Become a master of the Rundiffusion platform, a cloud-based workspace that supports Automatic1111 and InvokeAi, with this comprehensive rundiffusion tutorial.
What is Rundiffusion?
Rundiffusion is a unique platform that lets you rent high-performance servers equipped with data center GPUs. It provides a dedicated workspace for you to execute tasks related to Automatic1111 and InvokeAi. The environment is user-friendly and can be set up in less than 90 seconds.
Pre-requisites for using Rundiffusion
Before you dive into the Rundiffusion tutorial, make sure your setup meets these requirements:
- An updated version of Chrome or Edge browser. Brave browser is known for issues.
- Ad-blockers should be turned off as they can interfere with image uploads to the UI.
- Familiarity with Automatic1111 or InvokeAi.
- Reliable internet connection with low latency to the Central United States.
Understanding Server Tiers
Rundiffusion provides servers in three tiers: small, medium, and large. The cost is based on an hourly rate, charged per minute.
For instance, if you pay for 1 hour on a small server (50 cents) and shut it down after 30 minutes, 25 cents will be credited back into your available balance. Importantly, waiting time for a server to spin up or spin down is not charged.
Getting Started
To begin, navigate to https://app.rundiffusion.com/. Here you will choose your platform (Automatic1111 or InvokeAi), the server size, and the version of Stable Diffusion (V1.5 or V2.1).
The option will default to 72-hour storage for non-creator club members. If you are a member of the Creators Club, you will see your 100GB private storage.
Once you continue, you will choose the server uptime for this session and see your account balance.
Note: When finished with the server and time is still remaining, click “stop server” in the top right corner of the webui. This will shutdown the server and provide a balance back to your account with the amount of time that was left on the clock when shut down.
Creating Your First Images
Upon launching your server, you will be greeted with the file browser on your right-hand side. The login for this is “user” and “password”.
There are two options above the file browser. The highlighted one, as shown below, hides the side view file browser, and the button to the right of it, allows you to open a Fullscreen browser in a new tab.
The next three buttons are for:
- View/open another session
- Help/FAQ/Resources and Tutorials
- Account menu where it will show your available balance (and storage if you are part of our Creators Club)
Model Viewer
Over in the top left corner, you will see the model viewer. Here you will choose between a list of models preloaded.
Note: If you sign up for the creators club, you will gain access to private storage and will not have access to all the preloaded models. However, within the creators club documentation you received by email, you are able to upload custom models as your leisure.
Prompt Windows
Below are our prompt windows. The top is your positive prompt, and beneath is the negative prompt.
The positive prompt is where you are going to input everything you want to be in your picture, while the negative prompt is for everything you do NOT want in your prompt.
Prompting
Prompting is a crucial part of this rundiffusion tutorial and requires a lot of research/trial and error.
Typically, your prompt should be structured as follows:
- Subject: This is your main objective. For instance, “A vast scenery of an unexplored planet”, “A fluffy white rabbit”, “1967 Ford Mustang crossing the checkered flag”.
- Medium: This refers to the material used to make artwork. Some examples are illustration, oil painting, 3D rendering, and photography.
- Style: refers to the artistic style of the image. Examples include impressionist, surrealist, pop art, etc.
- Resolution/supporting descriptors/lighting: These include all supporting details to really bring your image to life!
- Artist names: are strong modifiers. They allow you to dial in the exact style using a particular artist as a reference.
Note: The negative prompt is populated with descriptors of things we want to stray from in our generated images.
Samplers
For beginners in Stable Diffusion, stick with the Euler A sampler. It is highly renowned and great for nearly all cases. However, as you advance, you might want to explore different sampling methods.
Sampling Steps
Sampling steps determine how many steps the AI will take to create your prompt. Too low, and you will get an unfinished image, too high, and you will see an overproduced image. A safe bet is to stick between 25-65 steps.
Width/Height Slider & CFG Scale
The width/height slider determines the final resolution of your image.
The CFG scale provides the AI guidance on how closely the AI will follow your input prompt.
Batch Count and Batch Size
This will determine how many images, and batches of images, Stable Diffusion will create with your prompt.
Generating Images
Once you’re ready, click “Generate” and watch the magic happen. Once complete, you will see your images in the viewer below.
Congratulations! You’ve created your first images.
Remember, this Rundiffusion tutorial is just the start. There’s a lot that goes into generating art within Stable Diffusion, but with the right mindset, creative thinking, and a helpful community behind you, you will be making your own masterpieces in no time!
