AI Resources

There are several different AI image generators that I’ve been using to make images and video. Some of them are better than others, although the ones that aren’t as good, still produce very interesting images…

Midjourney is by far the best for pure text-to-image generation. I wrote a post detailing some of that previously, so I won’t repeat myself. Suffice to say, it produces some beautiful stuff, if you’re lucky.

I’ll say a few words about some of the other ones I’ve used…


One of the first text-to-image generators I used was Craiyon. It was before public access to DALL-E was available. It didn’t produce good images, but it was amusing more than anything. Since I first started playing around with them in June 2022, the quality has improved… but it’s still not brilliant. I’m not quite sure why someone would want to use this other than to get bonkers AI images out of it without any effort

woman doing yoga” – June 2022

Woman doing yoga” – April 2023
I think AI generally has an issue with ‘yoga’ and often creates horrifying bodies from it.

A nearly 53 year old woman with brunette curly shag haircut green eyes bit tired of the bullshit frankly.” – June 2022
This looks more like the results of some kind of Frankensteinian experiment than a 53 year old woman, but there is an attempt to express ‘tired’ at least.

A nearly 54 year old woman with curly auburn shag haircut green eyes bit tired of the bullshit frankly.” – April 2023
VERY good hair, but ’54 year old woman’ still looks a lot older than it should.


As soon as DALL-E opened up to the public (previously, I’d been sitting on the waiting list for, well, ever without success), I started playing around. Again, image creation has got better since it was first launched, but it’s still not great…

Lonely Gia Milinovich dancing in a desert in the style of a Looney Tunes cartoon” – November 2022
After generating a bunch of disappointing images, I’d just decided to throw my name into a prompt to see what it made and… I liked it.

Lonely Gia Milinovich dancing in a desert in the style of a Looney Tunes cartoon” – April 2023
If I’m honest, these are less interesting as the previous ones. It’s lost something…

While DALL-E isn’t great for image generation, it is very good for expanding – or ‘outpainting’ – images you already have. Let’s say you have a something that is square and you want it to be 16:9…

I took this image of a dorodango I made suring the first Covid lockdown in 2020.

I uploaded it into DALL-E then got DALL-E to generate more of the image.

It generated four different options. This is the one I liked best.

You can use the AI-ness of it to create amusing images, too.

Someone posted this image of Trump on Twitter wondering how long his tie was – as Trump always wears ties that are far, far too long.

I used DALL-E to generate a reply.


There are a few AI video options easily available. Most of them you can try out for a bit without paying.
This creates ‘presented’ videos with people using only text as the input. You have control over which ‘AI person’ you choose, which accent, which voice, the background colours etc etc. It is technically only supposed to be used in corporate or educational settings.
The videos I’ve made from this have only really been of one ‘style’. It seems that the AI takes your input image and produces a new similar image from that, then takes that images and does the same, and on and on, morphing the images as it goes along. You can’t make someone walk, for example, you just get the image changing…sometimes in interesting, often times in grotesque ways.

Kaiber is another one that produces ‘evolving’ images from both an image and a text prompt. They are interesting when you first see them… less so when you realise there’s mainly one ‘style’.

Tome is cool. The images it creates aren’t great (yet), but it will write a story and make images for you based on a text prompt with as much or as little detail as you want. I’ve only played around with this and not actually taken anything out of my account yet, but this feels like it has potential.