Dall E Mini AI image generator: how to try art and picture demo online - what does ‘too much traffic’ mean?

Almost anything you can think of can be made real by this impressive piece of AI tech

If you’ve been seeing bizarre, almost nightmarish images pop up on your social media, like something out of the fever dreams of Salvador Dali himself, you’re not alone.

Users on Instagram and Twitter have been sharing grids of semi-surreal imagery inspired by two seemingly random phrases smashed together.

We’ve seen cult sitcoms meshed with Spielberg classics (”Peep Show Saving Private Ryan”) and even a lovable green ogre taking residence in a supermarket (”Shrek lives in Asda”).

But what’s the meaning behind the madness, and how can you create such hypnagogic images for yourself?

Here is everything you need to know about it.

What is Dall-E?

The images mainly come from an AI image generator called Dall-E.

Named after the lovable trash-compacting robot from Pixar’s Wall-E and the aforementioned surrealist, it is an AI model that generates images from any prompt you give.

The sophisticated piece of software, developed by AI research and development organisation OpenAI, is capable of producing a wide range of visuals in seconds based on a string of words.

OpenAI first invented a programme called Dall-E in 2021, which was able to take a written suggestion and turn it into an entirely new AI-generated image.

Is this treason? Here’s what Dall-E imagines the ‘Queen’s Binley Mega Chippy Jubilee’ might have looked like (Images: Dall-E)

But the images were frequently grainy, inaccurate, and time-consuming to produce. Now, OpenAI has improved the software significantly, resulting in Dall-E 2 — a powerful new model that performs far better.

The version of the software currently available to the public is Dall-E Mini, which resembles the original iteration of the programme much more closely.

So while it won’t produce the genuinely impressive results of its more advanced counterpart, the fact that it often gets things slightly wrong only adds to the hilarity of its final output.

How can I use the generator?

It’s quite straightforward. Simply click on this link to access the AI image generator Dall-E Mini.

Then simply type in your phrase and wait for it to work its magic. This can take some time, and the results aren’t always reliable, and that’s if the thing works at all.

Since the AI tool has gone viral, the website it is being hosted on is getting bombarded with traffic, far too much for the site to handle.

Repeatedly, as we click ‘Run’, we’ve been told there is “too much traffic” and to “please try again”.

Sometimes, it’s taken multiple clicks of the run button to even get the bot to whirr into action, and it can take a couple of minutes to get your results, depending on demand on the site.

And just because Dall-E begins its calculations, doesn’t mean you’ll ever see your results: we’ve had it “running” for at least a minute or more at times, only for it to crash under the weight of demand.

WeE’ve had marginally more success getting results from the site when using it on an Android phone’s Chrome browser, but your mileage may vary.

It’s a hit and miss experience - fingers crossed the site is upgraded to cope with the influx of visitors in the near future - but when it does work, it provides some very amusing results.

How can I use the ‘full’ Dall-E 2 generator?

Currently, the programme is being slowly rolled out to users via a waitlist, but there are no concrete plans to make it available to the general public.

By gradually releasing their product, the OpenAI team can track its progress, improve their safety processes, and prepare their product for the millions of users who will be imputing their instructions in the near future.