Diffusion Bee allows to generate images locally with Stable Diffusion

Diffusion Bee allows to generate images locally with Stable Diffusion

Are you starting to get around Dall-E? Then it’s time to try Stable Diffusion, another image machine learning model that runs locally on your Mac! Developer Divam Gupta has just released a new version of Broadcast Bee, a graphical interface for Stable Diffusion allowing to use it without technical skills on macOS. It offers (among other things) to generate images from a handful of words.

Unlike Dall-E, Stable Diffusion is a free project that requires get your hands dirty to start. Diffusion Bee solves this problem as it is simply a program to drag into your Applications folder. Everything works offline and no data is sent to the cloud.

The interface is very simple: a frame invites you to enter a description, and a click generates the image. Several options are available, and we can also use AI to modify an existing image by adding elements or letting it imagine what is happening outside the frame. Advanced settings are offered for users wishing to go further.

The application pulls on the graphics card and it is recommended not to use other programs during the image generation (it freezes quite quickly for me on my Intel iMac). Newer Mac owners should get better performance, as Diffusion Bee is optimized for M1 and M2 chips.

If you want to play with the AI ​​without downloading anything, remember that Dall-E recently removed its waiting list. However, image generation is limited to a certain number of credits per month, and you will then have to checkout to continue creating.