A trial version of the AI ​​image generation ‘Stable Diffusion XL’, which is much more powerful than traditional Stable Diffusion, was released as a test, so I tried to use it–GIGAZINE

Stability AI, the AI ​​research company that developed Stable Diffusion, has greatly promoted traditional AI image generation.Stable Spread XL (#SDXL)it is being tested. AI web service stabilityDreamStudioI’ve already been able to use the “Stable Diffusion XL” trial version, so I tried to see how good the new image creation model would be.

The number of parameters included in the traditional stable diffusion was 900 million, but the number of parameters in “Stable Diffusion XL” reached 2.3 billion. The “Stable Diffusion XL” beta is a web service from Stability AI”DreamStudioavailable at

Upon accessing DreamStudio, at the time of writing, a popup was displayed announcing that early access to “Stable Diffusion XL” was available as follows. Click “Get Started”.

If you do not have a DreamStudio account, you will need to register a new account. Click “Login” in the upper right.

You can create an account using a Google account, Discord account, email address, etc. This time, click Continue with Google.

Select the account you want to use.

Click Accept.

The account is now opened. DreamStudio gives you 25 credits at the time of initial registration, allowing you to create about 125 images with default settings, so you can use this free credit to try out “Stable Diffusion XL”.

DreamStudio offers several Stable Diffusion models to choose from, including the Trial Stable Diffusion XL. Click “Model” in the upper left and first select the classic “Stable Diffusion v2.1 (v2.1)”.

This time, you can search Stable Diffusion and Midjourney for prompts from photos and illustrations.”UndirectedAnd try to create an image with the default initial value and number of steps. After entering the prompt, click Dream.

Terms of ServiceAfter confirmation, select “I accept the Terms of Service” and click “Accept”.

Then 4 images were generated.

Looking at the four photos side by side, it looks like this.

Next, change the model to “SDXL Beta (Stable Diffusion XL Beta)” and try to create an image with the same prompts and settings. Please note that the number of credits required is more than double compared to “Stable Diffusion v2.1”, so you should pay attention to the amount of credits consumed.

The image created with “SDXL Beta” looks like this.

I put 4 photos side by side. The image quality was surprisingly high, and the power of “Stable Diffusion XL” was clearly realized.

Stable Diffusion XL is scheduled to be open after training, but the name at that time may not be “Stable Diffusion XL”.