Open-source alternatives to Taming Transformers

jetson-inference is a set of libraries and tools for executing optimized deep learning models on embedded GPU hardware. Its primary purpose is to enable real-time computer vision and AI inference at the edge with low latency and high throughput. The project distinguishes itself through high-performance streaming analytics and the ability to execute concurrent AI pipelines on auto-grade silicon. It provides specialized support for multi-sensor stream processing, utilizing zero-copy data transport to load camera frames directly into GPU memory. The codebase covers a broad surface of capabiliti

This project is a PyTorch implementation of StyleGAN2, providing a library and research framework for training style-based generative adversarial networks. It serves as a toolkit for high-resolution image synthesis, utilizing competitive minimax optimization to create realistic synthetic visual content. The framework incorporates specialized architectural components such as style-based latent mapping, multi-scale feature modulation, and self-attention layers to improve structural coherence. It distinguishes itself with advanced training stability techniques, including exponential moving avera

IOPaint is an AI image editor and Stable Diffusion inpainting tool providing a web interface for removing objects and replacing image content. It utilizes latent diffusion image processing to synthesize high-resolution replacements for erased sections of an image. The project features a specialized AI background remover for isolating subjects and an AI image upscaler that employs super-resolution models for general photos and anime artwork. The software covers a broad range of capabilities including image segmentation for object isolation, face restoration for improving facial details, and t

This project is a deep learning framework for AI image super-resolution and facial synthesis. It provides a diffusion model image upscaler and a generative facial image synthesizer capable of transforming low-resolution images into high-resolution outputs using pretrained model weights. The system utilizes iterative diffusion refinement and low-resolution guided sampling to restore fine details and sharpness. It supports both unconditional image generation, where images are created from scratch, and guided resolution enhancement for high-fidelity facial reconstruction. The repository include

This project is an AI image upscaling and high-resolution generation tool. It uses tiled diffusion to create ultra-large images by processing them in smaller, overlapping regions to prevent memory crashes on limited hardware. The system manages spatial composition through regional prompting, which routes specific text prompts to designated areas of an image. It maintains visual stability and global coherence during the upscaling process using noise inversion and structural guidance. Additional capabilities include tiled detail upscaling and memory optimization for the variational autoencoder

StyleGAN3 is a PyTorch implementation of a generative adversarial network designed for high-fidelity image synthesis. It functions as an image synthesis model and a deep learning research tool used to train and deploy networks that generate realistic synthetic imagery from custom datasets. The project is specifically an alias-free generative model, utilizing an architecture that eliminates jagged artifacts to produce smooth translational and rotational image sequences. This enables the creation of alias-free videos and the generation of high-resolution photos without visual distortions. The

pix2pixHD is a conditional generative adversarial network designed to transform semantic label maps into high-resolution photorealistic images. It functions as a high-resolution image synthesizer and an image-to-image translation model capable of producing synthetic images at 2048x1024 resolution. The system includes a semantic image editor that allows for the modification of high-resolution visuals by updating the underlying semantic label maps. This enables interactive image editing and the generation of photorealistic images based on source images or discrete label maps. The framework pro

This is a PyTorch implementation of a text-to-image model designed for synthesizing high-fidelity images from natural language descriptions. It utilizes a diffusion image generator to transform latent embeddings into visual data through an iterative denoising process. The system employs a two-stage latent mapping process, using a CLIP-based latent prior to map text embeddings to image embeddings before decoding them into pixels. It features a cascading diffusion decoder that produces high-resolution imagery by passing low-resolution outputs through a sequence of models at increasing scales.

This repository provides a complete framework for training generative adversarial networks (GANs) that produce high-resolution photorealistic images, up to 1024 by 1024 pixels. The core technique is progressive layer growth, where both the generator and discriminator networks start training at low resolution and gradually add new layers to model finer details, enabling stable synthesis of large images. The framework includes a high-resolution image generator, an image quality metric evaluator, a latent space interpolation tool for creating smooth transition videos, and a multi-resolution datas

Stable Diffusion is a generative machine learning pipeline that synthesizes high-resolution visual content by performing iterative denoising within a compressed latent space. By mapping natural language embeddings into pixel outputs through conditioned probabilistic processes, the framework enables the generation of images from text prompts and the transformation of existing visual inputs based on semantic instructions. The architecture utilizes a modular execution environment that decouples model loading, scheduler logic, and inference components to support diverse hardware configurations. I

dalle-mini is a text-to-image model and generative AI system designed to transform natural language descriptions into synthetic images. It functions as an image generation training toolkit and a generative model capable of creating visual representations from text prompts. The project provides a containerized deployment for consistent execution across different computing environments. It includes the necessary scripts and configuration files to train custom generative models from datasets. The system utilizes an autoregressive transformer architecture that treats visual data as discrete toke

TripoSR is a single-image 3D reconstruction system that generates a high-quality textured mesh from one photograph in under half a second. It uses a feedforward neural network to process a single image through a transformer architecture, compressing the input into a compact latent vector that conditions the entire reconstruction pipeline. The system outputs a separate UV texture map with configurable resolution, replacing vertex colors for higher-quality surface detail. The project is built around an end-to-end differentiable pipeline that trains the entire reconstruction system from image in

VAR is a visual autoregressive model and image generation framework that applies large language model scaling laws to visual data. It functions as an image generator that uses a coarse-to-fine next-scale prediction approach rather than traditional raster-scan tokenization. The system utilizes scale-based tokenization to represent images as a hierarchy of discrete tokens. It generates high-resolution content by iteratively predicting the next resolution level, refining coarse predictions into fine-grained details. The project covers a broad range of capabilities including autoregressive image

Kolors is a generative model implementation for synthesizing photorealistic images from natural language descriptions and visual references. It utilizes a latent diffusion model framework to produce high-fidelity imagery, operating within a compressed latent space to improve generation efficiency and quality. The system functions as a multilingual image generator, interpreting text prompts in multiple languages to produce semantically accurate visual outputs. It includes a custom model training pipeline that uses low-rank adaptation to teach the model specific subjects or artistic styles from

MAGI-1 is an autoregressive video generation model designed to synthesize high-resolution video sequences from text prompts and image references. It functions as a generative system for text-to-video, image-to-video, and video-to-video transformations. The model utilizes an autoregressive architecture that treats spatio-temporal patches as a sequence of discrete tokens to maintain temporal motion. It employs a variational autoencoder to compress the spatial and temporal dimensions of video data and uses distillation-based step scaling to allow for inference budget control. The system integra

Shap-E is a generative 3D modeling system that creates three-dimensional digital assets from natural language descriptions or two-dimensional images. It functions as a generative model capable of producing three-dimensional implicit functions and assets. The project includes a 3D latent encoder that converts trimeshes and 3D models into latent representations using point clouds and multiview renders. It utilizes an image-to-3D generator to produce assets from synthetic view images and a text-to-3D generator to build shapes from text prompts. The system implements a pipeline involving latent

This project is a PyTorch implementation of a generative adversarial network designed for high-resolution image synthesis. It provides an image synthesis model that produces realistic images from latent vectors and learned class conditions, supported by a latent space projection tool to find numerical vectors representing specific target images. The implementation features adaptive discriminator augmentation, a training technique used to prevent discriminator overfitting when training on limited image datasets. It also includes a generative model evaluation suite providing quantitative metric

ToonCrafter is a model that combines latent diffusion, reference-based colorization, and sketch-guided control for cartoon animation and interpolation. It functions as a cartoon video interpolation model, a reference-based colorization model, and a sketch-guided animation tool, all built on a latent diffusion animation framework. The project distinguishes itself by integrating three core capabilities into a single pipeline: generating smooth intermediate frames between two cartoon images using diffusion-based priors, transferring color and style from a reference image onto black-and-white ske

MegaTTS3 is a bilingual speech synthesis system that generates natural-sounding speech in Chinese and English, including seamless code-switching within a single utterance. It functions as a text-to-speech engine, voice cloning system, and speech-to-text alignment tool, built around an acoustic latent compression model that encodes high-resolution audio into compact representations for efficient processing. The system distinguishes itself through accent intensity control, allowing adjustment of a speaker's accent strength in generated speech, and voice cloning from short audio samples for pers

Sana is a framework for high-resolution image and video synthesis based on a linear diffusion transformer. It provides a toolkit for the training, fine-tuning, and execution of text-to-image and text-to-video models, as well as a video generative world model capable of simulating physical environments with precise spatial control. The project is distinguished by its use of linear complexity layers to handle high resolutions and its support for long-form, minute-length video generation in real time. It implements a two-stage inference paradigm that separates structural generation from visual t

This project is a diffusion model training framework and image synthesis pipeline. It provides the tools necessary to train generative models to learn image data distributions through an iterative denoising process. The framework includes a generative model evaluation tool consisting of automated scripts used to measure the quality and accuracy of produced samples. The system covers model training pipelines and performance evaluation for generative diffusion models.

LatentSync is an audio-driven video generator and latent diffusion lip sync model designed to synchronize a speaker's lip movements in a video to a target audio track. It provides a lip synchronization training framework for developing synchronization networks on custom video and audio datasets. The system utilizes a video preprocessing pipeline to clean, segment, and align face data. It includes a visual sync evaluation tool that calculates confidence scores to measure the accuracy of audio and visual alignment in generated videos. The project covers capabilities for custom synchronization

TRELLIS.2 is a generative image-to-3D system that creates high-resolution 3D assets with physically based rendering materials from 2D images. It utilizes a sparse voxel representation to handle complex topologies and internal structures without relying on iso-surface fields. The project features a structured latent space representation that maps geometry and texture attributes to maintain visual fidelity. It employs an optimization-free geometry reconstruction process to decode latent representations directly into voxel grids and includes a PBR texture generator for synthesizing base color, r

StyleGAN2 is a TensorFlow generative adversarial network and image synthesis model designed to produce high-resolution synthetic visual content. It functions as a deep learning architecture that learns patterns from image datasets to synthesize new images. The project includes a latent space projection tool for mapping existing images to latent vectors to analyze their representation within a generative model. It also provides an image quality evaluation framework to measure the visual fidelity and diversity of synthetic outputs. The system covers the full generative pipeline, including imag

OmniGen2 is a unified image generation model and multimodal large language model designed to handle text-to-image generation, image-to-image tasks, and image editing within a single framework. It functions as a causal language model visual engine capable of generating and editing images based on combined text and visual inputs. The system features in-context visual composition and subject-driven generation, allowing it to extract subjects from reference images and place them into new scenes. It also supports instruction-based image editing, where specific objects or styles are modified via na

Hallo2 is an AI video generation tool and audio-driven portrait animation framework designed to transform static images into speaking videos. It functions as a portrait image animator that synchronizes a single photo with an audio track to produce high-resolution talking head videos. The system includes a distributed animation trainer for fine-tuning deep learning models using custom datasets and distributed computing resources. It employs hierarchical video generation and temporal consistency modeling to produce long-form character animations that remain stable over extended durations. The