| System: | Ubuntu 24.04 |
| Edition: | 010326 |
AI for in-browser image generation
Launch Automatic1111 — a powerful and user-friendly tool for creating images from text descriptions — on a ready-to-use cloud server with a GPU. This Ubuntu-based image includes everything you need: a pre-installed web interface, NVIDIA drivers, popular Stable Diffusion models, and nginx for instant browser access.
No need to configure dependencies or compile kernels — simply choose your preferred configuration, launch the VM, and start creating.
Stable Diffusion is an open-source generative neural network that creates images from text prompts. Using the latent diffusion method, it gradually reconstructs an image from noise based on the semantics of your description. Trained on billions of text-image pairs, the model accurately links words to visual elements: styles, colors, and composition.
Automatic1111 simplifies this technology: you describe your idea, and the system handles parameter tuning, memory management, and result generation.
Linux 6.17.0-14-generic #14~24.04.1-Ubuntu SMP PREEMPT_DYNAMIC Thu Jan 15 15:52:10 UTC 2 x86_64 x86_64 x86_64 GNU/Linux
When using large models (SDXL or any other high-capacity models), it is essential to choose a configuration with a higher amount of System RAM. We recommend selecting configurations with 32 GB, 64 GB, or 128 GB of RAM. Otherwise, you may encounter stability issues, such as periodic crashes or disconnections of the Stable Diffusion service. This issue does not occur when working with smaller models like Stable Diffusion 1.5 or Stable Diffusion 2.1.
Tip: For a smooth experience with SDXL and extensions, it's best to choose an RTX 4090 or RTX 5090. This will ensure high speeds and prevent "out of memory" errors.
To connect, wait until the StableDiffusion link appears under your server's IP-address. Clicking this link will redirect you to the Automatic1111 interface.
Please note that access is provided via a link containing a token, which is required for proper authentication. If you wish to share access with someone else, you must copy the entire URL, including the ?token= parameter.
To achieve the best results, clearly describe what you want to see. A typical prompt structure looks like this: [subject], [action or pose], [background], [style], [quality], [details]
Example:
a futuristic warrior in a glowing armor, standing on a cliff, cinematic lighting, volumetric fog, ultra detailed, artstation
You can use commas or brackets to set priorities. Negative prompts are also supported, allowing you to exclude unwanted elements from your image.