Start Year: 2023
I am fascinated by generative AI and what it can do. Text-2-Img generators are very quickly evolving, but they still have a weakness, when it comes to arms and hands, and specifying how two people interact.
This project, for me, is an exploration into what is currently possible, finding and pushing the limits of the AI. When is tango in a generated image still tango, and when is it only the aftertaste of cliché?
Admit it: if you are a tango dancer, you have tried it. And failed to create something interesting. Usually there are issues with everything that makes tango interesting: the embrace (position, heads, arms, hands), the expressions, the pose. The AI reproduces stereotypes mostly, and doesn’t understand hands and basic dance anatomy. However, with some seed…
Over the last weeks, I have played with different Ai models to generate pictures of tango dancers. Sometimes with good results, sometimes with not-so-good results. In general, there are two main areas of problems with the images: Cliche and Hands.
Part 2 of a series. How to use DALL-E via Bing to create tango visuals.
Part 3 of a series. How to use DALL-E via ChatGPT to create tango visuals.
Part 4 of a series. How to use Stable Diffusion via NightCafe to create tango visuals.
Part 5 of a series. How to use Midjourney to create tango visuals.