But little by little, he’s done things such as fixing bugs and improving code, enabling it to get better at coming up with more complicated images, such as the Eiffel Tower landing on the moon. In the past, for example, it would be able to envision simple things like a landscape, Dayma said. But it didn’t get much attention beyond the AI community until the past couple of months, perhaps due to the limited quality of the images it could produce.
#Funny photo to text generator
He posted the text-to-image generator - then called DALL-E Mini - on Hugging Face, where anyone could try it out (it’s still available there under that name). Initially, Dayma said, he built it as a technical challenge he thought DALL-E was cool and he wanted to make it himself. And as with many nascent technologies, it is a work in progress in the near term, if left unchecked, it may produce outcomes that reinforce stereotypes and biases.ĭayma and some other coders built the AI system last July during a hackathon hosted by Google and Hugging Face, a company that builds and hosts AI models.
#Funny photo to text trial
In the process, however, Craiyon is effectively acting as a trial run for what could happen - good or bad - in the future if anyone can access such AI systems and solicit any kind of image from them with just a few words. “I think it’s important to be able to have an alternative where everybody has the same access to this type of technology,” Dayma said. But unlike Craiyon, which anyone can try, most of these are not available to the public: DALL-E 2 is open to users via invitation only, while Imagen has not been opened up to users outside Google.Īn image generated by Craiyon, an AI system that anyone can use to feed text to a computer and get a picture in response. There are similar, much more powerful AI systems than Craiyon, such as OpenAI’s DALL-E (Craiyon was initially named DALL-E Mini as an homage) and DALL-E 2, as well as Google’s Imagen.
Lately, people are typing in about 5 million prompts per day, Dayma said. Computers are getting better and better at ingesting words and producing increasingly realistic-looking images in response. The brainchild of Boris Dayma, a Houston-based machine-learning engineer, Craiyon is popularizing a growing trend in AI.
“I think that’s the main draw of it: You can make anything a reality,” Laming said in an interview with CNN Business. To use it, you just type what you’d like it to envision - “A rainbow lion eating a slice of pizza” - and it will spit out pictures in response. These and other pictures, which range from ridiculous to disturbing, were created with a freely available artificial intelligence system called Craiyon. An image created by an AI system called Imagen, built by Google.