RealVisXL_V4.0 is a text-to-image model hosted on Hugging Face. It belongs to the class of Stable Diffusion XL implementations and produces images from natural language prompts.
The model is distributed as a StableDiffusionXLPipeline compatible with the Diffusers library. Installation proceeds through pip install of diffusers, transformers, and accelerate, after which a pipeline loads the repository identifier SG161222/RealVisXL_V4.0 with bfloat16 precision and CUDA device mapping. A provided code example generates an image for the prompt "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k". Notebooks on Google Colab and Kaggle, plus local applications Draw Things and DiffusionBee, offer additional runtimes. The repository also lists availability through Mage.Space.
Its license is openrail++. The model card carries the DOI 10.57967/hf/5476 and is tagged with Safetensors format. The maintainer maintains a set of related models under the same account, including ParagonXL, NovaXL variants, RealDreamXL, and their Lightning editions.
RealVisXL_V4.0 is offered at no cost for download and local execution.
RealVisXL is an Image generation product. It focuses on generating photorealistic images from text prompts using a fine-tuned SDXL model. RealVisXL is an open-source project aimed at AI artists and developers. The project is open source (Open Source). The product ships for the web, the command line, and API.
It is developed by SG161222, and the product first shipped in 2024. Among its 3 catalogued features are text to Image, Realistic Output, and Diffusers Support.
Latest indexed changes and source events
SG161222/RealVisXL_V4.0 verified by the PulseGate indexer
