Web Analytics

TLDR: My notes on using Dall-e and Midjourney. May be helpful to fellow facilitators, trainers and workshop designers interested in how to use emergent and generative AI in their work.

Let’s talk about generating images

Dall-e is owned by the same people who run ChatGPT, so signing up for ChatGPT will get you into both. Midjourney is a different tool and it has a bit more of a technical interface. Both allow you to use very detailed prompts in order to create images. From the set of (usually) 4 generated options, you can refine and fine-tune the results, and you can then use the final image on your website, presentations and so on without any concerned about getting sued by the big stock image companies.

How I’ve used it (them): I have used both tools to get inspiration for what sorts of images I might use on my website and in presentations, although I’ve not used the generated images (I am a big fan of Unsplash for images). I have also used a combination of ChatGPT and image creation tools to write a comic for a workshop with teens which saved me LOADS of time. In the future, I may use the tools more as I currently pay around EUR300 per year for image banks.

How other people use it (them): to create digital art for sale, t-shirt designs, book cover designs, to create inspiring visual worlds for story building, video game backgrounds, illustrations – honestly, if you need an image for something, these tools can create them. Also, if you have to use a particular colour scheme for something – i.e. green for a political party  – you can generate images that all use the same colours, which may be interesting for branding or to create a particular atmosphere.

Pros: for me, these are inspirational tools. I use them to get my creative juices flowing, especially as in Midjourney, I can also see what other people are doing. The human mind and what it can imagine is truly amazing! I

Cons: huge time sink (although very enjoyable!). Some of the images are a bit odd (especially the faces), but with refinement, you can do some interesting things. There are lots of questions about where this positions artists in the future. Midjourney can involve a bit of a learning curve. The possibility of DeepFakes…

Try them out: both tools offer a number of free credits but require sign-up.

Learn more:  https://openai.com/dall-e-2/ and https://midjourney.com

What I created

I’ve played with both tools quite a bit and they are incredible. For the purposes of this experiment, here is a single prompt that I used for both Dall-e and Midjourney, with a few iterations. See what you think.

The prompt: hyper-realistic image of 2 facilitators, one young man and one middle-aged woman, delivering a workshop to 20 participants of various ages. The participants are working together in small groups. The room is light and bright. There are two flipchats in the room and lots of sticky notes on the walls. Colourful. Happy. Laughter.

Dall-e: lovely and colourful but problems with faces

Step 1:

Step 2:

Step 3:

Step 4:

Midjourney: I ran out of credits in my testing, but also some facial weirdness…

Step 1:

Step 2:

Step 3:

Step 4:

Step 5:

Step 6: