| Last Updated | Changes |
| 9/6/2023 | First version published |
| 11/10/2023 | SDXL Support added! |
| 4/11/2023 | User Interface Updated |
What is the Civitai Image Generator?
The Civitai Image Generator is a web-based interface for Stable Diffusion, which allows us to input text prompts and receive image outputs! You don’t need a powerful graphics card or PC, all the processing is handled by Civitai.com – you can ever run the generator on your cell phone!
It’s super easy to get started creating your own AI generated art with the Generator, and this guide will walk you through how to do so!
Who has access?
Everyone with a Civitai.com account can access the Generator and start creating images in seconds!
For help creating and configuring your Civitai account, check out this Guide to Civitai (Part 1)!
We also provide the ability to turn off the Image Generator entirely, from your Account Settings, Early Access Features! This might be useful if you generate locally with your own PC and have no interest in generating on-site.
Even if you do generate locally we’d encourage you to leave it on and try it out, as it provides an excellent way to take a model or LoRA for a “test drive“; getting a feel for the types of outputs it produces, before deciding whether to download!
What can it generate?
One of the best things about the Civitai Generator is that it has access to literally thousands of resources uploaded to Civitai, and can leverage them in the image generation! Your imagination is truly the limit, as the generator can be used with innumerable permutations of models, LoRA, and embeddings to create fantastic content, in various styles;
How do we use it?
Upon logging into Civitai, you’ll notice a blue “Create” button in the top right of the screen (all pages). Clicking this expands the Generator drawer from the right.
We can also access the Generator from a specific resource by selecting the blue “Create” button from the Model Card – which will pre-populate the Generator’s “Model” field. Note! Only models which the uploader has granted permission to be used on the Civitai Generator will show the “Create” button here, otherwise, the regular download button will be shown.
If we find a particularly exciting prompt we’d like to use in the Generator, we can select Start Creating from the bottom of the image metadata window.
Clicking Start Creating will send both the positive and negative prompts, along with the advanced options like the Sampler, CFG Scale, Steps and Seed to the Generator – pre-filling these options for you.
Note that you may not be able to generate an exact facsimile of the chosen image due to a whole range of complex factors, but it should be close!
If you’re browsing images and see something you’d like to try replicating in the generator, you can click the little “Remix” button at the top right of each image – this will pre-populate the Generator with as many parameters as are available, and should re-produce a similar image!
For users who’re new to AI image generation we’ve tried to make things as simple as possible! Advanced settings have pre-set values which are optional to tweak, but images will generate just fine without setting them manually.
All the currently available settings are explained (at a very high level!) below;
- + Add Model (or Swap)
Models are what power Stable Diffusion, which in turn powers our Generator. Clicking Add Model opens a popup listing all the selectable Models available for inference (generating images against) on Civitai.
- + Add additional Resource
Additional resources include LoRA and Embeddings.
- Prompt
A short description of the image you’d like to see!
- Negative Prompt
A short description, or a list of keywords, for content you don’t wish to see in your image.
- Mature Content
If you don’t wish to see adult/lewd/not safe for work content, you can toggle this selector off. It won’t completely block such content from appearing in your generations, but it will reduce the chance, somewhat.
- Aspect Ratio
For SD 1.5 models, we allow images to be created in Square (512×512), Landscape (768×512) and Portrait (512×768) dimensions.
For SDXL Models, we allow images to be created in Square (1024×1024), Landscape (1216×832) and Portrait (832×1216)
The Advanced settings are pretty complex concepts, and we’ll delve into those in a dedicated guide, but for now if you’d like to read more about what these are, and how they change the generated images, you can check out the Civitai Generative AI Glossary.
To view your Image Queue, and Image Feed, there are icons at the top of the Generation Panel;
Cost
There is a small Buzz ⚡ costs involved in Generating images! It costs a lot of money to support the generation of literally millions of images each day (we regularly generate upwards of 9 million images per day!), and it’s just not sustainable to offer unlimited free generations to all our wonderful users.
We’re very grateful the site has become so popular and continues to see more users everyday, but servicing that many people is becoming increasingly expensive and a value needs to be added to offset that cost.
We had considered just continuing to lower the generation limits but that would be a temporary fix, requiring more and more cut backs over time. This way we can remove all generation limits, in exchange for community content or revenue to keep the site running.
It’s very important to us that our free users can continue to use the Generator. To that end all users receive a certain amount of free buzz every 24 hours, just for logging in and opening the Generator. Additionally we have many ways (and more are coming) for users to earn Buzz by reacting to images, posting content, having people like and collect your posts, and engaging with other site content.
We’ll expand on this section very shortly, detailing exactly how much Buzz you can expect to pay for each type of generation.
Restrictions & Limitations
A number of advanced image generation features are on the way, including Hires. fix, ControlNet, and ADetailer, but are not yet implemented.
