DALL-E, the latest tool to hit the web, is a powerful AI image generator that has turned many creative minds.
DALL-E. You’ve probably already seen the latest trend, especially if you’re on Twitter: AI-generated images. The concept of the tool is quite simple – users enter two descriptive words or phrases that claim to represent two completely different topics and ideas. Think Darth Vader and kittens or Top Gun and whatever else you can think of. This tool combines the two sentences or ideas and creates individual images in which the two merge together and become one original image. Think of it like automatic photoshopping.
It’s easy to imagine how something like that would arrive. There are endless possibilities for weird pairings and shocking images that are guaranteed to spread.
How does the tool work, who developed it and what can we expect from this kind of apps in the future?
How does DALL-E work?
The most popular app you see on social media right now is “DALL-E mini ” (soon to be found on Crayion.com). She creates images in 9 blocks with variations on the same weird looking theme. The results are so weird that they aren’t hyper-realistic. They just look like weird drawings. . . nearly.
Who developed DALL-E?
Developed by OpenAI, a non-profit AI research organization (co-founded by Elon Musk), DALL-E is a multimodal implementation of GPT 3 (generative pre-trained transformer) with billion parameters that “exchanges text for pixels ”. The tool is trained on “text-image pairs” from the Internet. Basically, it scours the internet very quickly, linking sentences to images. It uses zero-shot learning to produce output from a description and a hint.
We had to look up what zero-shot learning is and found that – when a computer learns what a horse is but not what a zebra is – zero-shot learning is the idea that the computer can infer that a zebra looks like a striped horse. So he can make a kind of distinction. This is not much different than when we come to conclusions by deduction. . . right?
Similar to the reaction to deepfakes, there is also here some people questioning the ethics behind the whole thing. It’s not hard to understand this point of view, because any kind of AI that creates itself will make people a little uncomfortable for various reasons such as: B. Because of privacy and morality.
Apart from the question of morality, we can’t help but think about how this will be the role of photographers, retouchers and even videographers and editors in the future could replace as it becomes easier and easier to replace their skills with a technology like DALL-E.
Looking at the pictures we show here, you might think: “Are you serious? These images look like a fried nightmare that can only exist in the meme-verse, rather than anything cohesive, artistic, or professional!”. That’s because we’re using the free, not-so-advanced version of the tool. That’s right – there is a more powerful tool called DALL-E 2 that demands more from its users and regulates certain content restrictions and permissions.
So what does a more powerful DALL-E look like? Well hold on tight.
This image is taken directly from their website – a case study for the tool. The text description read: “An astronaut rides a horse in photorealistic style.”
So as you can see, it’s not just about “real” looking photos. There are many variations on what your images can look like. It is this artistic freedom that fires our imaginations, as the capabilities of the tool are so advanced that it is hard to imagine that this tool will not find lasting application in the creative industries.
Impact on Photography and Video
Although DALL-E does not pose a specific threat to video production, it is not difficult to see where that could lead. There are already a number of video creation tools out there, but they fall short of producing what the average filmmaker can create or what can be found in stock video galleries. Also, there are already a lot of AI-generated scripts and stories gradually seeping into the world, but no one has yet offered complete video and audio generation in one package.
Besides, it’s also lightweight imagine what this threat to creatives looks like. When a client is working on a new advertising campaign and wants a desert valley with a meandering river at golden hour and a parked car on the lower left corner of the street. . . Well, you could pay for this shoot, with everything that goes with it: models, location, photographer, photo assistant, accommodation, transport – the list goes on. Or they could do it themselves without leaving the office and save a ton of money in the process.
There are tons of fun uses for these kinds of tools. The @lincoln_baiyeaux Instagram account, for example, posts content intended to replicate the profile of a modern-day “film photographer.” This account uses the same hashtags for Kodak stocks, vintage cameras, and cliche slogans as captions. The only catch is that none of the pictures are real. They are all 100 % computer generated. Even “his” profile picture is computer generated. It’s a fun, comedic project that also makes you think about the potential of this technology, even if the consequences aren’t always positive.
The idea of AI generated users is not new though. We’ve dealt with bots in comment columns and forums for years. And it’s becoming increasingly common for internet users to be skeptical about almost everything they see online. Most creatives try to achieve a certain level of originality and specificity for their own artistic voice in order to create good and authentic art. So such art that cannot be recreated by machine.
Every time we’ve used the app, our reactions couldn’t have been weirder. Most of the time we laughed when we saw her for the first time. Then we found ourselves just staring at them in silence, almost mesmerized by the weirdness.
So here are some of our own creations in all their wacky, random glory.
Boba Fett, Barbie and the Bass Guitar
After doing some pop culture mashups tried it, we wanted to try it with a stock photo to see what we’re dealing with here. So we typed in “sunset on the mountain top” and this is the result. .
Our first thought was that it’s not terrible, but it’s not great either – it’s just what we expected.
Now that we know what DALL-E is, what it can do and how it works, what do you think? Will it stay like this? Will it completely change the creative landscape for professionals and their careers? Only time will tell.