelity Talking Head Synthesis with Audio-Driven 3D Gaussian SplattingEncodes 3D Gaussian attributes into a shared implicit feature that speech audio can steer, rendering pose-controllable talking heads at up to 120 FPS.#3D#gaussian-splatting#talking-head#real-time Talk3D teaserICCV 2025 WorkshopTalk3D: High-Fidelity Talking Portrait Synthesis via Personalized 3D Generative PriorDrives a personalized 3D generative prior with an audio-guided attention U-Net, so facial geometry stays plausible even under extreme head poses.#3D#talking-portrait#NeRF#generative-prior WACV 20233D GAN Inversion with Pose OptimizationInfers camera viewpoint and latent code together, so a single image can be reconstructed and edited with multi-view consistency.#3D#GAN-inversion#NeRF#editing AE-NeRF teaserarXiv 2023AE-NeRF: Auto-Encoding Neural Radiance Fields for 3D-Aware Object ManipulationAn auto-encoder that disentangles shape, appearance, and camera pose from a single image, then renders them back through a generative NeRF.#3D#NeRF#auto-encoder#disentanglement

Back to top

Copyright © 2017-2026 Patrick Marsceill and Just the Docs contributors. Distributed by an MIT license.