Civitai Dataset and Training Guidelines
In this document, we provide you with information about how to prepare and upload a dataset to us for the purpose of model training.
Uploading an Existing Dataset (as a Zip file)
This section assumes that you have already created a dataset outside Civitai and are ready to use it for training a model.
Accepted Files and Formats
If you elect to upload an existing dataset, then we ask that you provide the data in a zipped file (.zip extension). The file may include only images, or it may include images and text files which include captions.
Filename Convention
A directory should contain an image file with a name that matches the caption text file. These will be interpreted as a pair in training.
| Image File Name | Caption Text File Name | |
|---|---|---|
| Pair One | 1.jpg | 1.txt |
| Pair Two | 2.jpg | 2.txt |
Accepted File Extensions
For images, we accept any of the following image file extensions: .png, .jpg, .jpeg
For caption files, we only accept .txt file extensions.
If you do not provide captions with your images, you would simply have images in your dataset's directory.
Zipping Your Data
Important! If this step is not done correctly, we will not be able to parse your files correctly.
Typically, when you have a dataset created, you will have a directory structure that looks like the below:
When you zip this data up, you should only select your image/caption file pairs. Do not include the parent folder when preparing the zip file.
Creating a New Dataset on Civitai
This section provides information on how to create a dataset on civitai.com
Adding/Uploading Images
Images can be dragged and dropped to our uploader. These images can later be given captions (see "Providing Captions" below).
Providing Captions
Captions are an important component of training a successful model. You can train models without them; however, your model will be more flexible and better overall when you do provide them.
Great captions describe the content of the image on which they are written. There is a wealth of knowledge on the subject. See this post for in-depth info on captioning.