AI Image Bloopers #1

A selection of the out-takes from my AI image studio.

Brought to you by popular demand!

At some point after the first hour of intense concentration/typing/chitchat, at a Co-Work Corner session, I will spin my laptop around to amuse and delight my fellow co-workers with a particularly horrific AI image that I’ve generated… that didn’t make the final cut for one of my posts.

These are always greeted by much shrieking and laughter (ok, mainly shrieking) followed by a ‘you should do a post of these!’ suggestion.

So here we are with the ‘AI Bloopers’ that didn’t make the final edit.

The brief

AI prompt: ‘cartoon version of this profile pic’

Inspired by

Marie Vandoorne

’s The Bored Millennial, I first experimented with a cartoon version of me.

Some might say I have a pretty cartoonish look anyway, so surely it wouldn’t be too much of a stretch for an AI image generator.

Or so you’d think…

The result had more than a passing resemblance to Twilight actor Robert Pattinson, so after a whizz through all the ‘style genres’, I settled on ‘realistic’.

Although in a lot of these images the word ‘realistic’ is highly implausible.

I uploaded only one image for reference, the headshot profile pic above (left), which is referred to in the AI prompts as ‘the woman in the picture’.

IMPORTANT: I really must stress that the ONLY image of myself I have EVER uploaded to Gencraft AI is this headshot.

You would do well to remember that specific info later on, as things go very awry the further you get down this post.

It’s a long one, so you might need to view it on the web, if your email cuts out midway.

And so it begins…

The first post that I decided to use a characterisation of myself as the lead image was ‘Now I know how Taylor Swift feels!’, this was the final image selected:

AI prompt ‘The woman in the picture brandishing a pressure washer on a moss covered tarmac drive, in front of a red brick house. She is wearing boots with flowered print. She is covered in mud up to her face.’

Obviously I had a good go at trying to place Taylor Swift in this kind of set-up, but the result just didn’t make me laugh, whereas the slightly skewed face of the one above (and that it shared my dress sense) did.

AI prompt: ‘Taylor swift brandishing a pressure washer on a moss covered tarmac drive. Taylor is wearing floral boots, and covered in mud.’

Sadly, try as I might, after about 20 attempts I still couldn’t get me, Taylor Swift, Linda Hamilton or a Ghostbuster to be wearing anything like my actual floral print wellies or be ‘covered in mud’. Sigh.

Next up…

AI prompt: ‘The woman in the picture, standing in a laboratory wearing a white coat, brandishing a clipboard. She has just had a great idea, and is shouting for joy!’

I used this as the lead image for ‘Co-Work Corner: the detail’, however those with a photographic memory amongst you, will have noticed that I cropped out the third hand/arm on the published version.

This seems to be a recurring theme with AI image generation, it is so keen to show you what it can do, it almost always overdelivers, I never thought I’d need to add these phrases to an AI image prompt:

‘with only two hands, not three’

‘just two legs’

‘wearing matching shoes’

Despite my best efforts at reducing the number of arms down to two, Gencraft didn’t quite nail the other features, as these two alternatives ably demonstrate.

AI prompt: ‘The woman in the picture, standing in a laboratory wearing a white coat, brandishing a clipboard. She has just had a great idea, and is shouting for joy! With only two hands, not three.’

AI prompt: ‘The woman in the picture, celebrating with a bottle of champagne because she has just had a great idea!’

As per the caption above, I did not include the phrase ‘an elbow instead of a right boob’ to achieve this little gem.

What was it thinking?

Don’t even get me started on the limb count!

Moving swiftly along…

This was the eventual winner for ‘Income stream ideas: the vote’.

AI prompt: ‘The woman in the picture, holding a map, trying to decide which way to go’

However, this one nearly won, because I just look so damned hot…

AI prompt: ‘The woman in the picture, looking at a road going in three directions and not knowing which one to take.’

However, I felt it had too much cleavage, and so updated the text prompt to include the phrase:

‘Not showing her cleavage.’

And got this…

AI prompt: ‘The woman in the picture, looking at a road going in three directions and not knowing which one to take. Not showing her cleavage.’

Give me strength.

It is worth noting that whenever you see an AI picture of me that hints at cleavage, I have had a good 19 attempts at trying to get her to keep her baps in, but sometimes you just can’t fight City Hall.

I wonder if Gencraft AI was programmed by a male robot?

And then there were three…

This one is spookily accurate to how I actually look when woken up before 7am and so was an obvious choice, for ‘I will never be a 5am girl’.

I love the expression on my face enough to overlook the additional hands, which no amount of cropping could remove.

AI prompt: ‘The woman in the picture, being woken up by a huge alarm clock.’

However, I did try and include actual ‘5am girl’ Diane von Fürstenberg, resulting in this trio of horrors!

AI prompt: ‘The woman in the picture, standing next to Diane von Fürstenberg.’

I have no doubt that I will bump into these three doing a stage show at the Edinburgh Festival Fringe this year!

Last but by no means least…

I shall leave you with this utter doozy.

This is the published image for the post ‘Co-Work Corner: next steps’, illustrating quite subtly that we have ‘lift-off’.

AI prompt: ‘The woman in the picture, riding on the back of a rocket.’

Before I reveal the first attempt at this one, I CANNOT stress enough that I have only ever uploaded ONE IMAGE, my headshot, to Gencraft.

It has no reference real or imaginary for the rest of my body.

IMPORTANT NOTE: The AI prompt did not include ‘the Sports Illustrated calendar’, and I haven’t had any official firearms training.

So just for my inner circle, and once you see it, you’ll realise why.

The image that has solicited the highest number and loudest shrieks at any Co-Work Corner event…