Accomplice is graphic design and illustration software 100% powered by Artificial Intelligence. With just a sentence you can create endlessly amazing AI designs and illustrations for your projects. Accomplice uses high-powered GPUs and state of the art Machine Learning models CLIP + VQGAN to create, in minutes, what would take a regular CPU hours!
Prompts are the sentences you use to tell Accomplice what to create. Like “colorful sunrise over the mountains”. The Accomplice Community is a great place to go and see all kinds of different prompts and their results.
Once you give Accomplice your prompt, it starts with random static at Iteration #0, and then starts trying to turn that static into whatever your prompt is. Every 20th iteration Accomplice will show you its progress. And then Accomplice will stop at iteration #200, which is oftentimes (but not always) its best attempt.
One credit gets you one run of 200 iterations – where Accomplice shows you every 20th iteration – based on whatever your prompt was. So one prompt, where you ultimately get one unique image, equals one credit.
Yes! There may be similarities, but since Accomplice starts from random static every time, no two prompts will ever be exactly the same. And oftentimes they’ll be totally different!
No! Accomplice uses OpenAI’s groundbreaking CLIP model to connect text and images, and that’s how it knows what it knows about the world. But it is not just combining existing images together, it is using its knowledge of the world to create new images out of whole cloth.