Open-source alternatives to StackGAN

Generative Adversarial Text-to-Image Synthesis

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.

This project is a PyTorch implementation of a discrete variational autoencoder designed to compress high-resolution imagery into discrete latent representations. It functions as an image autoencoder that encodes visual data into discrete codes and decodes those codes back into reconstructed images. The system utilizes a latent space image compressor to convert images into a compressed, discrete format. This allows for generative image synthesis and the analysis of image compression by transitioning between raw pixels and discrete code sequences. The implementation covers latent space visuali

Learning What and Where to Draw

StyleGAN is a TensorFlow-based generative adversarial network framework designed for the synthesis of high-resolution synthetic imagery. It utilizes a style-based generator architecture to create realistic visual assets from latent vectors, focusing on the production of high-fidelity images. The system incorporates style mixing and stochastic noise injection to control visual attributes and fine-grained details. It uses adaptive instance normalization and progressive resolution upsampling to manage image quality and variety across different resolutions. The framework covers the full lifecycl

Wasserstein GAN

Code for the paper "Improved Techniques for Training GANs"

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

Official pytorch implementation of the paper: "SinGAN: Learning a Generative Model from a Single Natural Image"

Text to image synthesis using thought vectors

iGAN is a framework for producing synthetic images using generative adversarial networks. It provides a web-based interface for interactively creating and editing imagery across categories such as landscapes, architecture, and fashion using pre-trained models. The system enables precise control over visual output through latent space exploration, interpolation, and projection. Users can guide the generative process using an interactive editor featuring sketching, coloring, and warping brushes to refine specific regions or shapes in real-time. The project supports both automated scripted gene

StarGAN is a PyTorch image-to-image translation framework designed to synthesize visual styles and attributes across multiple domains. It implements a generative adversarial network that serves as a deep learning image translator for modifying specific visual characteristics within an image dataset. The framework uses a single unified model to handle translations between multiple image domains rather than requiring separate pairs of models. It is a research implementation that learns mappings between different image attributes without the need for paired training data. The project covers the

A list of all named GANs!

Invertible conditional GANs for image editing

CycleGAN is a generative adversarial network framework designed for unpaired image-to-image translation. It enables the conversion of images between two distinct visual domains using datasets that do not require direct one-to-one matching examples. The project implements a deep learning style transfer tool capable of artistic style transfer, object transfiguration, and domain-to-domain conversion. It uses a dual-generator architecture and cycle-consistency loss to ensure that images translated to a target domain and back recover their original state. The framework covers core machine learnin

TIP'19 AttGAN: Facial Attribute Editing by Only Changing What You Want

This is a PyTorch-based implementation of diffusion models for synthesizing photorealistic images and video. It provides a framework for text-to-image and text-to-video generation, as well as unconditional image synthesis. The system utilizes a cascading diffusion pipeline to produce high-resolution imagery by passing low-resolution outputs through a sequence of super-resolution models. It also includes capabilities for image inpainting, allowing the reconstruction of masked or missing regions of visual media guided by surrounding context and text prompts. The project includes tools for diff

Awesome-GANs is a curated resource list and research repository focused on the development and evaluation of generative adversarial networks. It serves as a structured index for academic literature and open-source implementations dedicated to the creation of synthetic data generators. The project provides a framework for training competing neural networks to produce outputs that mimic the statistical properties of original datasets. It emphasizes the use of configuration-driven pipelines to manage model hyperparameters and dataset paths, facilitating reproducible research workflows and standa

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

Z-Image is an AI image editing engine and generation framework designed for photorealistic synthesis and the refinement of diffusion models. It functions as a multilingual text-to-image renderer and a system for training custom foundation models to generate and edit visuals using natural language instructions. The project distinguishes itself through a reasoning-based prompt enhancer that expands simple descriptions into detailed visual instructions using a structured reasoning chain. It also features specialized capabilities for rendering high-quality Chinese and English typography within ge

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

qrbtf is an AI QR code generator and image synthesis system that blends machine-readable data with artistic imagery. It uses a latent diffusion model and spatial control networks to produce functional QR codes that incorporate visual art generated from descriptive text prompts. The system provides a dedicated interface and programmatic API for tuning visual output, allowing for the adjustment of control strength, padding ratios, and error correction levels. It supports deterministic sampling via random seeds and the use of negative prompts to refine the final aesthetic of the generated assets

OutfitAnyone is a diffusion-based virtual try-on system and AI person-garment integration tool. It functions as an image-to-image clothing transfer model designed to visualize how specific clothing items look on any person regardless of their pose. The system adapts garment textures and shapes to a person's body and pose to produce photorealistic results. It specifically focuses on adjusting clothing deformation based on body shape to maintain high fidelity and detail consistency during the fitting process. The project covers AI fashion visualization and virtual garment fitting, providing ca

This project is a deep learning style transfer framework designed to apply artistic styles to photographs. It functions as a photorealistic image stylizer that merges the content of one image with the visual characteristics of another while maintaining the original geometry and structural details. The system distinguishes itself through the use of matting Laplacian matrices and semantic segmentation masks to prevent distortion and preserve edge fidelity. These capabilities allow for region-specific styling, where different aesthetics can be applied to distinct objects or areas within a single

Primitive is an algorithmic art generator and geometric image reconstruction tool that transforms raster images into stylized vector compositions. It functions as an iterative shape optimizer and raster-to-vector converter, approximating pixel-based photos by layering geometric primitives such as triangles, circles, and rectangles. The project utilizes a search algorithm to determine the optimal position, size, and color for each shape to minimize the visual difference from the source image. Users can apply shape constraint definitions to control the properties and orientations of the geometr