Open-source Dall-E “Open Diffusion” is now available on a website

We have experienced a surge of AI-created images using Dall-E and other AI blackboxes (not accessible?by the general public). However designers, researchers and other AI experts have been able to play with these tools and generate original images.

The image generation is usually guided by a very standardized UI where the user follows an iterative process based on writing a text prompt describing the idea and repeat:

• Generate - adjust prompt - generate-....
• Select best image
• Edit (post-process)
• Publish

If you were not part of the limited group that had access to this technology, you will be very happy to know that DreamStudio just published an open source AI that will allow you to try an AI similar to DALL-E 2 for free!.

The name of this technology is Open Diffusion Model, and as with DALL-E 2, there are restrictions on prompts, such as sexual or violent images.

Currently you can run the cloud beta version in their website (see link below), but the model is expected to be released on Github in the next few days.

You can try to generate few images for free and with a pricing of just under $12, you can buy about 1000 prompts. The actual number of prompts available depends on the complexity of the calculations and the resolution of your image. See below some of the UI settings that you can adjust prior to run the model:

Below some of the "Picasso-style" images that I generated using this tool (love that we can get higher resolution than Dall-E 2... I could even apply an AI and resize the images to print them on a canvas):

Text Prompt: "Malaga city painted by Picasso" (not great...)

Text Prompt: "Barcelona painted by Picasso" (better interpretation...)

Text Prompt: "Madrid city painted by Picasso" (my favorite)