Blog Layout

BING.com just got blinged up with an exciting new AI feature

Jeffrey Edwards • July 7, 2023

Stable Diffusion image generation is now as easy as one click plus a few keyboard strokes away.

To go along with the ChatGPT powered search bar, now the image bar has Dall-E powered image generation. Similar to ChatGPT, prompts are used to direct A.I. to create output, this time with images. It has fun creative capabilities worth exploring. For those familiar with Stable Diffusion or Midjourney this is a tool similar to that.

Let's take it for a spin:

First open up Bing.com in a Microsoft browser like Microsoft Edge.

Next you have multiple options, either click on Chat, in the search box or on Images (as pointed to by the arrows below).

I went ahead and clicked on chat. We can begin talking with BingGPT, requesting it to generate images, like so:

Basic Use:

Oops that’s not quite what I wanted. We can use the same iterative approach like with ChatGPT to ask it to make adjustments and corrections. Note that the memory it has for images is more constrained.

The images are thumbnails. You can click on them to open them in a new window and enlarge them. There you can also download them.

Different ways to access: If we had instead clicked on images then on image creator, here’s the interface for that:

The creative process often involves inspiration. This interface helps with that, with its built in features and examples.

Simply click on the photo of a particular art you find tantalizing or an effect you want to emulate

Clicking on the lighthouse photo enlarges it and reveals the prompt used to generate that image. Notice it was asked to produce a painting in the art style of pointillism. I liked that, so I added that word to prompts I made in the past.

Here are some examples with and without the keyword “Pointillism”.

As you can see this is a great way to get inspiration to help try new creative options to make different style images.

It’s perfectly capable of generating artistic images or realistic images. If it starts doing one when you want the other simply add the descriptor into the prompt. Even detailed descriptors like particular art styles can exert finer control.

Also use similar strategies for getting the most out of prompts as with chatgpt: iterate, be detailed, be precise, etc.

Limitations:
Image generators are notoriously bad for adding text to images and also handling fingers.

As always, I encourage you to test drive the feature for yourself. Also consider not only saving the images, but also the prompts in case you want to revisit creating certain similar styles, but with specific alterations.

Jeff Edwards,

A.I. Content Creator

Follow my Social Media for the latest in A.I. info or philosophical takes, Oh and the occasional Cat GIF:

https://twitter.com/657Edwards

< Older Post Newer Post >