ExAvatar is a new type of full-body dynamic expression model that combines the full-body driving capabilities of SMPL-X with the strong appearance modeling of 3DGS. It creates models through simple smartphone scanning and supports animation rendering for various poses and expressions. Its hybrid representation method enhances the naturalness of facial expressions, reduces artifacts for new expressions and poses, and ensures complete compatibility with the facial expression space of SMPL-X.

Stable-Hair is an innovative hairstyle transfer technique that robustly transfers a diverse range of real-world hairstyles onto user-provided facial images for virtual try-ons. This method excels at handling complex and varied hairstyles, maintaining the original identity content and structure while achieving highly detailed and high-fidelity transfer effects.

AniTalker is an innovative framework that can generate realistic dialogue facial animations from a single portrait. It enhances expressive motion capture through two self-supervised learning strategies, and develops an identity encoder through metric learning, effectively reducing the need for labeled data. AniTalker not only creates detailed and realistic facial expressions but also emphasizes its potential in real-world applications for producing dynamic avatars.

ugly-avatar is an open-source avatar generator primarily targeting individuals and small websites, offering a fun and unique avatar generation service. Developed with Vue and JavaScript, it supports customizable configurations, making it easy to integrate and use. The project follows the Attribution-NonCommercial 4.0 International License, meaning it cannot be used for commercial purposes.

BakedAvatar is a novel representation for real-time neural avatar synthesis, deployable in a standard polygonal rasterization pipeline. The method extracts deformable multi-layer meshes from learned head iso-surfaces and computes appearance variations related to expression, pose, and viewpoint that can be baked into static textures, enabling real-time 4D avatar synthesis. We propose a three-stage neural avatar synthesis pipeline, including learning continuous deformation, manifold, and radiance fields, extracting layered meshes and textures, and refining texture details via differential rasterization. Experiments demonstrate that our representation achieves comparable results to other state-of-the-art methods while significantly reducing inference time. We further showcase diverse avatar synthesis results generated from single-eye video, including view synthesis, facial re-enactment, expression editing, and pose editing, all at interactive frame rates.

MagicAvatar is a multi-modal framework that can convert various input modes (text, video, and audio) into motion signals, thereby generating/animating avatars. It can create avatars through simple text prompts and also create avatars that follow given movements based on provided source videos. It can also animate avatars with specific themes. MagicAvatar's strength lies in its ability to combine multiple input modes to generate high-quality avatars and animations.

PicLooks is a website that offers free AI-generated realistic style avatars, aimed at providing designers and marketers with high-quality avatar images for design prototypes, marketing materials, and more. These avatar images are entirely generated by AI, eliminating copyright concerns associated with traditional stock photo websites. Users can quickly find avatars that meet their needs, saving substantial time on searches, and they can be used for commercial purposes with no copyright worries.

ExAvatar is a new type of full-body dynamic expression model that combines the full-body driving capabilities of SMPL-X with the strong appearance modeling of 3DGS. It creates models through simple smartphone scanning and supports animation rendering for various poses and expressions. Its hybrid representation method enhances the naturalness of facial expressions, reduces artifacts for new expressions and poses, and ensures complete compatibility with the facial expression space of SMPL-X.

Stable-Hair is an innovative hairstyle transfer technique that robustly transfers a diverse range of real-world hairstyles onto user-provided facial images for virtual try-ons. This method excels at handling complex and varied hairstyles, maintaining the original identity content and structure while achieving highly detailed and high-fidelity transfer effects.

AniTalker is an innovative framework that can generate realistic dialogue facial animations from a single portrait. It enhances expressive motion capture through two self-supervised learning strategies, and develops an identity encoder through metric learning, effectively reducing the need for labeled data. AniTalker not only creates detailed and realistic facial expressions but also emphasizes its potential in real-world applications for producing dynamic avatars.

ugly-avatar is an open-source avatar generator primarily targeting individuals and small websites, offering a fun and unique avatar generation service. Developed with Vue and JavaScript, it supports customizable configurations, making it easy to integrate and use. The project follows the Attribution-NonCommercial 4.0 International License, meaning it cannot be used for commercial purposes.

Avatar is a simple web application that can help users randomly generate avatars in different styles. The application offers various avatar forms, including half-body, sitting, and standing poses, which users can choose according to their preferences. The application uses generative adversarial network (GAN) technology to output high-quality, detail-rich avatar images. Compared to traditional avatar creation methods, it is more convenient and quick, and it can produce more personalized avatars. Overall, it is a practical and interesting web application suitable for various scenarios requiring avatars.

UltrAvatar is a realistic and movable 3D avatar generation model designed to bridge the gap between virtual and real-world experiences. It utilizes Score Distillation Sampling (SDS) loss, a differentiable renderer, and text conditioning to guide the diffusion model in generating 3D avatars. Compared to existing works, UltrAvatar presents a novel approach to 3D avatar generation by enhancing geometric fidelity and offering superior physical rendering texture quality. It employs a diffusion color extraction model and a realism-guided texture diffusion model to remove unnecessary lighting effects, presenting genuine diffusion colors, enabling the generated avatars to render realistically under various lighting conditions. Our experiments have proven the effectiveness and robustness of this method, significantly outperforming existing state-of-the-art approaches.

BakedAvatar is a novel representation for real-time neural avatar synthesis, deployable in a standard polygonal rasterization pipeline. The method extracts deformable multi-layer meshes from learned head iso-surfaces and computes appearance variations related to expression, pose, and viewpoint that can be baked into static textures, enabling real-time 4D avatar synthesis. We propose a three-stage neural avatar synthesis pipeline, including learning continuous deformation, manifold, and radiance fields, extracting layered meshes and textures, and refining texture details via differential rasterization. Experiments demonstrate that our representation achieves comparable results to other state-of-the-art methods while significantly reducing inference time. We further showcase diverse avatar synthesis results generated from single-eye video, including view synthesis, facial re-enactment, expression editing, and pose editing, all at interactive frame rates.