Self Host your own AI for Image Generation using biniou [Part 1]

A quick and easy guide on how to self-host your own AI for Image, Audio, and Video generation on a budget.

Self Host your own AI for Image Generation using biniou [Part 1]
A view inside the Crunchbits datacenter

Last time, we explored how to self-host our own AI GPT bot; this time, we will explore how to create AI-generated images. Why pay corporations like OpenAI or Google to use their Image Generation AI when you can just self host one?

So what exactly is biniou?

biniou is a self-hosted webui for several kinds of GenAI (generative artificial intelligence). You can generate multimedia contents with AI and use a chatbot on your own computer, even without dedicated GPU and starting from 8GB RAM. Can work offline (once deployed and required models downloaded).

I recommend Biniou over other tools and software because it is an all-in-one tool and very well-maintained. It can be used for text, image, and audio generation, and it supports multiple models for a wide variety of results.

I'll be using a Crunchbits Cloud GPU VPS [RTX 3090], which has instant deployment and GPU Optimized templates so we can get started immediately. These GPU Cloud VDSes start at just $65/month, one of the best market rates possible.

Crunchbits' cheapest GPU Server (RTX 3070) starts at just $65/month
Crunchbits' Control Panel with GPU-Ready Templates

Getting Started

As mentioned in their Git Repository, there are many ways to get started but we'll be exploring the docker option which is straightforward and fast to set up. Please refer to our previous blog article if you need instructions for installing docker and the required NVIDIA docker packages.

# Create the work directory
mkdir binou && cd binou

# Build the Image
docker build -t biniou https://raw.githubusercontent.com/Woolverine94/biniou/main/CUDA/Dockerfile

# Run the container 
docker run -it --gpus all --restart=always -p 7860:7860 \
-v biniou_outputs:/home/biniou/biniou/outputs \
-v biniou_models:/home/biniou/biniou/models \
-v biniou_cache:/home/biniou/.cache/huggingface \
-v biniou_gfpgan:/home/biniou/biniou/gfpgan \
biniou:latest

If you did everything correctly, it should be accessible on port 7860 of your server, through which you can open it in a web browser (e.g., 192.168.69.69:7860)

Biniou Image Generation Page

As you can see, there are multiple Algorithms and Models which are available for image generation. We will use the default settings, which are Stable Diffusion and the "v2ray/stable-diffusion-3-medium-diffusers" model.

Put any prompt you want in the prompt box, and hit generate.
Keep in mind that the first time you run any model/algorithm, it will take some time to download and do the initial setup. Hopefully you have a Crunchbits GPU server equipped with a 10Gbps port 😄.

Initial Run with Stable Diffusion
Biniou Image Generation Page with Results

You can see images generated above for my prompt. There are tons of customization and parameters available, which can be used for fine-tuning the generated image.

And that's it. You have your very own AI generating images for you. Biniou also has settings where you can lock this behind a login. Throw in a reverse proxy like Nginx to the mix, and you can share a URL with your friends to try it out!

Biniou is a very capable tool that can do not only Image generation, but Audio and Video as well. We will explore this in Part 2 of this blog post. Stay Tuned!