Welcome Creators! Thank you for your support of RunDiffusion! We are thrilled to have you as a member of our community, creating incredible art on our platform! To show our appreciation, we have created this document specifically for you, to help you make the most of your subscription. We are confident that it will be an invaluable resource for you.

Benefits

$10 per month of promotional balance awarded
Discount on Large Servers
Access to MAX (48GB VRAM) servers
100 GB Private Storage
Persisted Extensions and Custom Nodes
#CreatorsClub Discord
Train Models (DreamBooth)

Server Variations

We have four options when it comes to choosing the hardware to use when working on RunDiffusion. We keep this simple. We have tailored our hardware options to match the software for a good experience. Here is a brief explanation of the options.

Small (SM) - 8GB VRAM. Perfect for diffusion, image generation, and can do deforum at small resolutions.
Medium (MD) - 16GB VRAM. Faster than the small server. Perfect for diffusion, image generation, ControlNet, and Deforum. Does do Dreambooth training but takes about 30 to 45 minutes to train a model at 1500 steps.
Large (LG) 24GB VRAM. This does it all. Perfect for all the jobs you need done. Diffusion can be done in batches and your image output can be multiplied by a factor of 3. The extra VRAM enables for a very fast and stable Dreambooth experience. This server is perfect for use with LLMs. We have discounted this server for all our Creative Club Members because it gives the best experience.
MAX - 48GB VRAM. This is our largest GPU. Best for massives batches, inferring and training with Language Models. Use when you need to put the pedal to the metal.

Access to Discord

Once you have signed up for Creator's Club, click the "Connect to Discord" button on the top of the UI. This will open a dialog box to connect your discord account and give you the Creator's Club role, giving access to the Creator's Club section of the Discord.

Storage

As a Creators Club member, you will receive 100GB of storage space exclusively for your use. You can increase your storage space for an additional fee, and you can also choose to add backups for an additional fee. Otherwise, there will be no backups made. Reach out to RunDiffusion to make those arrangements. Please note that you are responsible for the contents of your storage space and should ensure that they comply with the Stability Ai usage agreement. Keep in mind that this space is reserved exclusively for you and is not accessible to anyone else. https://huggingface.co/spaces/CompVis/stable-diffusion-license

You will start with an empty storage, minus a few essentials like the base model for Stable Diffusion. The shared models do not take up your storage and are there to use.

Additional models can be found on HuggingFace, Civitai and across the web.

SDUser Volumes and Paths

Sometimes the software will need the path to find files you’ve uploaded. Here are the common formats you will be using.

Pop open the Filebrowser, we find it's best in a new tab - less crowded!

The file browser drops you into the root directory where you have a few folders.
1. Models - this is where all your models go. Checkpoints, LoRA, Embeddings, Controlnet and even audio and LLM models all go here.
2. Images - this is where the outputs from all the applications go. They are organized by app.
3. Custom - Certain app assets can be accessed from here, per application. For example, custom scripts for Automatic1111.
4. Logs - all the logs are stored here. They are organized by app name and date.
5. .system - this is a hidden folder, it can be accessed in the event you need to reset your box if you have broken it from installing extensions, etc.
The root path for these files is /mnt/private/
So your LoRA models would be in /mnt/private/models/lora/ for example.

Errors (Terminal)

If you are experiencing errors make sure to check the logs. Open the "server manager" on the left hand menu and you will see the logs there. You can also view your them via the Logs folder on your storage. Open the logs file there to see the error message that might be popping up. Great for troubleshooting or getting help.

Errors (Bad Gateway)

If you get a black screen, but the timer frame and file browser still stays there you may have a "Bad Gateway". If you highlight the black portion of the screen you will see those words! There are generally three causes of Bad Gateway errors:

Out of storage. Check your account in the top right of the Smart Timer and see how much storage you are using. If you are over 100GB you will need to delete some models or other files until it is below 100GB. Try to keep some room free at all times.
Installing an Extension. When you install an extension and then refresh your UI, you will get a bad gateway error. Hit refresh on your browser after a minute and see if the UI loads. Note some extensions take longer to get out of a "Bad Gateway" state.
Extension problem or misconfiguration. If you are trying different extensions, settings, scripts or VAE, you may find you get a bad gateway on boot. This can be fixed by either deleting the offending file if you know it, or by resetting to back up. Reach out with a support ticket to address this issue.

Custom Models Uploading/Download to Server

Because of the size of models, we don’t recommend uploading them to your server through Filebrowser. (Unless you have blazing fast internet upload speeds). Your server is equipped with a 5Gbps fiber internet that can download models in seconds. We recommend the following methods over anything else.

The Civitai Downloader (Small Civitai logo on left menu) let's you download models easily and quickly. You can search by url or name. They will automatically sort to the correct server based on the model metadata.

Through Filebrowser Terminal

Don’t be scared, this is easy.

First watch the video to see how it’s done in 90 seconds. https://streamable.com/f7cwr5 - Quick Version https://streamable.com/o3x2p6 - Detailed Explanation
Spin up a SM server.
Open Filebrowser and navigate to the top right to “open image gallery in” a new tab.
Go to the folder you want to upload the model in. We recommend uploading to models/v1 or v2 depending on if it is a 1.5 or 2.1 model respectively.
Now go back to the top right and click on the icon that looks like this “< >”.
At the bottom of your screen you will see a “Terminal”

You have three commands you can use.

aria2c

aria2c is a torrent client that has been installed for your use. You can download from any URL or magnet link. This can be very very fast if you use the right option to download.
Get your URL and type aria2c -x8 [URL]
e.g. aria2c -x8 https://huggingface.co/protogen_model.ckpt

gdown

To download something from Google Drive you have to first allow the file to be shared. You will then take that link and bring it into your server and use “gdown” to download it.
Log into your Google Drive
Find the file you would like to download to your server.
Right click and find “Share”
Under “General access” change “Restricted” to “Anyone with the link”
Copy this link to your clipboard
Open the terminal in Filebrowser. In the terminal type:
gdown --fuzzy [URL]
gdown --fuzzy https://drive.google.com/file/d/etcetcetc...
You will not add a space or a new line after the --fuzzy flag. The url is just too long for the line on this page.
The download will start and you’ll see the model in your filebrowser once you refresh

wget
wget is the easiest to use. It’s simple.
wget https://huggingface.co/protogen_model.ckpt
Note that we find that aria2c is generally better and faster!

Email: admin@rundiffusion.com
Discord: https://discord.com/invite/wH6dTyBpCf #support

Creator's Club Documentation