Menu Menu
[gtranslate]

New ‘DALL-E Mini’ website sparks arty social media trend

A new open source AI application lets anyone input prompts and generate images via a database algorithm. Social media has been messing about with its tools and coming up with some truly bizarre results.

The inevitable shift to algorithm-generated artwork is well under way, at least as far as surreal abstract pieces go.

You may remember that we wrote about the advanced AI programme called ‘DALL-E 2’ a few months back, which creates insanely accurate and impressive original art based on user prompts. It incorporates a database of references and merges them together.

Crucially, though, this programme was a private academic piece of software that won’t ever be used by the general public.

That is likely to change as the technology becomes more accessible and cheaper to develop, with third party companies and coders taking it upon themselves to launch their own products.

The new ‘DALL-E mini’ website that’s been popular on social media this week is one such example. Though it shares the same name as the original (created by OpenAI) it was developed by an alternative coder called Boris Dayma who used a smaller data sample.

Available via the website Hugging Face, users can type up their own prompts and wait a few minutes for a collage of images to be created. There are ample Twitter threads and posts documenting a wide range of different text prompts, some hilarious and others just plain quirky.

Check out this collection of Darth Vader pictures holding a bass guitar, for example.

As mentioned in our previous article, the technology does pose concerns.

Misinformation could become even more rampant if anyone can fabricate images of celebrities and political figures. Artists may be at risk too, as commissioned artwork and unique pieces fall in demand and AI replaces real human work.

For now, the newly-made pictures are mostly being shared as memes or novelty content. The DALL-E Mini is fairly limited in its ability to create specific details on things like faces, resulting in some slightly disturbing surrealist interpretations of human features.

Here’s a selection of borderline demonic interpretations of Britney Spears holding a vacuum cleaner in space that I experimented with on the app to demonstrate the limits of the current version. Terrifying, no?

Many posts online have gained a decent amount of buzz as users test out various pop culture characters and place them in unique situations. How about Fire Emblem meets Pingu? We’ve got you covered.

It’s fun to try new ideas and see how they turn out, but it’s worth remembering that this technology will significantly impact the art world in the near future. Obviously we don’t know exactly how just yet, but there’s no doubt it’ll have ripple effects on marketing, advertisements, commissioned artwork, and stock image libraries.

As our music industry feels the weight of AI voices and TikTok snippets, it’s safe to assume other mediums are soon to follow suit. Still, it’s fun to share bizarre concepts with the world while the tech is still in its infancy.

If you fancy giving it a go yourself, check out the DALL-E mini website here.

Accessibility