A Maltese developer has built a local picture generator as artificial intelligence takes over the planet.
“I’ve always loved computers, and even though I didn’t have a degree, I taught myself how to develop software using the internet. I began over. Picaii designer Sacha Cassar told Lovin Malta, “I worked at a low-wage factory, but I kept learning and improving.”
Users submit a few images for a variety of profile pics, from tattooed boxer to slick-suited CEO, futuristic cyborg to dirty mobster, for a price.
“My girlfriend helped me succeed too. I created this from her design. “Working on that project together and seeing it come to life was amazing.”
Cassar employs steady diffusion, a 2022 deep learning text-to-image model.
It generates detailed pictures based on text descriptions, but it may also be used for inpainting, outpainting, and text-guided image-to-image translations.
“At Picaii, we’re proud to build on top of some of the industry’s most talented and hardworking engineers. Experts spent years developing stable diffusion technology, and we’re honored to use it.
“One credit buys one model. This can be a person, animal, or couple. Take additional photographs to “train” the model. Five to 20 images from different angles and lighting are needed.”
Cassar advises people who want to know how AI sees them.