Character Generation¶
Characters & Avatars
Open Source¶
-
DreamCharacter-1: From 3D Generative Foundation Models to Product-Ready Character Generation Discovered by AI
2026
A lightweight post-adaptation framework from ByteDance that calibrates pretrained 3D foundation models toward high-fidelity, production-ready 3D character generation. It incorporates geometry and texture post-training along with inference acceleration to produce animation-ready 3D characters from a single reference image.
3d-characterbytedancefoundation-model -
EvoTale: Continual Character Customization for Expanding Story Worlds Discovered by AI
2026
A continual character customization framework designed for expanding story worlds, enabling the integration of new user-specified characters while preserving recurring character identities. It addresses identity conflicts and customization difficulties in multi-character narrative scenes.
character-customizationidentity-preservingstory-visualization -
OmniChar Discovered by AI
2026
An open-source AI character studio for portable identity, allowing creators to build multi-model consistent characters, train LoRAs, and generate consistent character sheets locally on their own GPUs.
character-studioconsistent-characterlora -
Rotate Your Character: Revisiting Video Diffusion Models for High-Quality 3D Character Generation Discovered by AI
2026
An advanced image-to-video diffusion framework (Rotate your Character Model, or RCM) from Tencent Hunyuan and HKU designed for high-quality novel view synthesis and 3D character generation. It transfers characters from any complex pose into a canonical pose, enabling consistent 360-degree orbital video generation at high resolution.
3d-characternovel-view-synthesistencent -
SDO: Subspace Deconflicting Operator for Multi-Adapter Composition Discovered by AI
2026
A parameter-space subspace deconflicting operator (SDO) designed for multi-adapter composition, preventing identity mixing and attribute leakage when generating multiple distinct characters in a single scene.
adapter-compositiondiffusionmulti-character -
StdGEN++: A Comprehensive System for Semantic-Decomposed 3D Character Generation Discovered by AI
2026
A comprehensive system built upon a Dual-branch Semantic-aware Large Reconstruction Model (Dual-Branch S-LRM) to generate high-fidelity, semantically decomposed 3D characters from diverse inputs.
3d-charactercharacter-generationsemantic-decomposition -
TGRHuman: Text-Guided Realistic 3D Human Generation via Diffusion Renderer Discovered by AI
2026
A novel framework for text-guided realistic 3D human generation that decouples geometry and texture generation. It uses a high-resolution generative module for multi-view normals and a diffusion renderer for dense, spatially consistent texture synthesis.
3d-humandiffusion-renderertext-to-3d -
CartoonAlive Discovered by AI
2025
A method for generating high-quality, expressive Live2D digital human models from a single portrait image within seconds.
character-generationlive2dportrait -
CharaConsist Discovered by AI
2025
A consistent character generation method tailored for text-to-image Diffusion Transformer (DiT) models, maintaining fine-grained identity details.
character-generationconsistent-characterdiffusion-transformer -
CharacterFactory Discovered by AI
2025
A framework that integrates GANs with diffusion models to sample and generate highly consistent characters across different scenes and poses.
character-generationconsistent-characterdiffusion -
Diffuman4D: 4D Consistent Human View Synthesis from Sparse-View Videos with Spatio-Temporal Diffusion Models Discovered by AI
2025
A spatio-temporal diffusion model designed to synthesize high-resolution, 4D-consistent human views and animatable avatars from sparse-view video inputs.
4d-avatardiffusionview-synthesis -
HumanDreamer-X Discovered by AI
2025
A framework for photorealistic single-image human avatar reconstruction using Gaussian Splatting. It integrates multi-view human generation and reconstruction into a unified pipeline, utilizing a Gaussian restoration process to enhance geometric consistency and visual fidelity.
3d-avatargaussian-splattingreconstruction -
InstantCharacter Discovered by AI
2025
A tuning-free, scalable framework for character customization built upon a foundation diffusion transformer. It enables character-preserving image generation from a single reference image while maintaining high textual controllability.
character-preservationdiffusion-transformertuning-free -
Make-A-Character 2 Discovered by AI
2025
An advanced system for generating high-quality, animatable 3D characters from single portrait photographs. It improves upon its predecessor by incorporating illumination correction and enhanced head-to-body integration, making it ideal for game development and digital human applications.
3d-characteranimatabletext-to-3d -
MHR: Momentum Human Rig Discovered by AI
2025
An anatomically-inspired parametric full-body digital human model developed by Meta. It combines a decoupled skeleton/shape paradigm with a modern, corrective-driven rig and pose corrective system, making it highly suitable for animation, graphics, and AR/VR applications.
digital-humanparametric-modelrigging -
PERSONA: Personalized Whole-Body 3D Avatar with Pose-Driven Deformations from a Single Image Discovered by AI
2025
A personalized whole-body animatable avatar creation system that supports pose-driven deformations (such as non-rigid cloth deformations) from a single image. It integrates 3D-based identity representation with diffusion-based deformation learning to achieve high-fidelity, animatable 3D avatars.
3d-avatarpose-deformationsingle-image -
PFAvatar: Pose-Fusion 3D Personalized Avatar Reconstruction from Real-World Outfit-of-the-Day Photos Discovered by AI
2025
A method that reconstructs high-quality 3D avatars from real-world Outfit-of-the-Day (OOTD) photos featuring diverse poses, occlusions, and complex backgrounds. It uses a pose-aware diffusion model and NeRF-based representation to preserve high-frequency textures and support downstream applications like virtual try-on and animation.
3d-avatarnerfootd -
ReMix Discovered by AI
2025
A unified framework for consistent character generation and editing, leveraging large-scale text-to-image diffusion models like FLUX.1 to produce high-fidelity outputs with consistent identities.
character-consistencydiffusioneditingflux -
SVAD: From Single Image to 3D Avatar via Synthetic Data Generation with Video Diffusion and Data Augmentation Discovered by AI
2025
SVAD is a pipeline that generates high-fidelity animatable 3D Gaussian avatars from a single image by leveraging video diffusion models for synthetic data generation. It incorporates identity preservation and image restoration modules to maintain visual consistency across novel poses and viewpoints.
3dgsavatar-generationvideo-diffusion -
Textoon Discovered by AI
2025
An innovative method for generating diverse 2D cartoon characters in the Live2D format from text descriptions. It leverages large language and vision models to quickly create interactive, layered 2D characters suitable for animation.
cartooncharacter-designlive2dtext-to-image -
Visual Persona Discovered by AI
2025
A foundation model for text-to-image full-body human customization that generates diverse images of an individual guided by text descriptions. It captures detailed full-body appearance, including face, clothing, body shape, and accessories, while maintaining consistency.
full-bodyhuman-customizationtext-to-image -
AvatarPopUp: Instant 3D Human Avatar Generation using Image Diffusion Models Discovered by AI
2024
A method for fast, high-quality 3D human avatar generation from different input modalities, such as images and text prompts, with control over the generated pose and shape. It decouples generation from 3D modeling by using diffusion-based image generation networks specialized for each task, followed by a 3D lifting network.
3d-avatarcharacter-generationdiffusion -
CanonicalFusion: Generating Drivable 3D Human Avatars from Multiple Images Discovered by AI
2024
A framework for reconstructing animatable 3D human avatars from multiple images by integrating individual reconstruction results into a canonical space using Linear Blend Skinning (LBS) weight maps and forward skinning.
3d-avatarlbsreconstruction -
Character-Adapter: Prompt-Guided Region Control for High-Fidelity Character Customization Discovered by AI
2024
A plug-and-play framework for generating images with consistent characters. It uses prompt-guided segmentation to extract fine-grained regional features of reference characters and dynamic region-level adapters to prevent concept confusion.
adaptercharacter-generationconsistent-characterimage-generation -
CharacterFactory Discovered by AI
2024
A framework that enables sampling and generating new characters with consistent identities in the latent space of GANs for diffusion models. It allows users to generate consistent character images in various poses, outfits, and backgrounds without requiring test-time fine-tuning.
consistent-characterdiffusionganidentity-preservation -
CharacterGen Discovered by AI
2024
A framework for efficiently generating high-quality 3D characters from single images. It introduces an image-conditioned multi-view diffusion model that calibrates input poses to a canonical form to address pose ambiguity and self-occlusion.
3d-characterdiffusionmulti-viewreconstruction -
CharNeRF: 3D Character Generation from Concept Art Discovered by AI
2024
A framework designed to generate 3D character models from consistent turnaround concept art, which is a standard input in the 3D modeling industry. It utilizes Neural Radiance Fields (NeRF) to create volumetric representations of characters, streamlining the 3D modeling pipeline for games and virtual reality.
3d-characterconcept-artnerf -
ConsistentID: Portrait Generation with Multimodal Fine-Grained Identity Preserving Discovered by AI
2024
A multimodal identity-preserving portrait generation framework that integrates face parsing and facial identity features into diffusion models. It enables fine-grained facial customization and high-fidelity identity preservation from a single reference image.
character-generationdiffusionidentity-preservingportrait-generationGitHub · arXiv · Hugging Face
-
DrawingSpinUp Discovered by AI
2024
A framework designed to generate high-quality 3D animations from a single character drawing by reconstructing 3D character meshes and transferring motion.
3d-characteranimationcharacter-generation -
En3D Discovered by AI
2024
An enhanced generative scheme for sculpting high-quality 3D human avatars from 2D synthetic data without relying on pre-existing 3D assets.
3d-humanavatar-generationzero-shot -
Gen-3Diffusion: Realistic Image-to-3D Generation via 2D & 3D Diffusion Synergy Discovered by AI
2024
A template-free framework that generates realistic 3D objects and clothed human avatars from a single RGB image by leveraging the synergy between 2D multi-view diffusion models and 3D diffusion models.
3d-avatardiffusionimage-to-3dGitHub · arXiv · Hugging Face · Website
-
Human-LRM NEW Discovered by AI
2024
Human-LRM is a feed-forward large reconstruction model that combines diffusion-guided priors and transformer architectures to generate 3D human digital characters from a single image without template constraints.
3d-humancharacter-generationlrmsingle-view-3d -
Motion Avatar: Generate Human and Animal Avatars with Arbitrary Motion Discovered by AI
2024
An agent-based framework that integrates 3D avatar generation with motion synthesis, allowing the creation of animatable human and animal characters from text queries.
3d-avatarcharacter-generationmotion-synthesis -
PSHuman Discovered by AI
2024
A framework for photorealistic and identity-preserving 3D human reconstruction from a single monocular image. It uses cross-scale multiview diffusion to generate consistent views and explicit remeshing to produce high-fidelity 3D human meshes.
3d-humandiffusionmeshreconstruction -
RodinHD Discovered by AI
2024
A high-fidelity 3D avatar generation framework that produces detailed 3D avatars from a single portrait image using diffusion models and a novel weight consolidation regularization.
3d-avatardiffusionreconstruction -
StdGEN Discovered by AI
2024
An innovative pipeline for generating high-quality, semantically decomposed 3D characters from single images. It utilizes a semantic field to reconstruct characters into layered, part-level components (such as body and clothing) for flexible customization.
3d-characteranime-charactersemantic-decomposition -
StoryMaker Discovered by AI
2024
A personalization solution that preserves the consistency of faces, clothing, hairstyles, and bodies for multiple characters in text-to-image generation.
character-generationconsistent-charactertext-to-image -
StructLDM: Structured Latent Diffusion for 3D Human Generation Discovered by AI
2024
StructLDM is a structured latent diffusion model for 3D human generation from 2D images, utilizing a semantic structured latent space defined on a statistical human body template. It enables controllable 3D human generation and editing, including pose, view, shape control, and part-aware clothing editing.
3d-humanavatar-generationdiffusion -
GETAvatar Discovered by AI
2023
A generative model for creating animatable 3D human avatars with high-quality textures and geometries represented as textured meshes.
3d-avataranimatablemesh-generation -
HumanGaussian Discovered by AI
2023
A text-driven 3D human generation framework that utilizes 3D Gaussian Splatting to generate high-quality, animatable 3D humans with fine-grained geometry and realistic appearance.
3d-humanavatargaussian-splattingtext-to-3d -
Make-A-Character Discovered by AI
2023
A user-friendly framework designed to create lifelike 3D avatars and characters from text descriptions. It leverages large language and vision models for textual intention understanding and generates animatable 3D characters within minutes.
3d-characteranimatableavatartext-to-3d -
SiTH: Single-view Textured Human Reconstruction with Image-Conditioned Diffusion Discovered by AI
2023
SiTH is a single-view 3D human reconstruction pipeline that integrates an image-conditioned diffusion model with implicit neural fields. It decomposes the reconstruction problem into back-view hallucination and mesh reconstruction, producing fully textured 3D human meshes in minutes.
diffusionhuman-reconstructionmesh-generation -
MPFB2 Discovered by AI
2022
A free and open-source human character generator and editor for Blender, offering parametric modeling of body features, procedural skin materials, and Rigify integration.
3dblendercharacterparametric -
CharMorph Discovered by AI
2020
An open-source 3D character creation tool for Blender that uses base meshes and morphs to allow easy creation and customization of humanoid and non-humanoid characters.
3dblendercharactermorph
Proprietary / Industry Tools¶
-
Character Creator (CC4) Discovered by AI
2022
A premier 3D character creation and customization software widely used in game development and animation. It allows artists to generate, morph, and dress realistic or stylized 3D characters, complete with full-body and facial rigs compatible with Unreal Engine, Unity, Blender, and Maya.
3d-characterrealisticriggingstylized -
Character Creator 4 Discovered by AI
2022
Reallusion's professional 3D character creation and animation rigging software designed to easily generate, import, and customize stylized or realistic character assets.
3d-humancharacter-creatorriggingstylized -
VRoid Studio Discovered by AI
2021
A popular 3D character creation software developed by Pixiv that allows users to design and customize humanoid anime-style avatars. It features intuitive brush tools for hair modeling, texture painting, and clothing customization, exporting to the standard VRM format.
3d-avataranimecharacter-creatorvrm -
Character Creator NEW Discovered by AI
Reallusion Character Creator is a 3D character design solution for generating, customizing, and auto-rigging realistic or stylized humanoid character assets for games and animation pipelines.
3d-charactercharacter-designdigital-humanreallusion
