About Mage Space

Mage.space is a powerful online image generator with a number of features that aren't present in most other online services. You can use it for free in some modes, and even the free modes offer some powerful features.

The first free mode is "FAST TEXT-TO-IMAGE." Type any simple prompt, describing what you want to see in plain English. There are controls to pick one of seven aspect ratios, so choose one of these to pick the shape or format of the overall image, and click the GENERATE button in the lower right to make an image (or press Ctrl+Enter to generate.)

To change the look of your images, there's a set of style presets called Prompt Filters. If you choose a pack of these filters, then choose a specific look from the drop-down, these let you achieve all kinds of looks and visual styles, without needing to type long, complex prompts to achieve them.

There's also a feature called DYNAMIC PROMPT that lets you choose words from your prompt and turn them into a Wildcard, which could be randomly replaced with other words from a list you create. This can add variety to aspects of your prompts. See this video for more about Fast Text-to-Image.

Another free mode is called "FAST SELFIE." This starts by letting you upload a picture of yourself (or someone else) and is optimized for portrait-style images that are just a close-up on one person's face.

After you provide it with an image of your face, you can type a prompt describing what you wanted the picture to look like. You might type "Professional portrait, formal, business suit, professional photograph of an employee" for example, and a cell phone selfie would be restyled into a portrait you could attach to a resume. Not bad for a free service!

You can add to the description if you want to specify more about the gender, appearance, hairstyle, and so on. There are also "Selfie-fast" sliders at the bottom (you may have to click "Show Details" to view them) that balance between the strength of the image you uploaded and the strength of the prompt. If your output doesn't match your own appearance enough, you can slide them to the left.

You can do creative and fun things with this. If you type "dog," you can make yourself appear more or less like a dog (have fun with those sliders), and if you type "selfie on a tropical beach" you can picture yourself on a vacation. See this video for more about Fast Selfie.

Mage.space uses the word "app" in reference to a saved workflow, which saves your prompt and all the settings in a re-usable way. Mage encourages users to build and share apps. You can find apps and try them, and they will come pre-configured to do a task such as dressing a picture you upload as a Harry Potter character.

Mage.space can be used to generate many kinds of content, with ratings you can set from G (All ages) to M+ (Mature Themes). When using Mage.space for free, you can only view or create G or G+ rated content.

All of the paid membership levels include verification to allow M and M+ ratings. With those ratings, mage.space seems to offer uncensored access to Stable Diffusion models, with no content filter limiting your prompts. However, be sure to read the full Terms & Conditions for yourself to make sure you are following their rules.

Mage offers three membership plans: Basic ($8/month), Pro ($15/month), and Pro Plus ($30/month.)

I signed up for the Pro plan, opening up a fuller set of features. Instead of just "Fast" Text to Image, there is another Text to Image mode that lets you choose which Stable Diffusion model you want to use, from a large built-in collection, as well as a choice of LoRAs to help style the image. The included Stable Diffusion models worked fine, although they were not the most famous and widely used models available for download elsewhere.

In what seems to be an implementation of IPAdapter, mage.space lets you choose Reference Images to control the look of the scene or match a reference face. ControlNet is also there, letting you control the composition, posing, or other aspects of your generations, based on a sketch or photograph you upload. There are a number of other controls and variables that will be familiar to people who have used Stable Diffusion interfaces such as Automatic1111 or ComfyUI.

Images are generated one at a time, and generation can sometimes take over a minute for a single image.

There is also a useful Image-to-Image mode, which allows you to resample and up-res images, using either the same prompt they were created with, or with a new and different prompt to transform the images. Again, this is similar to what you might be familiar with from other Stable Diffusion interfaces.

A major limitation on Mage.space is a resolution cap, which limits how much detail can be added to the images you generate. When you use HIRES FIX, or when you use the Image-to-Image mode to modify an image, you are limited to a maximum size of 1024 pixels in the image's smaller dimension. This makes HIRES FIX extremely useful when you use Stable Diffusion 1.5 models (with lower base resolutions) but only offers smaller gains with SDXL models. (If you generated an SDXL image in Landscape aspect ratio, the base resolution would be 1216 x 832. Turning on HIRES FIX will produce a small increase of resolution, to 1496 x 1024.)

This same limitation also applies to Image-to-Image mode. Even if you start with an image that is 2048 x 2048 resolution, then the resulting output would only be 1024 x 1024. Mage does allow you to Upscale images, which can give you pixel resolutions above the cap, although upscaling alone doesn't add any new detail based on your prompt.

Even with the resolution cap, there are still workflows that let you use Mage to refine high resolution SDXL-generated images. If you made a full-body image of a person, for example, and wanted more detail in the person's face, then taking that image into a paint program, upscaling the entire image, and copying just the face into a separate file would let you upload the face image to Mage, do Image to Image to refine it and add detail, then download the new face, paste it into place in your paint program, and soften the edges of the layer to blend it into the original scene. It takes some work, and Mage will help you make high resolution masterpieces.

Mage.space is a terrific free or low-cost platform for creating images. It doesn't have all of the features like Inpainting and Outpainting that you'd get if you ran a Stable Diffusion interface on your own computer, but for people who don't have a suitable graphics card to generate locally, it offers its own set of basic and not-so-basic functions that allow you to generate uncensored images and manipulate uploaded images in a variety of creative ways.